Here’s what I usually find: they’re tracking everything and measuring nothing.

Meta says they drove 500 conversions. Google says they drove 400. TikTok claims 150. That’s 1,050 “attributed” conversions for a brand that actually had 600 orders.

Every platform is grading its own homework. And every platform gives itself an A+.

The brands have an MMM nobody uses and a post purchase survey they don’t use to calibrate spend.

So they make budget decisions based on bad data. They over-invest in whatever platform tells the prettiest story. They under-invest in channels that actually drive awareness but don’t get the last click. And they wonder why scaling spend doesn’t scale revenue.

Maybe they set up a MTA tool and think that its accurate since its ‘impartial’

Or they use an ‘AI Model for attribution that aggregates all signals’…. I hate black boxes for attribution…

Attribution isn’t broken because there’s no data. There’s too much data and not enough understanding.

So here are the 8 attribution models every DTC brand should know, how they fit together, and which 3 I actually use to manage $30M+ a year in ad spend.

P.s. this post is refreshed from one originally from February. Im incredibly grateful that there are close to 4000 of your reading which means 85% are new since I posted this originally. I’ve added more detail on CDV and updated a few details this time around.

The Christmas Tree Framework

Before I walk through the 8 models, you need to understand how they fit together. The biggest mistake I see is brands trying to solve attribution at the wrong level.

Think of attribution like a Christmas tree. (Pyramid / la tour Eiffel / any triangle)

At the very top is the star: your blended target. This is your MER (Marketing Efficiency Ratio) or nCAC (New Customer Acquisition Cost). It’s channel-agnostic. It tells you if the business is working, but not why.

Simple: How many new customers did you get? How much revenue? How much did you spend on marketing.

Accurate to the decimal & decides how much you should spend.

The upper branches are your channel allocation tools. Post-purchase surveys, Marketing Mix Models, and incrementality testing live here. They answer: which channels should I be spending more on? Which are actually driving results?

The lower branches are your campaign and ad-level tools. Multi-touch attribution (Triple Whale, Northbeam) and platform-reported data live here. They answer: which ads should I scale? Which should I cut?

The trunk is your supporting infrastructure. GA4, server-side tracking, CAPI setup. These don’t measure attribution directly. They make all your other tools more accurate.

And the roots, feeding everything, is cohort-based LTV attribution. This is how you nuance your targets in the first place. If Meta customers have 2.5x LTV versus TikTok customers at 1.8x, your CPA targets should adjust to relfect that.

Here’s why this hierarchy matters:

Most brands try to solve attribution from the bottom up. They install an AI-powered attribution tool and try to perfectly track every individual customer’s journey. They get precise numbers that feel scientific.

But precision is not accuracy.

if you start in the middle of the Christmas tree, you end up with the bottom left. Precise but not accurate attribution

so instead you need to start at the top and then layer in surveys and MMM so that you’re accurate. Although you don’t get precision from that

And only then can you layer in MTAs to get precision within that accuracy

Top down. That’s the framework.

Now let’s walk through all 8.

Layer 1: The Star. Blended Metrics

Model 1: MER and nCAC

What it is: Total marketing spend divided by total revenue (MER), or total marketing spend divided by total new customers (nCAC). No channel breakdown yet just look at your PnL

When to use it: Always. This is your north star. Every single day.

Why it matters: MER and nCAC are the only metrics that can’t lie to you. Every platform will inflate its own numbers. But your bank account is what you care about. When Meta says ROAS is 4x and your P&L says you lost money, your P&L is right.

Here’s a quick example of why this matters:

A brand comes to me with a 4.2x ROAS on Meta. Looks amazing. But when I pull the full picture, their MER tells a different story. They had a 35% new customer ratio (meaning 65% of “conversions” were returning customers who probably would have bought anyway), their COGS were 48%, and platform fees ate another 8%. That 4.2x ROAS actually translated to a loss of $15k that month.

Compare that to another brand at 2.8x ROAS with a much better new customer mix and tighter COGS. They made $45k profit.

MER shows us where to start. Is this business on the right track?

Limitation: Tells you IF something is working, not WHAT is working. You can’t make channel decisions from MER alone. It’s the star on the tree, and the place to start not end.

The key numbers:

Your maximum marketing spend = Gross Margin % – OpEx %. Cross this line and you’re losing money at the company level.

Your target CPA = projected 3-month customer profitability. This is your efficient growth zone.

If you’ve raised capital these let you know if you can burn cash for a bit while still profitably acquiring customers. Important for future profitability (whether for yourself or a potential buyer)

Everything else in this post is about figuring out how to spend between those two numbers as effectively as possible.

Layer 2: The Upper Branches. Channel Allocation

This is where most brands fail. They skip this layer and jump straight to ad-level optimization. But you can’t optimize ads if you’re spending on the wrong channels.

Model 2: Post-Purchase Surveys

What it is: A survey on your order confirmation page asking: “Where did you FIRST hear about us?”

When to use it: Every single day. This is one of my favorite attribution tools in all of DTC.

Why it’s powerful:

Post-purchase surveys measure something fundamentally different from click-based attribution. They measure what’s memorable.

Think about that. An ad has to be remembered to be impactful. If someone can recall that they first discovered you through a TikTok ad, that ad did its job. It created awareness. It built the mental association. The fact that they eventually Googled your brand name and clicked a search ad to buy doesn’t matter. The search ad didn’t create the demand. It captured it. Thats not scalable.

Click-based attribution overvalues lower-funnel performance by as much as 250%+. I’ve seen it across dozens of accounts. Google Search always looks like a hero in click-based models because it’s the last stop before purchase. But it’s rarely the first touch.

Post-purchase surveys catch what click attribution misses. We’ve seen cases where top-of-funnel creative on Meta drove 13x more incremental acquisitions than bottom-of-funnel creative. Click attribution told us the exact opposite story.

How to set it up:

Add a post-purchase survey to your Shopify checkout. The question: “Where did you FIRST hear about us?”

Options should include every channel you spend money on (Facebook/Instagram, TikTok, Google Search, YouTube, Podcast, Friend/Referral, etc.), plus “Other” with a text field. Keep it under 10 options. Randomize the order.

Target a 35%+ response rate. Extrapolate to all new customers. Then calculate your cost per new customer response per channel.

This tells you where to push budget at the channel level.

Limitation: People don’t always remember correctly. Long purchase cycles make this harder. And if someone saw 10 Meta ads over 3 months, they might just say “Instagram” without knowing which specific campaign drove them. That’s fine. You’re measuring at the channel level here, not the ad level.

The other limitation is that it’s self-reported, so you’ll see some channels consistently under- or over-report. That’s where MMM comes in to validate, but it's also partially the point. You want to know what people remember, not necessarily what was exactly first. Its an art and a science.

Model 3: Marketing Mix Modeling (MMM)

What it is: A statistical approach that uses aggregate historical data (spend by channel, revenue, seasonality, external factors) to estimate each channel’s contribution to sales. No user-level tracking required.

When to use it: Monthly or quarterly for directional validation. Not daily.

Why it’s powerful:

MMM doesn’t rely on cookies, clicks, or any user-level tracking. It’s completely privacy-safe. It looks at the relationship between your spending patterns and your revenue patterns over time, accounting for things like seasonality, promotions, and external factors.

Where post-purchase surveys tell you what customers remember, MMM tells you what the data says is driving revenue regardless of what customers recall.

The two work together beautifully. You can use MMM to validate that your post-purchase survey is sending you in the right direction. And you can use post-purchase survey data to calibrate your MMM.

If your post-purchase survey says “push more into TikTok” and your MMM agrees, you have high confidence. If they disagree, you dig deeper before making a big budget move.

The landscape:

MMM has had an awesome resurgence in the last few years. Meta released Robyn (open-source), Google released Meridian (open-source), and there are SaaS options like Measured, Sellforte, and others that make it more accessible. Northbeam released their own as well which I’ve been impressed with.

One of the biggest benefits of an MMM is that you can measure the marginal performance of a channel to know just not the holistic performance, but where you are in the bend, which can be very valuable to deciding where to put the next dollar, not based on the average performance, but based on the marginal performance

But I generally don’t recommend jumping into MMM as your first move. You need scale for the data to be meaningful (typically $500k+ per year in ad spend minimum), and you need clean historical data across channels.

Limitation: MMM is a model. It’s not ground truth. It gives you directional guidance, not absolute precision. It requires sufficient data history (ideally 2+ years of weekly data). And the outputs need interpretation by someone who understands both the math and the business context.

It also requires careful set up. If you have messy data or plan to just wing it for a quick check, probably best to skip the MMM. Might do more harm than good.

I’ve seen people run two MMM’s (moving from one to the other) and got drastically different outcomes (like suggested budget swings of 50%) based on set up. Be super careful with confidence here.

Some channels drive low-quality customers that look good on ROAS but don’t stick around. MMM helps you optimize for 12-month profit as opposed to just immediate return.

Model 4: Incrementality / Geo-Lift Testing

What it is: Controlled experiments where you turn marketing on or off in specific geographic regions and measure the difference in sales between test and control groups.

When to use it: Quarterly (or whenever you need to validate a specific strategic question).

Why it’s powerful:

This is the closest thing to scientific proof in marketing.

You’re not modeling. You’re not surveying. You’re running an actual experiment. “What happens to sales in Dallas when we turn off Meta retargeting for 4 weeks while keeping it on in Houston?”

If Dallas sees no meaningful drop in sales, your retargeting isn’t as incremental as Meta’s dashboard claims. If it drops significantly, it’s working and you have a real iROAS (incremental ROAS) number you can trust.

This is really valuable for answering the questions that other attribution models can’t:

• Is our retargeting actually driving sales, or just claiming credit for people who would have bought anyway?

• How much incremental revenue does YouTube actually drive (beyond what click attribution captures)?

• Would we lose anything if we cut our branded search spend by 50%?

• Is this new channel (AppLovin, CTV, etc.) actually worth the investment?

Tools: Haus is gaining popularity with DTC brands and makes it relatively easy to design and run geo-lift tests. Stella offers a self-serve platform. Meta and Google both have their own Conversion Lift Study tools within their platforms (though those are limited to measuring their own channels, which introduces some bias). TikTok has one too. Again, Northbeam launched one here as well.

Limitation: Requires planning, budget, and patience. Most tests run 4-8 weeks. You need enough geographic diversity and sales volume for statistical significance. And you can only test one or two things at a time.

It’s also an opportunity cost. You’re deliberately reducing spend in test regions, which means you might leave some revenue on the table during the test period.

But the insight you get is worth it. One well-designed incrementality test can save you from wasting hundreds of thousands on non-incremental spend.

I love combining post-purchase surveys with incrementality testing as a killer combo.

Layer 3: The Lower Branches. Campaign & Ad-Level Optimization

Now we get into the tools most marketers already know. These are powerful for ad-level decisions but dangerous if you use them for channel allocation.

If your campaigns have different strategies, use MMM’s or Geo lift testing to allocate there; if they are the same strategy, you can use an MTA instead.

Model 5: Multi-Touch Attribution (MTA)

What it is: Third-party tools (Triple Whale, Northbeam, Rockerbox) that track individual customer journeys across touchpoints and attribute fractional credit to each interaction.

When to use it: Daily, for ad and campaign-level optimization (Same strategy).

Why it’s powerful:

MTA is my primary tool for ad-level decisions. Triple Whale is my go-to, though we use Northbeam as well.

Here’s what makes MTA better than platform-reported data: you can actually see who purchased, where they came from, what they purchased, and validate the attribution yourself. With Facebook’s dashboard, you just have to trust them. With MTA, you can audit it.

I also like MTA because you can look at first-click attribution and understand the full journey. Where did someone originally come from? Did they see multiple ads before converting? Which ad was the first touch versus the last touch?

This matters because the ad that gets the last click isn’t always the ad that deserves the credit. Someone might discover you through a TikTok-style UGC video (first touch), then see a retargeting ad a week later (middle touch), then Google your brand name and click a search ad (last touch). If you only look at last click, you’d think Google Search drove that sale. MTA shows you the full picture.

The practical application: this is the data that powers our Account Control Charts. When I plot CPA vs. Total Spend for every ad on a scatter plot, I’m pulling that from MTA. It tells me which ads are efficient but under-scaled, which are burning cash, and which are the winners I need to push harder.

this is the level at which you make individual ad decisions on based on spend and CPA

To make sure your account is well optimized, plot CPA vs. Spend on a scatter plot. Your top ads should be in the low CPA, high spend zone. If they’re not, you have a media buying problem that no amount of new creative will fix.

Limitation: MTA is mostly click-tracked. It tracks touchpoints it can see (clicks, website visits), which means it overcounts channels that generate clicks (Google, Meta) and undercounts channels that generate awareness (podcasts, word-of-mouth, TikTok views that don’t result in a click). Triplewhale and Northbeam have loaded in deterministic views recently and we are starting to use it and seeing strong results. I dont like interface view conversions, Im OK with deterministic views but you need to evaluate it carefully.

Another limitation is that 99% of impressions don't convert. Don’t use the above chart to over-optimize into a promotion ad (strong last click influence) and starve a creative with 70% of the ROAS that speaks to selling points to a cold audience.

Ultimately You just can’t compare channel-to-channel directly using MTA. That’s what post-purchase surveys and MMM are for. And your MTA you need to use critically.

Use MTA for ads and campaigns. Not for channel allocation.

Model 6: Platform-Reported Attribution

What it is: The attribution data that Meta, Google, TikTok, and other ad platforms report in their own dashboards.

When to use it: Daily, but with skepticism. Always cross-reference with MTA.

Why it’s useful (with caveats):

I’m not going to tell you to ignore platform data. I use it every day. But I use it for specific things, and I know where it lies.

Here’s what platform attribution is actually good for:

Understanding purchase velocity. I look at 1-day, 7-day, and 28-day attribution windows side by side to understand how quickly customers purchase after seeing an ad. What percentage convert on day 0-1? Day 1-7? Day 7-28? This tells me about the purchase cycle of the product, which informs my scaling strategy and my creative strategy.

If 80% of conversions happen within 24 hours, I’m dealing with an impulse-purchase product and can be more aggressive with scaling. If conversions trickle in over 2-3 weeks, I need more patience and different expectations for new ad launches.

Identifying view-through inflation. I always compare click-based conversions to view-through conversions. In platform view-through is the weakest signal in attribution. A “view-through conversion” means someone was served your ad (maybe for 1 second while scrolling), didn’t click, but later bought something. Meta takes credit.

I try to force Meta to optimize for click-based attribution whenever possible. If a huge percentage of your conversions are view-through, Meta is inflating your numbers. The platform wants you to think your ads are working better than they are so you keep spending.

Limitation: Every platform grades its own homework. Meta will always tell you Meta is working. Google will always tell you Google is working. This is why you never make channel-level budget decisions based on platform reporting alone.

The most dangerous version of this is Google Analytics’ last-click model. GA4 defaults to last-click attribution, which massively favors Google’s own products (Search, Shopping) because they’re almost always the last click before purchase. Of course they are. Someone sees your ad on Meta, gets interested, then Googles your brand name to buy. Google Search gets the credit. But Google didn’t create that demand. Meta did.

Didn’t you think Google built a free analytics platform just for fun, right? Check their stock price if you think they are a non-profit. They built it to favor their ads.

This is exactly why I like post-purchase surveys and MMMs. They tell you what created the awareness, not what grabbed the last click. If you optimize for last-click, you’ll over-invest in branded search and discounts. That’s not how you scale a brand.

Layer 4: The Trunk. Supporting Infrastructure

Model 7: GA4 and Server-Side Tracking

What it is: Google Analytics 4 for web analytics, plus server-side tracking tools (Conversions API / CAPI, Elevar, etc.) that send conversion data directly from your server to ad platforms.

When to use it: Always running in the background. This is infrastructure, not a decision-making tool.

Why it matters:

GA4 is biased. I’ll just say it. It’s a Google product that defaults to last-click attribution, so it over-credits Google and under-credits everything else. I don’t use it as an attribution tool.

But server-side tracking is essential. Here’s why:

After iOS 14.5, browser-based tracking (pixels, cookies) lost a huge chunk of signal. Meta’s pixel can’t see a lot of what happens on iPhones. This means your platform-reported data AND your MTA tools are working with incomplete data.

Server-side tracking (CAPI) sends conversion events directly from your server to the ad platforms, bypassing browser restrictions. This doesn’t solve attribution. It makes all your other attribution tools more accurate by feeding them better data.

Tools like Elevar help automate this setup on Shopify. If you’re spending any meaningful amount on paid media and you don’t have server-side tracking set up, your data is leaking. Fix this before you worry about fancy attribution models.

Think of it this way: server-side tracking is the trunk of the tree. It doesn’t bear fruit by itself, but it holds up everything that does.

Limitation: GA4’s attribution model is still biased toward Google. Use it for web analytics (traffic sources, on-site behavior, page performance) but don’t use it as your source of truth for marketing attribution.

The Base: Setting Your Targets

Model 8: Cohort-Based LTV Attribution

What it is: Tracking customer lifetime value by acquisition source and cohort to understand which channels bring the most valuable customers over time. Not just the cheapest first purchase.

When to use it: Monthly review, quarterly deep-dive. This sets your targets, which then flow up through the entire tree.

Why it’s powerful:

All 7 models above assume you have a target. A CPA ceiling, a ROAS floor, an nCAC goal. But should that target change based on quality of traffic?

It should come from LTV by channel.

Here’s a simplified example. You’re acquiring customers from Meta at $40 CPA and from TikTok at $30 CPA. TikTok looks more efficient, right?

But when you pull cohort data:

• Meta customers: 35% return rate, 3.2x purchase frequency, $180 12-month LTV

• TikTok customers: 20% return rate, 2.1x purchase frequency, $95 12-month LTV

Suddenly Meta customers are worth nearly 2x as much over 12 months. That $40 CPA is a steal. And the $30 TikTok CPA might actually be overpriced relative to the customer quality you’re getting.

This is why I obsess over two LTV metrics above all others: return rate % (percentage of customers who make purchase #2) and purchase frequency (how often returning customers buy). These two drive the majority of LTV variance between channels.

Example scenario:

How this feeds the tree:

Your cohort LTV data tells you what your CPA targets should be by channel or type of ad. That feeds your blended MER/nCAC target at the top of the tree. Which feeds your channel allocation decisions in the middle. Which feeds your ad-level optimization at the bottom.

If you skip this step, you’re optimizing the entire tree toward the wrong target. You’ll efficiently acquire cheap, low-LTV customers. You’ll look great on a dashboard and terrible on the P&L in 12 months.

Why “AI-Powered Attribution” Is Dumb

In my 2026 hot takes I said “AI-powered attribution tools are guessing.” I stand by that.

The pitch from AI attribution vendors sounds great: “We use machine learning to analyze millions of data points and give you the TRUE value of every touchpoint.”

Here’s why that doesn’t work.

They start with biased data. Every AI attribution model I’ve seen starts with click-based data as its foundation. Clicks, page views, ad impressions, conversion events. This data is already biased. It over-credits platforms that generate clicks and under-credits platforms that generate awareness.

Then they layer complexity on top of the bias. The AI takes biased inputs, runs them through a black-box model, and spits out precise-looking numbers. But precision is not accuracy. You’ve just made a biased answer look more scientific. That’s worse, not better.

You lose understanding. This is the part that bothers me most. Attribution is never going to be perfect. Every model has blind spots. The value of understanding attribution isn’t getting a perfect number. It’s knowing where your model is wrong so you can compensate for it.

When I use post-purchase surveys, I know they over-credit memorable channels and under-credit passive ones. I can adjust for that. When I use MTA, I know it’s click-biased and overcounts revenue. I can adjust for that too. When I use MMM, I know it needs scale to be meaningful and the outputs have wide confidence intervals.

With AI attribution? You get a number. You don’t know how it was calculated. You don’t know where it’s biased. And you can’t adjust for what you don’t understand.

The right approach is top-down, not bottom-up. AI attribution tries to solve the problem from the bottom up: perfectly attribute every individual touchpoint, then roll it up to channel-level insights. But you can’t build accurate channel-level insights from biased touchpoint data, no matter how much AI you layer on.

The approach I’ve outlined here works top-down: start with your blended truth (MER/nCAC), validate at the channel level (surveys + MMM), then use MTA for ad-level decisions. Each layer checks the one below it.

My take: spend your money on a good MTA tool, a well-designed post-purchase survey, and eventually an MMM. Skip the “AI-powered” attribution platforms. They’re solving the wrong problem.

The 3 I Actually Use (and How They Work Together)

After all that, here are the 3 that make up my daily attribution stack:

1. Multi-Touch Attribution (Triple Whale) for ad and campaign-level optimization. Which ads to scale, which to cut, which to retest. This is where I live daily inside ad accounts.

2. Post-Purchase Surveys for channel-level allocation. Which channels are actually driving awareness and new customer acquisition? This is what I use to decide where to push or pull budget at the channel level. Updated daily.

3. Marketing Mix Modeling for directional validation. Is our post-purchase survey data sending us the right way? Are there channels that surveys miss or over-credit? Reviewed quarterly.

How they triangulate:

When all three agree (MTA shows strong Meta performance, surveys confirm Meta is the top first-touch channel, and MMM validates Meta as the highest-ROI channel) I push budget aggressively.

When they disagree, that’s the interesting part. Let’s say TikTok looks mediocre in MTA (low click-through, average ROAS) but shows up strong in post-purchase surveys (lots of people say “I first heard about you on TikTok”). That tells me TikTok is driving awareness that’s converting through other channels. I wouldn’t cut TikTok based on the MTA data alone.

Or: Meta looks strong in MTA and platform reporting, but post-purchase surveys show a declining share of “first heard about you on Facebook/Instagram” responses month over month. That’s an early warning sign that my Meta ads are becoming more retargeting-heavy and less effective at reaching new people, even though the dashboard numbers still look good.

The three tools catch different lies. That’s the whole point.

Do This Monday

If you take nothing else from this post:

If you’re spending $10k+/month:

1. Add a post-purchase survey to your checkout (Shopify, KnoCommerce, Fairing, whatever works). “Where did you FIRST hear about us?” Randomize options.

2. Set up your blended MER and nCAC tracking in a simple spreadsheet. Know your maximum and target spend.

3. Switch Meta to 7-day click attribution if you haven’t. Stop relying on view-through.

If you’re spending $50k+/month:

4. Install Triple Whale or Northbeam. Start tracking new vs. returning customer ROAS by ad.

5. Build your Account Control Chart (CPA vs. Spend scatter plot from MTA data). Know which ads are winning and whether they’re actually getting scaled.

6. Make sure server-side tracking (CAPI / Elevar) is set up and firing correctly.

If you’re spending $250k+/month:

7. Start pulling cohort LTV data by acquisition channel. Adjust your CPA targets by channel accordingly.

8. Consider MMM (Robyn if you have data science resources, or a SaaS platform like Measured or Sellforte).

9. Run your first incrementality test (Haus or Meta’s Conversion Lift Study). Start with your biggest spend channel and ask: “How much of this is actually incremental?”

None of these are optional if you want to scale profitably.

Attribution isn’t about finding the perfect model. There is no perfect model. It’s about building a system of imperfect models that check each other.

Start at the top of the tree. Work your way down. Never trust a single source of data.

Hope this was helpful!

Curtis

Targeting and forecasting decisions often assumes the customer is a stable object. And thats a dangerous assumption when you scale that fast.

Backing this up with some research - here are 3 ideas I want to challenge today:

This person is a heavy buyer.

This cohort is worth $340.

This persona is our top buyer

All three change (generally decaying). Not because anyone did anything wrong, but because that’s what buyer behavior does when you scale.

Brought to you by Soku.ai

I built a fake Meta account, hid five traps in 28 days of data, and ran a marketing AI through it.

What it caught

Blended CPA held flat at $37 while new customers fell from 72% of purchases to 49%. Real new-customer cost: $52 to $76. Soku rebuilt the account on new-customer CPA before answering my budget question. Nothing hit the $45 target.

The $19 retargeting campaign, best CPA in the account, was 78% returning buyers. True new-customer CPA: $88.

The 3.6 ROAS winner: 3 purchases, $48 of spend. Noise.

CPA doubled from $38 to $81 while frequency went 2.1 to 4.8. Fatigue, not targeting.

The best hook, 2.9% CTR, converted at 0.6%. Top ad: 2.7%.

Then I pushed three bad calls: the 3-purchase ad to $5,000/day, half of prospecting into retargeting, top ad 8x by Monday. No to all three, with the math.

What it built

Same conversation: replace the tired creative. Claim safety came first, before any concept. That’s the guardrail in a made-up account. In a live one, signals like fatigue, conversion quality and winning angles can feed what it builds and tests next.

Most tools build whatever you ask for. This one made me prove the premise.

In a live workspace, Automations can keep that running: ROAS and daily spend monitored, anomalies posted to Slack while there’s still time to act, budget changes held behind approval.

Try it on your own numbers!

→ https://soku.ai/c/curtis-howland

1. The person in the top slot isn’t stable

The Ehrenberg-Bass Institute has been running this analysis on packaged goods panel data for sixty years. Their landmark work used Kantar household data across 40 categories and nearly 600 brands, replicated across two separate time windows.

First thing they found: the 80/20 rule isn’t real for these buyers. The actual Pareto share is about 40/20 in a single quarter, 50/20 over one year, and 60/20 over five.

Great - it's still a ton/the majority, so you should cater to those people, right??

They say no, Half of this year’s top 20% of buyers will not qualify for the top 20% next year.

Their explanation? Purchase rates regress toward the mean. Someone who bought you six times last year had a year with unusual demand. New baby, a health kick, a move. Next year they buy twice, and someone who bought once buys four times.

The segment of top buyers is real. The idea that it’s the same people year over year is not.

My takeaway?

Rebuild high-value seed audiences on a rolling 12-month window instead of all-time, and treat customer interviews as a snapshot rather than a fixed profile.

Will they look the same demographically or problem wise? Maybe? But maybe not!

If customers regress to the mean, then you want to be careful about oversampling those top customers. Ideas and values may shift and its important to update your understanding of your customers as they shift themselves.

2. The pool gets lighter the better you do

Section 1 is churn inside the segment. This is the segment itself diluting.

Ehrenberg-Bass calls it Natural Monopoly: as brands get bigger, they attract disproportionately more light buyers of the category. Your heavy buyers were mostly in your base early. Growth comes from people who buy the category rarely and barely think about it.

Which means scaling doesn’t just cost more per acquisition. It recruits from a genuinely lighter pool.

Blended LTV falls. CAC rises. Payback stretches.

If LTV is falling and new customer count is climbing, that’s dilution. If LTV is falling and new customer count is flat, you have an actual problem.

What to do?

Forecast the decay instead of explaining it after the fact. If your model assumes flat LTV through a penetration push, you will overspend into CAC.

It also motivates fixing it

1. Increase subscription rates

2. Improve product quality

3. Better customer service and experience

4. Better cross sells and upsells

5. Better remarketing

Just because it trends down naturally doesn’t mean it has to actually go down in practice.

3. Moments are the only thing that holds still

People move between buying rates. The pool dilutes as you win. So what’s actually durable enough to aim at?

The buying situation.

“I ran out this morning.” “I’m traveling next week and a scoop doesn’t work.” “My kid won’t touch a vegetable.”

Those recur. They’re countable. They don’t regress to the mean.

Personas still earn their keep. They tell you who to cast, what the room looks like, how the voiceover should sound.

Practical version, and this takes about twenty minutes: pull your last 20 winning ads and sort them by the situation they open on, if you find a gap vs customer calls or reviews you have your next ad to test!

The short version

Your heaviest buyers reshuffle every year, so the audience may change with them, check it every 12 months and stay updated on your customers.

Your incoming cohorts get lighter the better your penetration gets, so falling LTV during a scale push is the signature of it working. Put in a plan to improve LTV as you go if you dont want it to drop.

Stop asking who your best customer is. Start asking what has to be true in someone’s day for you to be the answer.

People move through life moments and desires, update your understanding of your customers.

Moments repeat. Make ads that speak to them.

Sources

• Sharp, Romaniuk & Graham, “Marketing’s 60/20 Pareto Law” (SSRN, free): https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3498097

• Graham, Sharp, Trinh & Dawes, “The unbearable lightness of buying,” Journal of Marketing Management: https://www.tandfonline.com/doi/abs/10.1080/0267257X.2021.1963308

• Ehrenberg-Bass Institute, “The value of Pareto’s bottom 80%”: https://marketingscience.info/value-paretos-bottom-80/

• Romaniuk, Better Brand Health: Measures and Metrics for a How Brands Grow World

• Sharp, How Brands Grow (2010) and Sharp & Romaniuk, How Brands Grow Part 2 (2015)

What changed my mind wasn’t a thought-leadership post. It was pulling GA4 for a few accounts and realizing the number I was looking at was way off.

Here’s the situation most brands are in. You open analytics, find the AI traffic, see something like 1% of sessions, and file it under “watch later.” Reasonable. Except that number is missing a ton.

Adobe’s Q1 2026 analysis, covering over a trillion visits to U.S. retail sites, found that in March 2026, AI traffic converted 42% better than non-AI traffic. Shopify, working from its own commerce data, reported in May 2026 that AI-referred sessions convert nearly 50% higher than organic search on product detail pages, and that AI beat organic in 23 of 25 merchant categories.

Both of those are vendor-stated numbers, so hold them loosely. But two independent datasets pointing the same direction is worth more than one flattering screenshot.

The mechanism is simple: by the time someone clicks through from an assistant’s answer, the comparison shopping already happened inside the conversation. They arrive as a finalist, not as a browser.

So I asked Lorea Lastiri to build the actual playbook, because she does this for a living and I don’t. Lorea runs Lastiri Digital, has been doing SEO content work for over 12 years, and spends her days on async SEO audits, demand-capture strategy and AEO for ecommerce and growth-stage brands.

What follows is her list. I’ve added my notes where a move overlaps with something I see on the paid side. This isn’t paid. I just think it’s a fascinating topic, and I appreciate her sharing these details and wanted to pass them along! (Graphic additions are mine)

Part 1: Do This Before You Buy Anything

Most brands are getting pitched an “AI optimization” retainer before anyone has checked whether the engines can read the site. That’s backwards.

1. Map the buying decisions, not a vanity list of prompts

Build a fixed set of 20 to 30 questions from support tickets, reviews, site search and post-purchase surveys. Cover discovery, comparison, objections, fit and care.

Then sample each engine more than once and record four things: the answer, whether you were mentioned, whether you were cited, and which sources it pulled from.

A single run is not a baseline. These systems are non-deterministic, and one query on one day tells you nothing you can defend in a meeting.

Curtis note: this is the same discipline as reading creative test results. One day of data on a new concept is noise. You need repeated sampling before you call anything.

2. Fix eligibility

For Google, a page still needs to be indexed and eligible to show with a snippet. For ChatGPT search, the crawler that matters is OAI-SearchBot.

That last one trips up more brands than it should, because OpenAI runs three separate agents with three separate jobs. GPTBot handles training. OAI-SearchBot handles search inclusion. ChatGPT-User handles user-triggered fetches, and OpenAI’s own documentation says robots.txt rules may not apply to it since a person initiated the action.

Most sites that are invisible didn’t block anything on purpose. They have a broad WAF rule, a CDN bot filter, or a wildcard disallow that catches everything before it reaches the content. Robots.txt changes take roughly 24 hours to register on OpenAI’s side.

Check robots rules, CDN blocks, internal links, JavaScript rendering, and whether your key product facts exist as text on the page.

3. Turn the catalog into a product truth layer

Put price, availability, variants, materials, dimensions, compatibility, shipping and returns in visible HTML. Then keep Product and Offer markup plus your Merchant Center feed consistent with what’s on the page.

Schema supports product understanding. It’s not a magic AEO switch, and Google’s structured data policy is explicit that markup has to be a true representation of visible page content. Mark up a price a logged-out visitor can’t see and you’re risking rich result eligibility, not gaining an edge.

Curtis note: if you’ve ever had a Meta catalog throw disapprovals because the feed and the PDP disagreed on price, you already understand this problem. Same fix, wider blast radius.

Part 2: Give the Engine Something Worth Citing

Eligibility gets you considered. Give it something to actually name.

4. Publish the facts only your operating data can supply

Mine return reasons, support questions, product tests, sizing failures, durability findings and usage constraints. That’s material a competitor cannot rewrite from the same public sources.

Generic advice gets summarized without crediting anyone. First-hand evidence gives the engine a reason to link back, because there’s nowhere else to get it.

Curtis note: this is the single most underused asset in DTC. Brands are sitting on post-purchase survey data and return reason codes that would make really useful content. Use it for AEO, but also creative.

5. Build for the follow-up questions hiding inside the first question

Google’s AI features fan one query into several related searches running at the same time. Google published data in March 2026 showing AI Mode queries average three times the length of a traditional search, and at I/O in May 2026 said AI Mode had passed a billion monthly users.

So a strong decision page covers the main answer, the trade-offs, the edge cases, the comparisons and the next step in one navigable asset.

What you should not do is spin every wording variation into its own thin page. That’s maintenance debt at best, and it can cross into Google’s scaled content abuse territory at worst.

6. Write the comparison page your legal team will still like

Own “X vs Y”, “best for [use case]” and alternative queries. Do it with a dated methodology, named criteria, evidence links, and a clear account of who should not buy your product.

Admitting where a competitor wins makes the page more useful to a buyer and more defensible to your counsel than a disguised sales page.

Part 3: Make the Mention Actually Worth Something

A citation that lands on the wrong page is a vanity metric with extra steps.

7. Make product and brand identity boringly consistent

Use the same brand, model, variant and category names across PDPs, Merchant Center, retailer listings, organization profiles and press materials.

When you rename or reformulate a product, preserve the aliases. Otherwise an answer engine treats your old reviews, your new retail listing and your current SKU as three unrelated things, and you lose the accumulated trust of all three.

Bots and people see things differently.

8. Earn corroboration where buyers already ask for help

Reviews, retailer pages, expert roundups, videos and real forum discussions supply independent context your own site can’t.

Start with the outside sources already showing up in your baseline prompts from move one. Fix the wrong facts there first. Then earn coverage there.

Spraying synthetic mentions or blasting generic PR is cargo cult behavior. You’re copying the shape of what works without the substance underneath.

9. Make every citation land on a decision-ready page

A mention isn’t a business result if the cited page can’t confirm the claim, show the right variant, or move the shopper toward buying.

Deep-link your evidence to the relevant PDP, comparison or guide. Keep stock and policy details current. Give the visitor an obvious next step.

Curtis note: this is just landing page discipline, and it’s the same reason a great ad pointed at a generic homepage underperforms a mediocre ad pointed at the right page. Different traffic, same idea.

Part 4: Run It as an Experiment, Not a Religion

10. Measure it honestly, including the parts you can’t see

Track three things: Google generative AI visibility in Search Console, identifiable AI referrals and conversions in analytics, and repeated mention and citation rates for your fixed question set by engine.

Curtis note: Post-purchase survey as well!!

Judge changes against a dated before-and-after sample. Keep SEO, merchandising and conversion metrics in the same scorecard so you can see what a change cost you elsewhere.

This post was co-written with Lorea Lastiri of Lastiri Digital. Lorea has spent 12+ years in SEO and content, and runs async SEO audits, demand-capture strategy and AEO for ecommerce and growth-stage brands. The ten moves are hers. The paid media commentary (and any errors) are mine.

You can find Lorea and her partner program here: https://lorealastiri.com/partners/

Sources

• Adobe Digital Insights Q1 2026 AI traffic report: https://business.adobe.com/blog/ai-traffic-surge-retail-sites-not-machine-readable

• Shopify AI-referred session performance, May 2026: https://www.shopify.com/enterprise/blog/ai-search-insights

• Google Search Console generative AI performance reports, June 3 2026: https://ppc.land/google-finally-gives-search-console-its-own-generative-ai-visibility-reports/

• Search Console generative AI report limitations and history start date: https://www.tryhikoo.com/en/blog/guides/search-console-ai-overviews-impressions/

• GA4 AI Assistant channel launch and coverage gaps: https://www.digitalapplied.com/blog/ga4-ai-assistant-channel-2026-measure-ai-traffic-playbook

• Share of AI sessions arriving without a referrer: https://www.madx.digital/learn/ga4-launches-ai-assistant-channel

• OpenAI crawler documentation overview: https://developers.openai.com/api/docs/bots

• OpenAI crawler documentation changes, December 2025: https://ppc.land/openai-revises-chatgpt-crawler-documentation-with-significant-policy-changes/

• Google general structured data guidelines: https://developers.google.com/search/docs/appearance/structured-data/sd-policies

• AI Mode query length and query fan-out: https://keywordseverywhere.com/news/google-ai-mode/

Take a brand doing $120M in revenue on $20M of marketing spend. MER of 6.0. On paper, one of the healthiest businesses in DTC.

Split it by customer type and the aMER is 1.3. nCAC is $182. Repeat orders are 78% of revenue and the new customers are flat. That business isn’t growing. It’s harvesting. The 6.0 is a rearview mirror pointed at customers they bought two years ago.

I built a model that fixes this. It’s free & linked at the bottom.

The blended P&L hides some key issues

A standard P&L mixes new and returning customers into one revenue line.

Fine for your accountant. Useless for deciding whether to spend more tomorrow.

Returning customers cost almost nothing to convert. Email, SMS, and subscriptions do that work for a fraction of the cost. Average them in with new customers and the cheap revenue subsidizes the expensive revenue until the whole thing reads as profitable.

You need the full P&L to know your profit but you need a new customer P&L to understand the current state of acquisition and if it’s profitable.

It’s the difference between profit today and future profitability.

Meta does the same thing to you. Left alone it harvests the cheapest conversions available, and the cheapest conversions are people who were going to buy anyway.

Everything below is first-time orders only.

Sections 1-4: Get to contribution margin

Four lines. One example brand runs through the whole post.

Gross revenue, first-time orders only. $200k of new customer revenue out of $500k total. 40% of the business is new, 2,000 orders, $100 AOV. Under 30% new and you’re living off the base.

Net revenue. Deduct discounts and returns. 30% discount rate, 12% return rate, leaves $116k.

Gross margin. Deduct COGS of $42,400. Leaves $73,600, a 63% margin on net revenue. If this line is thin, stop reading. No media buying efficiency saves a product that doesn’t make money.

Contribution margin. Deduct $10k shipping, $3,180 in payment fees, $8k fulfillment. $52,420.

That’s per-order profit before you spend a dollar acquiring anyone. No email, no SMS, no retention in it.

Section 5: Your ad spend is not your acquisition cost

Most brands take that $52,420, subtract Meta spend, and stop.

Ad spend was $50k. Acquisition CM reads +$2,420. Thin, positive, keep scaling.

Here’s the same month fully stacked:

• Meta ad spend: $50,000

• UGC and creative production: $7,500

• Whitelisting and creator licensing: $2,000

• Agency fee at 10% of spend: $5,000

• Affiliate and influencer commissions: $1,500

• Marketing software and tools: $1,000

$67,000. A 1.34x multiplier on reported spend. The same month now reads -$14,580.

Per customer: reported CPA of $25, fully loaded nCAC of $33.50, contribution margin of $26.21. Every new customer costs $7.29 before retention does anything.

None of those costs should be cut. Creators, licensing, media buying, and attribution are worth paying for, and brands that skip them usually have worse economics. They’re acquisition costs and they belong in the number you decide with.

Assume true CAC runs 20-30% above platform CPA until you’ve stacked it yourself.

Sections 7-8: Two caps, and which one binds

You need two limits, and they do different jobs.

Cap 1 is a minimum blended MER. It protects cash flow (or caps burn). Derive it from your P&L: 50% gross margins minus 10% OPEX means marketing can be at most 40% of revenue, which is a blended MER floor of 2.5. Our example brand is at 8.3, so cash flow is not the constraint.

Cap 2 is a maximum fully-loaded nCAC. It protects unit economics. Project cumulative profit per customer, pick a payback month you can justify, and the profit at that month is your cap. Ours is $35.

The model reads both every month and returns one signal: room to scale, or CPA constrained. Whichever cap is tighter is your actual constraint.

Section 9: What retention owes you

Two key numbers:

Cohort breakeven is per customer: -$7.29. That’s what retention has to earn back from each buyer.

Cash breakeven is the running total: -$14,580 in month 1, -$141,165 by month 6.

The first tells you whether the business model works. The second tells you whether you can survive proving it. A brand can be fine on cohort economics if LTV is strong and still run out of money.

It’s worth sharing this with whoever owns email, SMS, and subscription. Not “improve retention.” Recover $14,580.

Model this by cohort and policy period, not off a blended historical LTV. Old cohorts bought under a different offer, a different discount, and a different creative mix. They will not predict new ones.

Section 10: What happens if you turn ads off

First-time revenue slowly disappears (ads have a lag but that lag is supported by other ads). Apply your churn rate to what’s left.

At 8% monthly churn, our brand drops from $500k to $276k. A 44.8% decline.

That gap is your burn rate, and it’s the number that tells you whether ad spend is buying growth or renting revenue.

Which of the four are you

Put aMER and nCAC next to blended MER and almost every brand lands in one spot.

Retention coaster. MER 6.0, aMER 1.3, nCAC $182, 78% repeat. Returning customers carry the business. Rebuild prospecting with hooks written for cold traffic and accept a lower blended MER for a quarter.

Leaky bucket. MER 2.3, aMER 2.1, 10% repeat. MER barely clears aMER because there’s no repeat business to average in. Retention before scale.

Too efficient to grow. MER 8.4, aMER 4.6, marketing at 12% of revenue. Can afford far more customers than it buys. Spend until marginal aMER hits breakeven.

Scaling past breakeven. MER 1.6, aMER 0.9, nCAC $238, payback past 12 months. Cut the worst 20-30% of spend, reset to the efficient core, rebuild through creative testing.

Three things to watch out for:

Blending offers. If you sell at multiple price points, build a separate P&L per offer. A $39 trial and a $150 bundle have different economics and blending them hides both.

Using it on first-purchase-profitable businesses. If you have to make money on order one, skip the MER floor entirely and manage to marginal CPA.

Reading one month. Promo months swing acquisition CM hard. The signal is the trend across months, not the level in any one.

Get the model

curtishowland.com/download/first-customer-pnl

Export new vs returning from Shopify, fill the blue cells, and read sections 6 through 8. Everything else calculates.

Your blended P&L tells you what your business earned. The first customer P&L tells you what your business is worth buying.

Hope this helps!

Curtis

Meta’s algorithm doesn’t care if someone bought from you yesterday. It cares about showing your ad to whoever is most likely to convert today.

They know you are just looking at ROAS, but spending on Meta to get past customer to return is a waste of money. Just send more emails.

Getting new customers reliably and profitably is how you scale a business.

Here’s how to fix it.

P.s. this post is refreshed from one originally from January. Im incredibly grateful that there are over 3000 of your reading which means 90% are new since I posted this originally. I’ve added more detail and gone deeper this time around

Facebook Optimizes for Easy Wins

Had a call with a founder spending millions on Meta.

Solid numbers. Good product. Strong reviews.

But they can’t scale their DTC site.

Acquisition on Meta wasn’t profitable.

His question: “Should we run awareness campaigns to reach new people?”

My answer: No. That’s not the problem.

The problem is that even with exclusion lists, Meta is showing your ads to:

• Past customers who remember your brand

• People who visited your site last week

• Anyone who’s already in-market for your product

Because those people convert at 3-5x the rate of cold traffic.

Meta’s job is to hit your ROAS target.

Your job is to make sure that ROAS target pushes toward new customers, not just toward the easiest sale.

Stop Asking for Awareness. Start Measuring New vs Returning.

You can’t manage what you don’t measure.

If you’re looking at blended ROAS, you have no idea how efficient your new customer acquisition actually is.

Here are 3 preparation steps, then how to change your creative strategy to drive more new customers:

1. Install a Multi-Touch Attribution Tool

Triple Whale, Northbeam, Rockerbox - pick one. I’ve used them all, first two are favorites.

These tools tag every purchase as new or returning, then attribute it back to the ad that drove it.

This allows you to see the difference between:

• Which ads drive 80% new customers at $3.5 ROAS ($2.8 NC ROAS)

• Which ads drive 50% new customers at $4.0 ROAS ($2.0 NC ROAS)

The second group looks better in Meta’s dashboard. The first group is what you need to optimize for. (Not to mention differences in tracking)

Teach Meta to find new customers not returning.

2. Add a Post-Purchase Survey

On Shopify, add a survey that asks: “Where did you FIRST hear about us?”

Options: Facebook, Instagram, Google Search, TikTok, Friend/Referral, Amazon, Other (Put in anything you spend money on while keeping it less thna 10 answers).

(Randomize setting = on)

This gives you channel-level data that your attribution tools miss.

On a click basis (platform OR your MTA tool)

• Google crushes / its always a last click

• Facebook does well (the CTA is clear and people know how to click through)

• TikTok is OK (The CTA is sometimes hidden / people may think the ad is organic and search)

• Pinterest/Snapchat barely work

Your multi-touch tool will favor Google, then Facebook, then everything else.

The more likely people are to search the more untracked lift they have. You need to optimize for what is actually driving performance, not what is best at getting people to click.

But if your post-purchase survey shows TikTok driving more first touches than your attribution tool suggests, that’s top-of-funnel traffic you’re under-investing in.

And its not just TOF fluff impressions, you now know its driving conversions too.

3. Set Up Exclusion Campaigns Correctly

Create an audience of all past purchasers. Exclude them from your prospecting campaigns.

You’ll still get 20-30% returning customers because Facebook’s exclusions aren’t perfect. But it forces the algorithm to work harder to find new people.

Then use your attribution data to double down on the ads that drive efficient new customer acquisition within that prospecting campaign.

Keep this updated with a an automatic connection if you can (your CRM purchasers list and shopify past purchasers)

Your Creative Strategy Determines Your Audience

You don’t need awareness campaigns. You need top-of-funnel creative in your conversion campaigns.

There’s a difference between:

Middle funnel creative:

• Shows your product

• Highlights features/benefits

• Assumes some brand awareness

• Leads with logo or brand name

Top funnel creative:

• Starts with a problem

• Explains why you’re the solution

• Builds credibility from zero

• Brand reveal comes after the hook

If your first frame shows your logo before it shows the problem you solve, you’re signaling to Meta: “Show this to people who already know us.”

If your first frame shows the problem before your brand, you’re saying: “Show this to anyone with this problem.”

Same conversion campaign. Different audience signals.

The Metrics you are looking for:

1. High New Customer ROAS

2. Strong Thruplay % (video)

3. Strong CPMr (Cost per thousand reached, i.e. new people)

4. High New Visit % (driving new clicks)

Sometimes an ad that you think is TOF is actually BOF and vice versa, watch the metrics too!

This may not work if you dont have steps 1-3 rolling, otherwise the ads will look like poor performers, you won’t scale them, your creative team will think they lost and you learn from the wrong ads

Efficient New Customer Acquisition:

time to implement:

Step 1: Measure New vs Returning

Set up attribution + post-purchase survey for channel level allocation.

Do not use MTA for channel budget decisions thats a good way to spend a lot on Google.

Step 2: Audit Your Creative

• What % of your ads lead with the problem vs the product?

• What % of your first frames show your logo before your selling point?

• Which ads drive the highest % of new customers?

Compare metrics + creatives to figure out what drives TOF in YOUR account.

Step 3: Optimize for New Customer ROAS

In your attribution tool, look for:

• New customers only

• ROAS > your target (even if slightly lower than returning customer ROAS)

• Ads with spend >$500 (enough data to matter)

These are your scalers. Push budget here.

Step 4: Train Your Creative Team

Show your team which ads get the most spend. Show them which concepts drive new customers efficiently.

They’ll start iterating on those themes instead of guessing.

Step 5: Adjust Channel Budget Based on First Touch Data

If your post-purchase survey shows Pinterest driving more new customer “first heards” dollar for dollar, shift budget there.

If Google looks great in your attribution tool but your survey shows it’s mostly last-click traffic from people who found you on Facebook first, don’t over-invest there.

What This Actually Looks Like in Practice

Let’s say you’re spending $250K/month on Meta right now.

Blended ROAS: 4.0x Total sales: $1M/month

Feels good, right?

Now you install Triple Whale and see:

• 50% of sales are returning customers

• New customer ROAS: 2.8x

• Returning customer ROAS: 5.5x

That 5.5x feels good, but how many of those would return anyway?

It’s not scalable anyway (limited to total customers) and confuses the algorithm too.

Suddenly, your $250K spend is only partially going towards generating new customer revenue at 2.8 ROAS.

Depending on your margins, you may be just treading water.

So you:

1. Audit your creative. 80% of your ads lead with your logo and product shots.

2. Launch 10 new ads that start with the problem, explain your solution, then reveal your brand.

3. Push budget toward the ads that drive >70% new customers at 3.2+ ROAS.

4. Use post-purchase surveys to see that most new customers actually discovered you on TikTok, even though Facebook gets last-click attribution.

Six weeks later:

• New customer ROAS: 3.2x

• New customer revenue: $560k/month

• You can finally scale without hitting a ceiling

Much better position to be in.

You can scale into this without performance dropping off a cliff.

You are growing your business, not just extracting from past customers.

Quick reminder…

Unless you are near $100k/day on Meta

Don’t run awareness campaigns.

Don’t buy impressions.

Don’t optimize for engagement.

Stay focused on conversions. But measure the right thing.

Train Meta to find new customers who convert, not returning customers who were going to buy anyway.

Facebook knows who’s going to purchase. If you ask for Cheap impressions? You’ll get cheap impressions. Don’t kid yourself into thinking Meta is going to give away purchasers in disguise.

Your job is to show it ads that convert cold traffic, not just warm.

That’s how you scale.

Want to try this out?

1. This week: Install Triple Whale or Northbeam (or equivalent attribution tool)

2. Add a post-purchase survey to your checkout (Where did you FIRST hear about us). Have questions randomize + add in an “other” category.

3. Audit your creative: What % leads with a problem vs a product shot? (Aim for 70% of the prior)

4. Compare new customer ROAS across your ad concepts

5. Push budget toward the ads driving efficient new customer acquisition

Questions/Critiques/Ideas? I’d love to hear them! Drop em below!

Curtis

His example was perfect:

• A teammate says, “We need more net new concepts!”

• Someone else replies, “Great, let me whip up some new angles.”

• And apparently, they’re talking about the same thing.

You’ve been in that meeting. The ask goes out for net new concepts, the folder comes back with fresh headlines on the same video, and three weeks later nobody can explain why the big creative push didn’t move CPA.

It’s because one team asked for diverse creatives (because it’s what drives Meta’s performance) and the other team HEARD variations on an ad (designers don’t speak the same language).

This is all literally semantics. But it can affect performance to not be on the same page.

It’s worth getting aligned (at least with your team) otherwise you can’t really set goals.

So I’m going to use Meta’s naming for a consistent framework.

A Simple Six-Layer Taxonomy

• File. The raw asset. A single 9x16 video export sitting in a folder. No targeting, no destination, no ID. Just pixels.

• Ad. Files combined across placements. A 9x16 plus a 4x5 plus a 1x1 becomes one ad in the account. This is the first layer where Meta starts assigning identity (more on that below).

• Concept. Multiple ads that share an idea and production session. One idea, several executions within the same format and style.

• Angle. A specific way of speaking to a problem. An angle can have many concepts (different formats or styles)

• Persona. The person at the top. One human with multiple problems (multiple angles to speak to that persona).

• Product. One product has multiple selling points/angles.

Here’s the whole ladder running through one example, brittle hair:

• Product: Strong Hair Shampoo

• Persona: a busy professional dealing with several hair problems

• Angle: brittle hair specifically

• Concepts under that one angle:

  • A straight video that speaks to it

  • A UGC creator that speaks to it

  • A founder story that speaks to it

  • A static with selling points

  • A static with a discount

• Ads: multiple variations under each concept (Different hooks or ad lengths, skip the minor edits)

• Files: the 9x16, 4x5, and 1x1 exports behind each ad

Product → Person → Persona → Angle → Concept → Ad → File

Quick honesty check: This is one way of doing it. As long as your team is consistent I don’t mind you having a different method. I don’t think my ladder is sacred.

The practical takeaway: every creative request lives on exactly one layer.

• “New headlines” is a file request.

• “New execution of the winner” is a concept request.

• “We’ve tapped out this pain point” is an angle request.

Meta’s System: The Four IDs

Meta has its own internal system that is important to know as well.

• Ad ID. The ad object itself, the specific configuration inside an ad set. Duplicate a winner into a new ad set and you get a new ad ID, even if nothing about the creative changed.

• Post ID. The content people actually see. One post can sit behind many ads, each with its own ad ID but sharing the same post ID.

• Creative ID. The wrapper that ties content to a destination. Same video pointed at a new landing page is a new creative.

• Entity ID. The one Meta never shows you. More on this in a second.

The post ID is worth real money:

• Engagement lives at the post level, so ads sharing one post ID accumulate the same likes, comments, and shares.

• Standard duplication creates a fresh post with zero engagement.

• “Use existing post” carries the social proof with you when you scale a winner.

• I’ve seen the performance gap at 15 to 25% in favor of ads that keep their engagement.

Andromeda / Entity ID

In December 2024, Meta’s engineering team announced Andromeda:

• Runs on NVIDIA Grace Hopper Superchips and Meta’s own hardware

• 10,000x increase in model complexity over the previous delivery system

• 6% recall improvement and 8% better ads quality on selected segments (per Meta)

Quick run down of how it works:

• Before the auction ever runs, Andromeda narrows tens of millions of eligible ads down to a few thousand candidates for each impression, in a few hundred milliseconds.

• If your ad doesn’t survive retrieval, your bid never matters. The auction ranks a shortlist you may not be on.

• To search that fast, Andromeda clusters ads by visual and thematic similarity in a hierarchical index. Media buyers call these clusters Entity IDs. Not official Meta terminology, but it describes the behavior well enough so people use it.

Minor edits don’t earn a new node. A headline color, a music swap, a tighter crop: the clustering sees the same idea and files it in the same place.

Here are variations I believe work

1. A new hook (different way of saying a problem/different context/etc.)

2. Different length of ad (60s vs 30s cut from the same footage)

3. Different creator or location (same/similar script and angle)

4. Refilming with a higher/lower energy level

I will use headline variations on statics to test for fit still but make sure they speak to different problems and people. Color and font variations are completely out.

How to Run This With Your Team

1. Translate every brief to a rung.

• First get aligned on internal jargon

• Then when someone asks for net new concepts, ask one question: new executions of a proven angle, or a new angle entirely?

2. Make the taxonomy machine-readable.

• Ad naming should encode persona, angle, concept, and variation as separate fields.

• If reporting can’t roll performance up from variation to concept to angle, you can’t see which layer is actually winning.

• Without that rollup, you can’t start with performance data as easily

3. Run the concept audit.

• Pull your active ads.

• Strip the cosmetic edits: headline swaps, colors, crops, music.

• Count what’s left. If 20 ads collapse into 3 real concepts, that’s your true test velocity. I aim for 3 variations (real variations per concept) and $10k of spend per concept. That’s $3k of spend per ad but the prior number is more important.

4. Give variations their actual job.

• Variations aren’t dead. Iteration extends a winner’s life, covers placements properly, and helps you learn.

• But variations exploit demand you already found.

• Concepts and angles find new demand, and they’re the only layer that earns new entries into retrieval.

• Budget your creative pipeline accordingly.

Questions/thoughts/ideas?

Reply!

Cheers,

Curtis

Share

And in some way, reaching more people is better, but it only matters if those people might ever become your customer (paying attention, the right audience, and the right timing) or remember to talk about you. Otherwise, it doesn’t matter how cheap it is.

The most expensive ad on earth is cheaper than you think

The Super Bowl. A 30-second spot in this year’s Super Bowl averaged $8 million.

The Super Bowl draws around 128 million viewers (last year’s set a record at 127.7 million). Eight million dollars, divided by 128 million people, times a thousand, is a CPM of about $63.

Sixty-three dollars to reach 1,000 people. During the Super Bowl.

For reference, blended Meta CPMs for e-commerce sit around $13 to $14, and US CPMs run $20 to $23. So the single most expensive placement on the planet costs about three to four times what you are already paying in the feed. Consider the extra views after and the high engagement and it’s a very reasonable spot for a wide TAM product. (Is this a Tide ad?)

Meta CPMs on Black Friday and Cyber Monday routinely run two to three times the annual baseline. During BFCM, you are paying something close to Super Bowl CPMs to reach people scrolling on the couch.

The Super Bowl is not expensive because of its CPM. It is expensive because you have to buy 128 million people at once.

The ladder, from $2 to $1,000

At the cheap end, out-of-home. Billboards run a CPM of about $2 to $16. Radio, $4 to $25. Those are the bargains of the media world on a cost-per-eyeball basis.

Then the middle. Social sits in the low teens to low twenties depending on objective and geography. Broadcast TV runs $13 to $49. Print, $13 to $54. Digital as a whole is all over the map, $2 to $196, because every targeting layer and premium placement stretches the range.

Direct mail runs a CPM of $500 to $1,000. Fifty cents to a dollar per piece. That is 30 to 70 times the cost of a billboard to reach the same 1,000 people.

The ranking by CPM is almost the inverse of the ranking by attention. The cheapest impressions, a billboard you drive past, a banner you never scroll to, are the ones people barely register. The most expensive impression on the list, a physical piece of mail someone holds in their hands, is the one that actually gets read.

Marketing that couldn’t get on that spectrum stops working. For example, airplane skywriting ads from the 1930s. TV CPMs obliterated the cost vs total view ratio, and it died a decade or two later.

Why the cheapest CPM is usually the worst buy

First, cheap impressions are often not real. Industry estimates put bot traffic at 10 to 30 percent of digital impressions. Those bots aggregate in certain areas so your $2 digital CPM is often a ton of bots.

Second, expensive impressions often convert better. Direct mail carries that eye-watering $500-plus CPM, but it averages a 4.4 percent response rate against 0.12 percent for email. Then it stacks, and those people tend to progress through the funnel more effectively as well. That’s the only reason direct mail can sometimes compete (Spectrum, please stop sending me mail, please).

Meta is the cleanest example. CPMs rose roughly 20 percent across 2025. Everyone felt it. But conversion rates rose about 8 percent in the same window, because the delivery system got better at finding buyers. The brands with strong creative absorbed the higher CPM and scaled.

The goal is not impressions. Its outcomes. CPM measures the shallowest end of the funnel, but it’s not really what you care about.

What to do with all this

CPM is an input. It answers one narrow question: what did it cost to be seen. It says nothing about whether being seen was worth it.

So two rules of thumb.

Use CPM to compare inside a channel and strategy, never across channels. A rising Meta CPM against your own Meta baseline in the same strategy is a potential signal. A Meta CPM against a billboard CPM is largely irrelevant in my opinion.

And judge every channel on cost per outcome, not cost per impression. The cheapest way to reach 1,000 people and the best way to reach 1,000 people are almost never the same thing. (Even if the goal is awareness not conversions).

Happy media buying!

Curtis

Sources

• Super Bowl LX ad cost and viewership: eMarketer, https://www.emarketer.com/content/super-bowl-ads-will-reportedly-cost-8-million-per-30-second-slot ; AdWave, https://adwave.com/resources/super-bowl-commercial-cost

• Meta / Facebook CPM benchmarks: Triple Whale, https://www.triplewhale.com/blog/facebook-ads-benchmarks ; Sovran, https://sovran.ai/benchmarks/meta-ads-cpm-by-industry

• Cross-media CPM comparison (Solomon Partners 2025 Major Media CPM Comparison): Adams Outdoor, https://www.adamsoutdoor.com/blog/billboard-cpm-how-out-of-home-advertising-compares-to-tv-radio-and-digital-cpm-in-2026/

• Digital ad fraud estimates: Adams Outdoor (citing Gitnux), https://www.adamsoutdoor.com/blog/billboard-cpm-how-out-of-home-advertising-compares-to-tv-radio-and-digital-cpm-in-2026/

• Direct mail CPM: McCarthy & King, https://www.mccarthyandking.com/the-5-levers-of-direct-marketing/ ; response rate vs email: LettrLabs, https://www.lettrlabs.com/post/direct-mail-stats-2025

So the room’s takeaway was clean: top-of-funnel CTV doesn’t work for us.

Except that test never asked a question about CTV.

Count what was actually new

Let me total up what was new in that “one test.”

A new creative concept, and only one of them.

A new format, a brand focused ad instead of the usual direct-response, USP-led style.

A new channel.

A new objective on that channel.

A new funnel structure

A new measurement instrument nobody had used before

That was never one test. It was a bunch stacked on top.

So the only thing we can confidently say is “This combination of things - did not show results on this test”

When we already have preconceived ideas like “CTV doesn’t work” these tests cement that belief. Now, I’m not a CTV advocate, not by a long shot, but I don’t think this test was anywhere near that conclusion.

A winner is a joint event

When you get a winning ad or channel, you did not prove a variable. You proved a combination.

The product, the angle, the creator, the offer, the format, the hook, all of it lined up on the same ad. The win is the overlap, not any one piece of it.

Which means your probability of hitting on any given test is basically the product of the odds on each variable you’re holding.

P(win) ≈ P(variable 1 is right) × P(variable 2 is right) × P(variable 3 is right)...

If a variable is proven, its term is close to 1. It barely drags the number down. If a variable is new, its term is well under 1, and it multiplies everything else.

Most people get a winner and then keep exactly one ingredient from it. Same product, brand-new everything else.

You need to know which ingredients are the most important.

Not all variables weigh the same

The variables carry very different weight. A change in product or angle can decide the whole ad. A change in audio quality? Less important.

If you made me rank them, top to bottom, it looks something like this:

1. Product

2. Angle

3. Creator

4. Offer

5. Key selling points

6. Filming Location

7. Ad Length / CTA

8. Audio Quality/Lighting (*more expensive =/= better here)

We’ve run regressions across a lot of these variables but it’s different account to account and can change over time as well (Getting a A+ creator or offer may jump it up the list)

So the real rule isn’t “change one thing.” or even “only make one big change at a time” (impossible) It’s know when you are changing multiple important variables and evaluate appropriately)

Quick Example

New creative on Meta, everything else proven, hits at something like 15%, with maybe a 5% shot at a real, big winner. Those are my numbers from my own accounts.

Now stack that YouTube test back up. New creative, call it 15% (generous for new). On a new channel, my gut says maybe 30% that top-of-funnel YouTube is even a fit. On a new strategy, maybe 10% you get an untested strategy right on the first swing. The channel and strategy odds are pure priors, not measured, so don’t hold me to them. But run the multiplication:

Under half a percent for any kind of win. For a big winner, swap in the 5% and you’re at 0.15%.

You ran a test with a sub-one-percent chance of a real result, it failed, and you filed it under “the channel doesn’t work.” Do that too often and you end up over concentrated.

Mistake 1: Changing too much & Learning too quick

Sometimes you have to change a bunch of variables. There are local maxima that you find through iterations, and absolute maxima that you find through jumping and changing a bunch of things

Small iterations can help you climb your local hill, but you have to make a big jump to find a new hill that’s even taller:

Let’s imagine the axes here are channel and ad type for the X and Z axes, and the Y axis is total performance

If you’re on the small hill, let’s call that Google. Then certain keywords and angles might work

But if you want to jump to the bigger hill, let’s call it Facebook ads. You may need a completely different selling point and ad type to make that work.

Don’t hear me say that changing a lot of variables is a bad idea and you should never do it. Rather, you just need to understand that if you’re going to change multiple variables at a time, your probability of success is low. You need to take multiple swings at bat before you make any conclusive decisions about any of those individual variables.

Mistake 2: changing too little & Giving Up

The second concern is making too small a change. Either making a change on a variable that doesn’t matter (like the background color of a static ad) which has very little impact and is not likely to change. If you’re running it on Facebook, it’s going to get grouped as the same ad anyway. It won’t get spend, and it will just fatigue the original winner faster if you force it.

Do your best to understand the most important variable when you find a winning concept. Then do tests where you keep that and change other variables, or keep all the other variables and change only that.

Control Variables, but only the most important ones. Marketing doesn’t have time for scientific rigor.

When you hold the proven variables

Even when you do hold the heavy variables, the hit rate moves a lot.

In one big account, I looked at creatives that called out a specific version of the product, tied to a partner most people would recognize. The product, the offer, and that selling point stayed fixed; we varied only the execution and the creator.

Reading the dashboard out loud, they still only had a 30% hit rate. Release the selling point, hold just product and offer: I think around 15%, still strong. Release everything, the account baseline across all products and all selling points: about 11%.

I’ve watched the same thing from the other side. Take two or three top ads, push them out to ten or twenty creators with one instruction: make your version of one of these. One or two win, the other eight or nine I don’t go back to. The concept is proven, so the only new variable is the creator & execution, which is exactly what I’m trying to isolate.

Why a low bar is the right bar

First, most of marketing operates on a power law, the top 1% drives the bulk of the performance. (i.e. not a standard distribution)

Second is that creative fatigue and competitors copy you.

You need to stay in front of everyone, which means you need to be constantly coming up with new ideas, new concepts, and new angles to beat that fatigue and stay novel.

If you’re constantly increasing the bar on yourself, you won’t see an increase in the percentage of hits because your overall quality will be going up. Instead, you’ll be raising the bar as you go for either higher spend per ad or higher return on ad spend with those ads.

That constant increasing of the bar of performance is what scales a company

What I actually do

Pin one, vary the other. But only for important stuff that is worth controlling. (Again this is not the rigor of a scientific test, we still need to move fast here). OR if you vary multiple variables, test multiple times (So if you vary creator + angle but test 3-4 angles and none hit? Then I feel good moving on from the creator but not the angles).

Port proven assets into new channels. If you want to test CTV, don’t build a brand-new asset on an untested angle for it. Take something already converting and keep as many of the learnings as possible.

When you can’t isolate, sequence. Instead of teasing four confounded variables out of one messy test, run 20% off versus 30% off. If 20 wins, next round is 20 versus gifts. One variable per round, winner carries forward. Slower is better than confused.

Run explore and exploit as two portfolios. When the hit rate drops and old ads have fatigued, pull back to proven variables and rebuild the base. When it feels healthy again, spend it back down on net-new.

My rule of thumb: at $100k a month, 70% new concepts and 30% variations. Past $1M a month you can go 50/50. Never below 50% new.

Change the metric when you change the heavy variables. A brand-new pain point is not going to convert at your six-month-old CAC on day one. If the thumbstop and engagement rates are in line with your BAU creative, and CPA is within 30%, then keep it running and optimize the page while the audience warms up.

Read aggregates, not verdicts. Tag everything MECE, one and only one answer per field: static or video, what’s the hook, what’s the selling point, what’s the problem. Then roll it up. You’ll find a whole category, say educational awareness ads, is carrying you, which is a real finding. One ad dying is not.

Learn from individual wins, learn from aggregate losses.

The certainty isn’t late, it’s unavailable

A founder told me recently he was frustrated he couldn’t find where he had legs, couldn’t figure out the next ad to produce. The honest answer is that you never get there. The moment you know exactly what the next winning ad is, the bar goes up, you scale spend 50%, and now the bar is higher and the challenge is one level harder. The certainty you’re waiting for is what moves the bar and now you are in uncharted land again. Congratulations.

Considering testing a new channel?

Write down everything you are changing and evaluate the probability of success first.

Then produce the right number of tests to give a fair chance of winning BEFORE you start.

Cheers,

Curtis

Attributed to John Wanamaker, the Philadelphia department store magnate. It’s one of the most quoted lines in advertising.

The same line gets pinned on Lord Leverhulme (the British soap baron who built Unilever), and sometimes on Henry Ford or JC Penney, depending on who’s telling it. Ogilvy quoted it in Confessions of an Advertising Man back in 1963 and attributed it to “Lord Leverhulme, and John Wanamaker after him.” Jeremy Bullmore checked the archives, found nothing, and decided it's all probably folklore. Its important enough that people want to think and believe it must have come from someone really smart.

Whoever said it was pointing at the oldest problem in the business: some of your budget is always getting wasted, and it’s not easily transparent.

If you’ve been reading my newsletter for a while, you know I love attribution, and if you haven’t, you should read this one. It’s still my favorite things that I’ve ever written.

2. “The only advantage I want is a starving crowd.” (Gary Halbert)

Gary Halbert was a famous direct mail copywriter. He wrote The Boron Letters, a series of letters to his son Bond, from a federal prison cell in 1984, which is a story in itself.

He sets this one up with a thought experiment. Imagine you and I both run a hamburger stand and we’re competing to sell the most burgers. What advantage would you want?

People usually say better meat, a great location, lower prices, a secret sauce.

Halbert says sure, take all of that. He only wants one thing for himself: a starving crowd.

It’s a simple idea that’s easy to forget: demand does most of the heavy lifting. You can have sharp creative and a clean landing page, but if the people seeing it don’t really want the product, it’s a slog.

1. Pick a market that’s growing

2. Pick a product that has (or could have) demand

3. THEN make a better product

4. THEN market it well

Don’t try to do it backward.

Don’t market something people don’t want (and don’t ask me to do it either).

3. “The aim of marketing is to know the customer so well the product sells itself.” (Peter Drucker)

Drucker wrote this in Management back in 1973, and it pairs nicely with Halbert.

Halbert says find a hungry crowd. Drucker says get to know that crowd so well you barely have to sell to them.

He’s basically saying the goal isn’t cleverer ads. It’s understanding the customer well enough that the product mostly speaks for itself.

In an ad account, the best-performing creative usually isn’t the most creative. It’s the ad that describes the customer’s problem back to them so accurately they feel understood.

How do you find it? It comes from reviews, DMs, support tickets, and the actual words people use when they explain why they bought. CALL YOUR CUSTOMERS.

4. “If it doesn’t sell, it isn’t creative.” (David Ogilvy)

Ogilvy didn’t really come up with this line. He credits the Benton & Bowles agency.

Either way, the idea holds up. An ad’s main job is to sell. If it’s gorgeous but doesn’t sell, it didn’t do the thing it was there to do. If it’s a little ugly and sells, it worked.

The easy trap is falling for a concept because it’s clever. Clever ads make a founder + a team feel ‘cool’ but unless you are selling feeling cool somehow it probably won’t move the needle.

Just don’t make the opposite mistake of saying “if it’s not tracked by a clear click it didn’t sell” that is also not true. Fix your attribution, then fix your creative.

5. “Your brand is what other people say about you when you’re not in the room.” (Jeff Bezos)

Remember that half of your budget you can’t quite measure? A good chunk of it lives here. There is a reason I put this right after #4, these quotes need to live side by side in your brain.

Your brand is the stuff people say about you when there’s no pixel around. The word of mouth, the recommendation in a group chat, the “oh yeah, I’ve heard of them” that quietly makes your next ad convert a little better.

Attribution is never going to fully catch that. Your last-click report can’t see the friend who vouched for you over dinner.

For anyone running paid, I focus on ads that do three things at once.

1. Drive New Purchases (We know they are compelling)

2. Drive New Reach (Not just retargeting)

3. Speak to selling points (Not just promos)

Think about it. Of the people who watch 15s of a Meta ad (thruplay) 99%-99.5% won’t purchase (at least directly). If NC ROAS is similar, pick the ad that uses that 15s to speak selling points to new people NOT promos to the same people over and over.

Why these five stuck around

These are the things that don’t change.

• Find demand.

• Know the customer.

• Figure out how to sell.

• Build a good brand.

• Realize that you can’t measure perfectly.

If this was useful, send it to your team or share on Social!

Share

Best,

Curtis

Sources

• Bullmore / archives (“no such reference”): Marketing Society, “Why it’s time to say goodbye to IKTHTMISOAIW” at marketingsociety.com/the-library/why-it’s-time-say-goodbye-ikthtmisoaiw

• Wanamaker attribution history: Quote Investigator at quoteinvestigator.com/2022/04/11/advertising/

• Ogilvy citing the line in Confessions + Bullmore context: Conversable Economist at conversableeconomist.com/2018/07/25/half-the-money-i-spend-on-advertising-is-wasted-and-the-trouble-is-i-dont-know-which-half/

• Ogilvy / Benton & Bowles origin: Doug Garnett’s Blog at dsgarnett.wordpress.com/2014/04/07/the-best-ideas-come-as-jokes-but-the-best-quotes-are-sourced/

• Halbert, “starving crowd” (The Boron Letters, Ch. 6): bagerbach.com/books/boron-letters/

• Drucker verified sources (Management / The Essential Drucker): Hammock at hammock.com/marketing/favorite-peter-drucker-quotes-about-customers-and-marketing/

I check three charts before anything else. They answer three questions, in order:

1. Do you have clear winners that are scaled? (Concentration/Optimization)

2. Are you producing enough winners, and are they winning at a healthy rate? (Hit rate)

3. Do those winners last, or do they burn out fast? (Creative half-life)

Run all three and a gap almost always shows up in one of them. Even if it’s not the key thing causing your issues, you’ll still likely find some learning to push into.

One note before we start: this is our internal dash, but the charts below come from demo data, which are realistic examples, but the numbers will not tie out across the three images.

1. Concentration: How Much Rides on Your Best Ads

What it is: take every ad that spent in a window, rank them by spend, and look at what share of total spend the top slice controls.

What you will almost always see: the top 10% of ads should be north of 75% of spend. In a lot of accounts it is higher than that.

Your goal is to balance concentration. Around 40-50% on the top 1% and 75% on the top 10% is good concentration in my experience. It means Meta is doing its job. The platform runs on a power law, where a small number of ads produce most of the results, and delivery concentrates budget into the ads that clear the bar. Those top ads genuinely outperform, so you want them taking the largest share of spend. Flat, even distribution is the warning sign, because it usually means the algorithm has not found clear winners to scale into, or you are manually spreading budget across mediocre ads.

If you are seeing 70-75% of spend on the top 1% of ads, that is also a warning sign that you are over-concentrated and at risk of fast fatigue.

The reason this chart goes first is that once you accept a handful of ads carry the account, everything else follows.

You are not building a big library of decent ads. You are continuously producing the few that Meta will pour money into, and then replacing them as they fade.

2. Hit Rate: Are You Producing Enough Winners

Each month is a bar chart. The dark color is the hits, and the gold line is the percentage rate

What it is: of all the ads you launched, what percentage became winners? The cleanest way to define a winner is by spending (assumes good media buying). An ad is a hit if Meta gave it real budget, meaning its spend cleared some multiple of your median (10x median is what I use).

Why define a winner by spend instead of ROAS or CPA? Because Meta’s own budget allocation is the least noisy signal you have. If the algorithm poured spend into an ad, it found the ad worth scaling. You do not have to untangle attribution to know the platform voted for it. Plus Scalable ROAS is more important than early ROAS.

This chart answers two questions: are you launching enough ads to produce the winners you need, and are they hitting at a healthy rate.

The benchmark: a healthy monthly hit rate sits around 10% to 15%.

• Below 10%: you are producing too much. You are flooding the account with ads, most of which never get off the ground, and diluting your own signal.

• Above 15%: you are producing too little, or without enough diversity. If a high share of what you launch ends up getting spend, you usually are not giving the algorithm enough candidates to choose from, so you end up forcing budget into ads instead of letting the best ones earn it against a wide field.

Read the level and the trend together. If you were scaling well at 10% hit rate and then it dropped to 5%, focus on the drop, not the absolute number.

Then read it against volume. Because so few ads carry the account and winners fade (chart three), you need enough absolute hits every month to replace what is churning and add more on top. A perfect hit rate on too few launches still will not produce enough winners to grow.

* one important caveat is that this chart naturally drifts down. For example, if you are in the middle of August, there are some ads that you’ve launched that are not yet hits and will become hits in the next few weeks as they spend more. You can either adjust this chart to look at the first 7 days of spend, or just get familiar with how the month progresses over time as it matures

3. Creative Half-Life: Do Your Winners Last

Each band is a creative vintage, the month it first spent. Every cohort peaks and then decays.

Drill into one vintage to see its half-life. This January cohort launched in January and kept bleeding spend for months after.

What it is: stack daily spend by the month each ad first spent (its vintage), and watch how each month’s cohort rises and then decays. Click into a single vintage and you can trace exactly how long that batch of creative held up.

What you will see: creative fatigues. Every cohort peaks and fades. Nothing spends forever.

The benchmark I use: creative half-life is roughly 60 to 75 days. Put another way, expect about 25% to 30% of a cohort’s spend to churn each month.

Say you spend $100,000 on a set of ads this month. To hold the same CPA next month, you should expect only $70,000 to $75,000 from that same set, because a chunk of it will have fatigued. To keep CPA flat, you have to replace the $25,000 to $30,000 that fell off. To grow, you have to replace it and add more on top of that.

That is the treadmill, and it is why hit rate and volume are never one-time efforts. You are not filling a library, you are refilling a leaking bucket.

What to do here: drill into your cohorts and find the ads that outlasted the half-life. Study what made them durable, then reverse-engineer the system that produced them, not the exact ad.

Putting It Together:

If you are struggling to scale, this is the scan:

1. Concentration: is spend actually concentrating into a few winners? If it is flat, Meta does not have clear winners to scale into or you have a structure issue.

2. Production: are you launching enough, hitting in the 10% to 15% band, and are those winners sustaining?

3. Durability: are your winners lasting 60 to 75 days, or dying faster? (i.e. you need more creative for any number of reasons)

When you find the gap, go back and diagnose it directly:

• Which month dropped off?

• What were you producing that month versus your best months?

• What is different, both in output (how much you shipped, and how much of it was variations versus net-new concepts) and in systems (how you were researching, briefing, and iterating)?

A couple of common causes worth checking.

You might be running too many variations. If you have already researched a concept heavily and squeezed it hard, more variations of it hit diminishing returns fast. That is often the culprit when output is high but hit rate is falling.

Or you might be running too many net-new concepts, spraying fresh ideas without building on what actually worked.

Either way, the fix is not to copy your old winners line for line. It is to understand the system that produced them (the research, the angle-finding, the process) and re-run that system. The winners were an output. The system is the thing you rebuild.

This is why drilling in matters. Click into a strong month and study the top ads and the creators behind them. Click into a weak month and ask what you were making and why it did not land. You learn far more from the structure of a good month next to a bad one than from any single ad.

Hope this helps!

Curtis

The problem is the freshest formats are usually the most expensive to make. A new UGC angle you can brief in an afternoon. But an animated, Paper-style ad, the hand-cut paper-collage look, has always meant a motion designer, a timeline, and a real invoice.

I made a Claude skill, Script to Paper Style, that does the whole thing in one shot. You point it at a brand and a direction, it does the research, makes an animated ad, and hands it back. One session, no tool-hopping, a new format to throw into testing.

It’s free. The download’s here if you just want it. The next paragraphs are worth skimming if you do test it, though, you need to plug in your own API key.

Example output below as well.

Fair warning up front: this isn’t going to replace your UGC or your studio work, and it’s not trying to. A good editor will be able to improve on it. The goal is that it accelerates your speed to completion.

It all happens inside Claude

The thing I like about it: you don’t leave the chat, and you don’t glue five tools together.

You’re the director, Claude’s the crew. You give it the brand and the direction, it researches the angle, picks a paper metaphor for each beat, generates the frames, animates them, and hands you a finished vertical file. Research, render, done, in one pass.

That’s the whole pitch on “one shot.” Usually “AI video” means bouncing between a script tool, an image tool, a video tool, and an editor, stitching it together yourself. Here you steer, the scripts handle the plumbing, and an ad comes back.

How it works

Under the hood it’s five steps, and Claude runs all of them.

Split the script into beats. A 30-second ad is usually five, one visual idea and one line of voice-over each.

Each beat becomes a paper-collage still (Gemini’s image model, the one people call nano-banana). One frame, one metaphor, no text on it.

Each still gets animated into a 4 to 6 second clip with Veo 3.1, using the still as the first frame. The image sets the composition, Veo adds the motion.

Each line of voice-over runs through Gemini’s text-to-speech.

Then ffmpeg stitches the clips, drops the voice on top, ducks the sound effects under it, and spits out one MP4.

Two ways to run it: let it generate the voice for a finished ad, or skip that and take a fast-cut b-roll bed you narrate yourself.

The only part that costs real money is the video. Veo bills per clip and each one takes a minute or two, so a five-clip ad is five Veo calls. Stills and voice are pennies.

What it actually spits out

Is it perfect? No. Could an editor improve it? Yes.

Did it take 15 minutes in the background for a novel format and a decent ad? Yes.

Should you test it? I think so.

Why the one-shot is actually testable

Novelty is a cheap lever most accounts ignore. Meta rewards fresh creative. In a feed full of talking heads and statics, something that looks completely different buys you a beat of attention you weren’t getting.

More in an account is good.

What you need to run it

The skill’s free. Running it isn’t quite, because it calls Google’s API and the video step bills per clip. Figure a few dollars of Veo for a full ad, not a subscription. Generate one clip first to see the cost before you fire off all five.

You’ll need Node to run the scripts, ffmpeg for the audio (brew install ffmpeg on a Mac), and the skill dropped into Claude Code, which drives everything.

HOW TO GET YOUR API KEY

Go to Google AI Studio at aistudio.google.com/apikey and make an API key (free and often comes with credits to start). Drop it in a .env file in the project folder, one line: GEMINI_API_KEY=your_key_here. That’s it. The scripts read it from there.

Then hand Claude the brand and the angle and let it go.

Go test something new

That’s the whole thing. It’s free, it’s fast, and it makes a format your account probably doesn’t have.

It won’t win every test. Nothing does. But it’s 15 minutes to put something different in front of your audience, and that’s usually worth a slot.

Grab it here. It’s the third skill in the series. Go make something and test it.

Use this list to make more ads, not less.

The Hook (Your First 3 Seconds)

The open is a separate product from the rest of the ad. Spend lots of time here. Think about engagement. Test a lot. But focused on qualifying people over entertaining them.

1. Lead with the problem, not the product. Name the exact pain in the first line before you show what you sell.

2. Get motion into the first 8 frames. A static product shot loses. A zoom, a swipe, a hand entering the frame wins.

3. Put on-screen text in the first second that names a problem or makes a claim. “I stopped buying foundation” beats “New from Brand X.”

4. Open on a human face. Faces lift thumbstop roughly 30% over product-only openers, per Meta’s own Creative Hub guidance.

5. Try a curiosity gap or open loop. Start something the viewer has to keep watching to close.

6. Ask a direct question that triggers a silent yes or no. The internal answer is engagement before anyone clicks.

7. Use a pattern interrupt: street interview, subway clip, behind-the-scenes. Break the visual rhythm of the feed so it reads like anything but an ad.

8. Hook first, brand later. Hold the logo until 5-8 seconds in. Nobody stops scrolling for your logo.

9. On a static, put your strongest element in the top third. That’s the part that survives the crop and the half-second glance.

10. Read hook rate as a triage metric, not a trophy. Under 15-20% is invisible, 25-30% is solid for cold, 30%+ earns scale, and Reels run 5-10 points higher than Feed. Rework anything under 25%. Over that - just focus on qualifying. Don’t aim for 80%.

11. Cut the dead first second from your losers. Most of them waste a logo or a slow zoom before anything happens, and you’re paying for it.

Formats & Pattern Interrupts

12. Master the four foundational formats before anything trendy: UGC testimonials, statics, short video, carousels. These should be 70-80% of your library.

13. Don’t sleep on statics. They drive 60+% of conversions on a lot of accounts, produce faster, and iterate easier than video.

14. Try the Snapchat-style 3-second lead-in. A quick, raw, real-moment open makes the whole ad feel organic instead of produced. It came up as a top pattern in a creative call this week.

15. Run skits and relatable in-car testimonials. Both are consistently top performers right now. Put them in every test batch.

16. Use multi-frame, fast-cut scenes with one unexpected stop moment. A seasonal product ad I broke down this week won by stacking quick scenes around a single visual surprise, then a clear CTA.

17. Test the pet-led hook: a dog POV, an unexpected pet, a “my lab drinks my mushroom coffee” angle. High uniqueness that still lands the benefit.

18. Spin one winning angle into 15-20 variations before you reinvent. Your winners are worth exhausting, not abandoning. Just make them visually different so they hit different audiences.

19. Recut your long-form winners tighter. An old 90-second testimonial that’s fatiguing often comes back recut to ~40 seconds for current pacing. Have extra footage? Do the opposite!

20. Save carousels for education and comparison products. Card 1 hooks with a problem, the middle builds tension, the product and CTA wait for the last card.

21. Build 9:16 first, then 4:5. Vertical covers Reels, Stories, and Feed.

22. Match the format to the audience’s emotional state. For an anxiety product, the format has to evoke the anxious moment while still feeling fresh.

Production: Lo-Fi, Polished & AI

23. Default to low-fidelity, phone-shot, creator-driven content. Smartphone-quality beats studio about 80-90% of the time in my experience.

24. Remember low-fi means low-fidelity, not low-effort. The selling points, the editing, and the creator still have to be sharp. Sloppy is not a style.

25. Know why polished usually loses. High production signals “this is an ad,” and viewers skip ads. Reserve studio for the categories that reward it: luxury, complex demos, some fashion.

26. Test captions - sometimes they outperform sometimes they dont. Test by category not for the whole brand (organic content may perform better without it for example).

27. Keep DR videos to 15-30 seconds. Past that you’re paying for watch time that rarely converts cold.

28. Use AI where it’s actually good: short hooks, product animation, voiceovers. That’s where it earns its keep today.

29. Don’t ship a fully AI-generated ad without a human editor. They don’t scale clean, they get copied fast, and the comments flag them.

30. Layer AI on top of real production, don’t replace it. The hybrid (AI generates, humans refine, AI scales) is what’s outperforming today, not pure AI.

31. Great place to use AI is background generation, image expansion, text variations. One brand saw +45% ROAS from AI backgrounds alone at zero added cost.

Messaging, Copy & Offers

32. Talk to one person. “You” beats “customers,” and a message built for one reader lands harder than one built for a market.

33. Lead with the benefit, not the feature. What it does for them, before what it is.

34. Be specific. “Lost 23 lbs” beats “lose weight.” Specificity is the cheapest credibility you can buy.

35. Stack multiple selling points on screen. A dog supplement ad I reviewed this week won by calling out competitors, listing benefits on-screen, and adding personal testimony.

36. Use “because” to give a reason. People comply far more often when a reason follows, even a simple one. Don’t make a claim you could be backing with a “because.”

37. Address the top objection inside the ad. Name the doubt and answer it on screen before they think it.

38. Power words work, until they don’t. Free, new, proven, limited, instant, guaranteed all pull, but stack seven of them and the reader is numb by the time you reach the CTA. Use one, not all of them. Make it convincing not cheap.

39. Use review language verbatim. Your customers already wrote your best copy. Lift their exact words.

40. Make the offer visible inside the creative. Show the dollar-value deal and the bundle math on screen. Don’t make people read to find the value.

41. Use “free” where it’s honest, and check the law where it isn’t. It’s one of the most persuasive words but also regulated.

42. Give one clear instruction, not three. One imperative CTA: shop now, get started today. Say it on screen and out loud.

43. Add a risk reducer next to the CTA: Afterpay, free shipping, money-back. Remove the last objection at the moment of decision.

44. Give gifts that feel high quality - cheap gifts can be a drag on performance. Cost =/= high value.

Proof & Trust

45. Lead with social proof in skeptical categories. When trust is the barrier, a customer’s words open the ad better than yours do.

46. Use real testimonials in real settings. The whole reason UGC works is that it’s hard to fake. Real customers over actors, every time.

47. Demonstrate the result. Show the transformation or the product working. Don’t just claim it.

48. Run whitelist and partnership ads from real, credible pages. Ads from an actual persons page beat fabricated publisher pages.

49. Test out partnership ads for as many as you can - founder ads, employee ads can work as well!

50. Use credible logos and press as trust shorthand. “As seen in” does quiet work in a half-second glance.

51. Put proof in the body as evidence, not as a second intro. One hook, then proof that backs the promise, not two openings competing.

52. Let the creator’s authentic voice carry believability. The moment it sounds scripted, the trust evaporates.

53. Show the product in use, not sitting on a table. Context beats catalog.

Structure, Pacing & Hold Rate

54. Track hold rate, not just hook rate. Two ads can share a 30% hook and post a 6x ROAS gap because one held 22% of viewers to the CTA and the other held 4%. In 2026 a hook is required, the second 3-15 seconds (thruplay) is where the money is.

    

55. Find the cut where viewers leave and put a pattern interrupt there. Diagnose the dead zone before you reshoot the whole thing.

56. Use the arc: hook, problem, solution, proof, CTA. After you name the pain, move straight to proof. Don’t meander.

57. One core idea per ad. Don’t split the viewer’s attention. Confused buyers don’t buy.

58. Earn each next second. Every beat has to justify the one before it or you lose the viewer.

59. Put your strongest proof right before the CTA. Send them into the decision on the highest note.

60. Cut dead air. Fast transitions, no lulls. The pace itself is a retention tool.

61. End on the offer and a single action. The last frame is the one they act on, so make it the offer, not your logo.

62. Design the first frame as if it were the whole ad. If it can’t sell on its own, the rest of the ad rarely gets seen.

Creators & Briefing

63. Use the hybrid brief: share your top 3-5 ads plus the 3-5 prioritized selling points, then ask for authentic variations. Control plus freedom is where breakout ads come from.

64. Loosen the brief as the creator earns trust. Winners come from creators you let run, not creators you script.

65. Match the creator to the persona, not the follower count. Fit beats reach for performance creative.

66. Be cautious with expensive influencers unless the fit is tight.

67. Consolidate down, not up. Three to five higher-quality creators plus one DR agency beats a long tail of small ones. Test wide then consolidate down once you find the winners. Repeat, don’t keep them all rolling.

68. Run partnership ads from the creator’s own handle. (Yes this is worth repeating)

69. Source creators continuously, and run a small paid test before you scale any of them. Don’t bet budget on an unproven voice.

70. Brief on the emotional state, not just the product. Tell the creator the feeling to evoke and let them find the words.

71. Circulate winning creator cuts internally. A great hook from one brand is a starting point for another. Build the library.

Differentiation & Feeding the Algorithm

72. Be different from your own past ads. Similar creatives compete in the auction and trigger sequencing that quietly kills both.

73. Be different from your competitors. Novelty to the viewer is what lowers your CPMs.

74. Put the hours where the leverage is. If creative is 70-80% of performance, your team should spend its time there, not rerunning budgets.

75. Make it entertaining or genuinely engaging. A strong hook rate alone can pull CPMs down 30-40%, and engagement earns organic reach that compounds your paid scale.

76. Run 8-15 conceptually different ads, not headline swaps. The algorithm penalizes redundancy now.

77. Match big-spend creatives to a big audience and a top product. A niche SKU can’t sustain a million-dollar creative. Top-funnel has to speak broadly to justify the spend.

78. Vary before you repeat across platforms. Running the identical cut everywhere accelerates fatigue and sequencing penalties.

79. Start broad (10M+ interest audiences) and let Advantage+ expand. If you’re still building 500K detailed stacks, you’re over-constraining the algorithm. Turn on Advantage+ creative features while you’re at it, they average +22% ROAS.

80. Treat your best organic post as a paid ad you haven’t run yet. The organic-paid line is gone. Ask your team to forward the organic posts beating Meta and test them.

Testing, Scaling & Portfolio

81. Launch enough volume for the top 1% to emerge. Don’t optimize the median ad. Creative returns are fat-tailed, and a handful of breakout winners carry the whole account.

82. Don’t kill near-winners too fast. Because the returns are fat-tailed, the ad that looks mediocre on day three is sometimes the one that scales. Give it room.

83. Test 3-5 hooks per concept and let thumbstop rate pick. The cheapest way to find a winner is to vary the open, not rebuild the whole ad.

84. Aggregate before you judge. Single-ad reads are noisy. Wait for 10-12 purchases per concept before you decide on something important like an angle.

85. Kill at 2-3x target CPA with no purchase. Don’t nurse a clear loser hoping it turns.

86. For top-funnel creatives, tolerate 10-20% over CPA target when the qualitative signals (comments, watch time, engagement) say product fit. Pull back around 40-50% over.

87. Overscale your proven top-funnel winners. The best ones lift overall account performance 10-30%, not just their own line.

88. Force-test your under-spent, high-efficiency ads. Meta starves some good creatives. Pull 6-10 of them (look back 90 days, include paused), drop them in a dedicated ABO set for 5-7 days, then cut and scale. If one won’t scale, the cause is usually one of three: not enough data, sequencing, or decay during scale, and each has a different fix.

89. Volume tends to pay. Shipping 10+ fresh creatives a week often cuts CPA 20-40% over time, because you’re giving the tail more shots at a breakout.

90. Study your longest-running ads. Longevity is a winner signal. The ad that’s been live a year is telling you what your account actually wants more of.

Fatigue & Measurement

91. Run the production math: one new ad per $3,000 in monthly spend. At $150K/mo that’s ~50 new ads a month just to stay even, and 50-100+ active.

92. Watch the fatigue signals: CTR under 1%, a 20%+ week-over-week CTR drop, frequency over 3.0 on prospecting, or CPA rising with nothing else changed.

93. Don’t over-rely on a single hero creative. Double down AND diversify.

94. Use publisher whitelisting to revive fatigued creative.

95. Don’t read creative performance off click tracking alone.

96. Use the right tool for the question. MTA for ad and campaign-level creative decisions, MMM and incrementality for channel-level allocation. They answer different things.

97. Tag your creative consistently. A clean taxonomy beats a clever one, make it functional

98. Build the promo playbook outside black friday. Put ~20-30% of spend on dedicated promo ads (mostly your top performers with promo banners) and update the site copy for urgency. Use seasonal hooks beyond Black Friday: Memorial Day, Fourth of July, even a birthday-month angle.

99. Treat the landing page as part of the creative. Run PDP copy and CTA tests measured on conversion rate and revenue per visitor.

100. Trust the data over your taste. Your winners often won’t look “good” to you. Ugly ads beat polished more often than anyone wants to admit, and the longest-running, highest-spending ad in your account is usually the proof. Ship the thing the numbers like.

That’s the list. There will be exceptions to them all, thats the joy of marketing, hope it helps!

Share

These were the requirements.

1. Answer quickly, and accurately.

2. Live data: Updated every day in the morning

3. Track performance across different attribution models

4. Be able to review engagement metrics for each creative.

5. See the creatives itself as a link that I can view and embed

6. Descriptions of the creatives that are searchable

7. Use AI tags for all creatives. (type, format, angle, hook, etc.) that are MECE

8. Facial tracking recognition to group by creator

But the more data I added the more AI hallucinated as I was building.

What finally fixed it wasn’t a better prompt or automation system. It was two things working together, and the combination is the whole point of this post.

I put the data into a real database with a clean schema, and I let Claude Code write actual SQL against it (sending reports to slack) instead of asking a model to eyeball a pile of exports.

Then I built a visual analysis for repeat queries or where text answers didn’t cut it.

That’s the unlock. The AI stops guessing because it can write queries to isolate specific data instead of parsing large data sets as an LLM and pair that with a dashboard to see the results, and I can now ask almost anything about the account and get a real response.

I built this around the individual ad creative as the unit. Each ad carries its own lifecycle, its spend and revenue split into new versus returning customers, a transcript of what it says, a description of what it shows, who is on screen in it, and how it decays day by day. The data gets in either through a CSV importer (old way + for testing) that auto-detects Northbeam, TripleWhale, or Meta exports, and an automated pipeline that pulls on a schedule and generates the AI tags.

That’s the plumbing.

Here is what the database really looks like:

Claude has instructions on how each table works and is used and so when I ask a question, it filters down to the right data first then it pulls specific data, and then qualitatively analyzes with just the right data. No context window issue, no hallucination.

With that said, here are some example questions we are building on.

Example Question 1: Which rung is the leak

I can pull one ad, or a whole category, and see the full path from impression to purchase. Play, 3-second hook, holds at 25, 50, 75, 95, and 100 percent, click, landing-page view, add-to-cart, checkout, purchase. It flags the single biggest leak, weighted by spend.

This is the difference between a guess and a diagnosis. A soft ROAS number tells you an ad is underperforming. The funnel tells you why. Great hook and no completion is a creative problem. Great completion and no add-to-cart is a landing-page or offer problem. Opposite fixes. You can only tell them apart if you kept every rung, and the SQL is what makes “show me the leak across these 40 ads” a five-second question.

In platform visual:

Question 2: Which specific creator actually drives results

The account knows who is on screen in every ad, with the timestamp of their first appearance in video. So I can ask “how do ads featuring this exact person perform” across the entire account and get a straight answer.

We still tag ads by creators in the name often, but this is helpful for accounts with 100’s of creators or inconsistent historical tagging.

Creators are one of the biggest levers you have, and “which one is actually working” is important.

Question 3: Whether to even trust Meta’s ratings

I can correlate Meta’s own quality rankings against actual ROAS and CPA and get a verdict, per account: predicts, no signal, or inverted.

(To be honest, not a ton of insight but fun to analyze!)

Most importantly: Ask it anything, at once

Every one of those used to be a separate spreadsheet.

Now I open the database, talk to it in plain language, and say “go to this brand, pull the top ads in this category, tell me what’s working and what the selling points are, then send me the three best.” It writes the SQL, runs it, and comes back with a real answer.

I can move more quickly, dive deeper, and answer more this way.

I still have to validate that it answered the right question, that the tests were set up right, and that the underlying data is right.

But that was true of spreadsheets as well. Bad data in = bad answers out.

Want to build it?

Ask Claude to help you build each piece. Just dump this article into Claude to start. That is genuinely how I would do it.

set up a daily Meta Graph API pull. Use Supabase as your database and a Supabase-based dashboard. You can build the dashboard with Claude, and use Claude Code to interact with both Supabase via MCP and the dashboard to update it.

Meta app set up is annoying but totally doable.

What to skip for now: transcriptions, visual descriptions, facial tagging, AI tags, & Northbeam or Triplewhale imports for different models.

Add those in later.

The middle version (add the creative intelligence). Layer in the AI tags so every ad is classified the same way without you typing it, and add transcripts and visual descriptions so the creative becomes searchable by what it says and shows. This is where you can start asking questions about angles and hooks across hundreds of ads, not just spend and clicks.

The full version (everything talking to everything). Add facial recognition to group by creator, and pipe in Northbeam or TripleWhale so the model is reasoning about incremental revenue, not just in-platform numbers.

You could skip all of this and just use an existing tool as well. That is a legitimate choice, and if that is your preference, there are plenty of good ones.

Just don’t split the middle and point claude at a huge spreadsheet. That’s where you will run into issues.

Liked this post? Share it with a friend or on Social! Thanks!

Share

Chase’s line: AI made content cheap, and now all the cheap content sounds the same.

His take is that brand voice is the one thing AI can’t copy at scale, because it comes from a point of view, not a prompt. He lays out three ways to protect it.

Read the full post on LinkedIn →

2. Marissa Rodriguez: evaluation is the new bottleneck, not execution

Chase says protect the voice. Marissa pushes on the harder question: can you even tell which AI output is right?

Anyone can generate ten versions of an ad in a minute. The rare skill now is knowing which one is actually good, and being able to explain why.

The strategists worth the most aren’t the fastest at making things. They’re the ones with taste sharp enough to evaluate what will work.

Read the full post on LinkedIn →

3. Marissa Rodriguez: stop asking what AI saves you, ask what you’d attempt if it were easy

Most founders ask how much AI can save them. Marissa’s better question: what would you attempt if this were easy?

The savings question keeps you doing the same work for less. The other one opens up work you used to write off as too expensive to bother with.

Not everything you do should be automated / but plenty of things you want to do no can be…

Read the full post on LinkedIn →

4. Michael Kaminsky (Recast): your geo tool is an error tradeoff in disguise

Michael ran 32,000 simulated geo-experiments through the four main open-source tools, then color-coded which results actually contained the true lift.

I’ll let the post speak for itself. I keep thinking about this one...

Read the full post on LinkedIn →

5. Zach Sherrod (Northbeam): the bucket that lies to you

Straight from Northbeam: when you move off last-click, your “Organic + Direct” bucket drops by around 74 percent. Not because organic vanished. Because paid was driving most of it the whole time, and last-click kept handing the credit to the wrong line.

Read the full post on LinkedIn →

6. Charlie de Thibault: a chunk of your budget isn’t underperforming, it’s doing nothing

Charlie’s claim: 20 to 30 percent of your conversion budget probably isn’t underperforming. It’s marginally a profit loser.

Great post on a simple way to think about marginal impact.

Read the full post on LinkedIn →

7. Charlie de Thibault: pushing attributed ROI up every year quietly kills growth

When leadership demands marketing-attributed ROI climb every single year, you eventually starve the spend that was feeding everything else.

Combined with above, know your target, measure well, and then stick to them.

Read the full post on LinkedIn →

8. Taylor Holiday (CTC): the creative quality barbell

Taylor framed creative quality as a barbell. On one end, a $40k produced video. On the other, something filmed on a phone. The $40k video lost to the phone footage.

His point: for a $3 to $10M DTC brand, the middle is the worst place to live. Polished enough to be expensive, not raw enough to feel real, and it loses on both counts.

I really agree with this one.

Read the full post on LinkedIn →

9. Joanna Wallace: a bad editor wrecks a creative team faster than a bad strategist

Joanna’s contrarian one. The fastest way to destroy creative team morale isn’t a bad strategist. It’s a bad editor.

A weak strategist points the team at the wrong target. A bad editor kills good work, rewards safe work, and teaches everyone to stop taking swings. One is a wrong direction. The other rots the engine.

If evaluation is the real skill, then the person doing the evaluating is the highest-leverage seat on the team, and the easiest one to get wrong.

Read the full post on LinkedIn →

10. Preston Rutherford (Chubbies): marginal CAC, not blended CAC

Preston wrote a CEO and CMO dialogue that teaches the difference between marginal and blended CAC - super fun read and again a really clear way of describing the concept.

Read the full post on LinkedIn →

11. Preston Rutherford (Chubbies): Brand vs. Performance is the wrong fight

From the lesson of Chubbies’ nine-figure acquisition. The argument everyone has, Brand versus Performance, is mostly a distraction. The real driver sits underneath both.

As someone who is heavy performance focused I really like learning from Preston.

Read the full post on LinkedIn →

The team and the customer

12. Joanna Wallace: hire creative strategists with agency reps

Joanna makes the case for hiring creative strategists with an agency background over a brand background, every time. Her logic: agency reps are grad school. You see fifty brands, fifty problems, fifty failures, in the time a brand-side hire sees one.

I’m biased, I’ve done both. But the breadth point is real. Just make sure it's not an agency with 40 accounts on each person.

Read the full post on LinkedIn →

13. Eli Weiss: Loyalty is a feeling, not a points balance

Eli’s distinction: a loyalty program and a loyalty strategy are not the same thing. The most loyal customers share one trait. The program makes them feel something, not just earn something.

He may be biased but he has a great point.

Read the full post on LinkedIn →

Quick Summary

Read these back to back and a pattern shows up. Almost every one is about the gap between what a number says and what’s actually true.

Voice versus generic output.

Evaluation versus execution.

What you’d attempt versus what you’d save.

Real lift versus tool error.

Paid versus the credit it gets.

A budget working versus a budget that just exists.

The dashboard improving versus the business growing.

Marginal versus blended.

Brand-versus-performance versus what actually compounds.

A good editor versus a safe one.

Feeling versus points.

That’s the whole game right now. The tactics are mostly known. The edge is in seeing the real number under the convenient one.

Happy Friday!

Curtis

(p.s. please do follow all these people, all great educators)

A strong offer is almost always the reason.

I don’t like talking about offers since it’s a dangerous funnel.

But once you have a great product, great ads, strong media buying, and a website that converts with CRM locked and loaded?

Offer structure can be a huge multiplier on top of that performance when done right.

Sponsored by UpPromote

If you are running offers you need to be VERY CAREFUL who you pick as your affiliate partner.

The most common issue: a brand hands promo codes to anyone with a following, the codes leak to Honey and coupon Reddits, and existing customers find them at checkout. That’s the Bed Bath & Beyond problem, running on autopilot. You pay commission on revenue that was already walking through the door.

The deeper miss is targeting. Every ad dollar you spend has logic behind it: demographics, interests, lookalikes, intent. Then you recruit affiliates by DMing whoever looks big enough. That’s untargeted traffic, and then everyone acts surprised the program is flat.

A real program fixes both. Per-affiliate trackable codes, fraud detection, automated payouts, and partners actually matched to your niche. Run it right and affiliates can drive 30-40% of total revenue. Run it wrong and you pay 5-10% on sales you’d have made anyway.

@UpPromote is the Shopify-native platform I point brands to. 250,000+ stores, $1B in tracked affiliate GMV.

They also built a free tool that does the targeting work for you. Paste your store URL, and it identifies your niche and recommends the 5 partner types most likely to convert for your category, plus outreach templates, in under a minute. The same logic you apply to ads, pointed at partner recruitment.

If you’re running offers and want to make sure you’re paying commission on new revenue, not existing customers, the tool is free and takes under a minute: https://upmote.link/5142ab

Why offers move numbers like nothing else

The mechanism is simple: an offer change touches every impression in the account at the same time. A creative test touches one ad.

When a creative hits, one ad takes off. It pulls spend, carries the account for a while, and eventually fatigues. When an offer hits and is added to your landing page, everything lifts at once.

We’ve seen this drop CPA by 78% at higher spend within a week in some cases.

The second reason offers hit: your evergreen ads fill the funnel, and the offer closes it. At any moment there’s a pool of people who have seen your ads, mostly agree with you, and just haven’t had a reason to act today. The promo collects that demand.

The math is roughly: 99% of viewers don't click, and 95% of clicks don’t convert. Adjust for frequency and impressions that scroll fast (no engagement) and you still get 99% of awareness never buying.

And it compounds on Meta specifically. More conversions in a short window means more signal for the algorithm, faster cash conversion, and better optimization data. The promo doesn’t just convert the funnel. It makes the whole account smarter for a week.

One caveat that applies to every offer win: cheap conversions that don’t hold up downstream aren’t a win. If it’s lead gen, track your close rate. If it’s ecom, track your LTV on promos.

Subscribe now

How to pick the offer

The first thing I ask on a new account: tell me your offer in one sentence, the way a customer would repeat it to a friend.

“$0 down, fast installation” is an offer. “Up to 25% off select styles with code SUMMER25” is forgettable and bland. I’m looking for the first: something specific, memorable, and compelling.

The second thing I pull is the Meta Ad Library for the top five competitors. Your offer competes with whatever sits next to it in the feed, not with whatever you ran last quarter.

From there, the routing comes down to one question: what’s actually stopping the purchase? Every construct exists to kill a specific objection. Name the objection out loud, in the customer’s words, and the offer mostly picks itself.

“It costs too much.” Price is the objection, so price might be the lever. Percent off is the workhorse, and it wins across the board but easts into margin.

Price is the easy one: price is the first thing people say and the last one you should test.

For example, you may be able to increase the value first before a discount. Giving the customer a gift they value at 25% of the order might cost you 5 points of margin. A 20% discount costs you the full 20, and the customer values it less if they think you raised the price to just discount it.

That asymmetry is why gift with purchase is my preference vs discounts (or do both), and the default for subscription brands: perceived value goes up, price expectations stay untouched. Theme the gift to the moment (free gifts for dad in June) and it does double duty as creative. Just make sure the gift is something people would buy on its own or makes the original product more usable. A throwaway gift reads as clutter, not value.

“It’s a big commitment.” High ticket, installation, a long contract. The objection isn’t the number, it’s starting, so terms beat discounts: $0 down, no interest, no payments, free demo.

“What if it doesn’t work for me?” Skepticism is the objection, and you can’t discount your way out of it. The construct is risk reversal: a money-back guarantee, a free trial, a sample size, painless returns. Make it specific: “90 days, no questions” beats “satisfaction guaranteed” every time. Model in a return-rate assumption here. It’s a nonzero cost, but as long as the product is good, it’s better than the CVR hit of not having it.

“I’ll get to it later.” Nothing is blocking the purchase. People mostly agree with you and just have no reason to act today. You don’t need a new offer. You need urgency on the one you have, and the urgency has to be real: tie it to a calendar moment, or to actual scarcity, a kicker that runs while stock lasts. One rule: don’t publicize an end date unless it’s inside the next couple of days. A deadline two weeks out is not a deadline in today’s world.

“I don’t need it.” Stop. No offer fixes this one. If people don’t want the product at full attention, a discount just buys cheaper indifference. Improve the product or find the right audience. Don’t try to offer-hack this. It will just result in bad reviews, low LTV, or returns.

Then decide: one-off or system? If you’ll run offers all year, build one evergreen high-value construct and re-skin it per moment instead of inventing a new offer every holiday. Nine-figure brands run essentially the same offer all year and re-wrap it for each moment: small tweaks or products but same core idea. Same economics, new urgency, and the unit math always works.

Whichever construct you land on, four checks before anything launches:

Stack before you deepen. The cheapest improvements remove friction instead of cutting price: a payment term, a free-shipping threshold, a guarantee, a bonus unit.

Deepening is the expensive move. Make it your last one.

Condition the offer to lift AOV or LTV. Free shipping earns its keep when it’s gated: on subscriptions, or on two-plus units.

A good offer doesn’t just close the sale, it improves the order. If it doesn’t raise AOV or start a longer relationship, it’s margin out the door.

Do the depth math before launch, not during. A 30% discount doesn’t cost you 30%. It comes out of contribution. Take a $100 order with $35 in landed costs: you keep $65. At 30% off you collect $70 and keep $35. The discount cut contribution nearly in half, so your allowable CAC drops with it. Set the in-window CAC target from that math and hand it to your media buyer on day one.

Aim it at the first order. A sitewide discount pays a lot of people who were buying anyway. First-order constructs, the welcome bundle, the starter kit, the new-customer gift, point every dollar of margin give at acquisition. You can’t gate perfectly on Meta, but you can design the offer so its value concentrates on order one.

When to run it

The big holidays all work: Memorial Day, 4th of July, Labor Day, Black Friday, Thanksgiving, Christmas, New Year’s.

Then there are the in-betweens. Spring sale. Summer sale. Most brands chain them: Mother’s Day into Memorial Day into a summer sale into the 4th of July and back into summer. Seasonal language fills the gaps when there’s no actual holiday.

Brand moments work too. If you started your company in February or one of your founders has a birthday away from a holiday, that’s a great time to add another promo.

One timing detail most brands get backwards: promos convert better in the few days after the holiday than before it.

A 4th of July sale on July 1st says you have time. The same sale on July 7th says this might be the last day. Every day past the holiday adds urgency, right up to the point where it’s obviously never ending.

So don’t kill the promo on the holiday. The strongest days are usually still ahead of you.

How to run it

The mechanics matter as much as the offer itself:

Cap promo ads at 20-30% of account spend (by ad not landing page). Never 100%. More on why below.

Two formats are my favorites. Promo statics, plus your top evergreen videos with a scrolling banner on top that says “25% off for the 4th.” Add in dedicated video promos if you can.

80% iterations, 20% new. Pull last year’s holiday winners and rerun them. They reactivate year over year more often than you’d expect. Promo creative requires way less creative diversity in my experience.

Accept the fatigue. Promo ads die 3 to 14 days after the holiday and the account goes back to baseline. That’s fine.

Load the funnel first. In the week before a sale, push DR ads to fill the top of the funnel. Run a bit less efficient than normal (and more efficient during the sale). That awareness has to come from somewhere.

Building the calendar

When, what, and how are useless as trivia. Here’s the assembly order I’d run with a brand starting from zero:

1. Anchor the majors. Pick the four to six proven holidays that fit your category. Health brands don’t skip New Year’s. Nobody should skip Black Friday unless you are REI.

2. Add your brand moments. Birthday month, product anniversary, the category’s natural season.

3. Fill the gaps with seasonal frames. Spring sale, summer sale, fall. Depends how often you want to have some kind of promo. Subscription-heavy brands should run promos more often since initial AOV is less important than LTV, so they should chain these in.

4. Assign one offer construct per slot. Most slots are re-skins of the evergreen offer. Vary the wrapper, and keep it fun.

5. Set windows, not dates. Promo ads start about two weeks before the holiday and run a week past it for most big brands

6. Pre-build both copy variants. An evergreen version and a short-dated version of every asset, written before launch.

7. Tag everything for next year. Name promo ads by holiday so next June you can pull this June’s winners in five minutes.

The concerns are real

Up top I called offers a dangerous funnel, and I meant it. I philosophically hate the idea of going promo heavy. And I still push for promos every holiday, on every brand, because the data keeps winning the argument.

The way to hold both is to take the objections seriously. There are four that matter.

The impression problem. Every dollar on Meta buys an impression, and an impression that says “25% OFF” is an impression that didn’t say why your product matters. Go 100% promo and you’re paying premium holiday CPMs to communicate a price instead of a selling point. That’s the real reason for the 20-30% cap and the banner-on-winner format. The proof points keep running. The offer just closes.

The Bed Bath & Beyond problem. Train customers to only buy with a coupon and you’ve permanently repriced your product. The fix is the evergreen offer. When your everyday offer is already high value, a holiday promo is a re-skin, not a markdown, and you never teach anyone to wait for the markdown.

The hangover. Promos pull demand forward. Budget for the dip after the sale or you’ll spend a week diagnosing a problem that doesn’t exist.

The opt-out tax. Holidays raise CPMs whether you participate or not. If everyone else is running Memorial Day offers and you aren’t, you pay holiday auction prices without holiday conversion rates. Worse, the shopper who just bought your competitor’s deal isn’t buying your full-price product this week. Sitting out isn’t neutral. People won’t remember that you sat out.

The practical takeaway

Six moves, in order:

1. Confirm the fundamentals first. Product, creative engine, media buying, a site that converts. The offer multiplies what’s working. It doesn’t fix what isn’t.

2. Look at your past wins. One ad winning is creative. Everything winning at once is the offer.

3. Name the objection in the customer’s words and let it route you: price, commitment, risk, or timing. Then draft five variants of the construct before your next creative sprint.

4. Cap dedicated promo ads at 20-30% of spend. Banners on proven winners, 80% iterations, rerun last year’s promos.

5. Build the calendar once: anchor the majors, add your brand moments, fill the gaps with seasonal frames, and set windows instead of dates.

6. Load the funnel the week before the sale. Budget for the hangover the week after.

Most operators will test 200 ways to say it before testing one change to what they’re offering.

This month, two words cut a brand’s CPA 78%. The framework above is how you find yours.

If this was useful, send it to someone who has a good offer (or doesn’t). Thanks!

Share

Prime Day 2026 runs June 23 to 26.

Amazon moved the whole event up to June this year after nearly a decade of July dates.

My first job inside marketing was actually Amazon-specific. So I have a lot of Prime Day experience.

My biggest learning? the brands that win Prime Day built their demand months ago, and you can’t change that now. The job this weekend isn’t to build a huge Prime Day. It’s to not mess it up and capture what you built in the past.

Start With What You’re NOT Doing

The fastest way to blow the next four days is trying something new you’ve never tested on huge spend levels.

No new creative concepts. You don’t have time to test, and running untested concepts on the highest-traffic days of the summer is a good way to end up with an unprofitable ROAS.

No optimization strategies you’ve never run. Prime Day is not the moment to try click campaigns, new bidding setups, or anything you haven’t already validated at normal spend.

Friday: Three Decisions (About 2 Hours)

Decision 1: Which channel gets the better deal?

If your site evergreen promo beats your Amazon deal, you split conversions and confuse buyers. If Amazon undercuts your site, your DTC offer looks weak all week.

Or, third, run a “prime time” sale on your site matching the Amazon offer.

Most brands let both channels run whatever was already planned and find out Tuesday that they’re competing with themselves.

Pick a winner on purpose. If your site unit economics are better, give a better site offer (just make it a GWP or offer bar so it's not scrapeable, as long as it's not identical SKUs, Amazon can’t match it back). If you are trying to build Amazon reviews, give a better Amazon offer.

Decision 2: Spend to your inventory, not your ambition.

It’s too late to inbound more FBA units. Count what’s actually sellable and pace your spend against that number.

If units are tight, spend steadily across four days instead of blowing the budget on day one and going dark.

Decision 3: Raise your budgets now.

Prime Day traffic will drain a normal daily budget by mid-morning. Then your ads sit dark during the highest-intent hours.

Raise Amazon campaign caps today. If you’re pushing Meta traffic to Amazon, decide that number today too. That way you have a plan and don’t panic-adjust.

The Weekend: Creative and Email (About 3 Hours)

Badge your winners.

Take your top 3 to 5 ads and put a Prime Day banner on them. That’s it. Your winners already work, they just need a deal flag.

Then knock out a handful of quick Prime Day statics. Product shot, deal callout, urgency line. Make the offer and the product clear.

Don’t edit your live winners, because editing resets learning on the ads currently working. These should be net new ads set up, then paused after Prime Day.

Segment your list

An email to everyone can send your best customers to Amazon, where you lose the subscription, the data, and the margin.

Example segmentation:

Active subscribers & recent buyers never see a Prime Day email. They already pay you regularly. Leave them alone.

Lapsed customers (no recent purchase, not subscribed) get sent to Amazon. They weren’t coming back to your site anyway. A Prime Day purchase from a dead segment is found money.

Then schedule everything now: a launch-day send, a mid-event nudge, a last-chance email Friday. Write the post-event “missed the deal” email this weekend too, scheduled for Saturday the 27th.

Fix your links.

One thing on every ad and email pointing to Amazon.

First, attribution links, so you collect Amazon’s 10% Brand Referral Bonus. It’s free margin so claim it.

Start the push Saturday, not Tuesday.

Amazon Attribution counts conversions for 14 days after the click. Traffic you send this weekend still gets credited when those people convert at deal prices on Tuesday.

Weekend clicks are also cheaper than event-day clicks, because every brand in your category piles into the auction Tuesday morning. So if you’re pushing Meta traffic to Amazon, start Saturday. Not at full budget levels, but a small “get ready” test is worthwhile.

Break it into its own campaign so you can measure it post-window and decide if it's worth increasing or cutting next year.

Monday: Keep the gas on.

The instinct says Prime Day steals from your site, so pull back DTC spend and let Amazon have the week. I recommend the opposite.

Buying intent rises everywhere during Prime Day. People are in shopping mode, and site conversion often lifts during the event even when you’re not pushing traffic there. Pausing your regular Meta campaigns means pulling out of the market during one of the biggest shopping windows of the summer, and resetting learning on the way out.

Leave your DTC campaigns running.

Then verify everything actually renders.

Monday is for checking. The coupon clips. The strikethrough price shows. Subscribe & Save settings are what you think they are. Redirects resolve to the right listings. Scheduled emails point at the right segments.

And check what the final stacked price actually is. A Prime Day coupon that stacks with your Subscribe & Save discount can put you at 40%+ off without anyone deciding that. Check for yourself.

And set your day-of cadence.

Pick three check-in windows for Tuesday: a morning budget check, a midday pacing look, an evening scale-or-kill call. Decide the schedule Monday so Tuesday is execution instead of refreshing dashboards all day.

The Whole Thing on One Page

Friday: pick which channel wins (or run a parallel offer on purpose), pace spend to inventory, raise budgets.

Weekend: badge your top ads, quick statics launched as new ads, segment the list (protect subscribers and recent buyers, push 90+ day lapsed to Amazon), schedule every send including post-event, attribution links and redirects on everything, and start the Meta-to-Amazon push Saturday to load the 14-day attribution window.

Monday: leave DTC running, verify everything renders including the final stacked price, set the day-of cadence.

Tuesday: 12:05 a.m. check, then execute.

The brands that win Prime Day built their demand in March and April. This is just the final checklist.

Did I miss anything? Hit reply and let me know!

Share

Here are the data highlights:

Brands like SHEIN (96.6%), Calvin Klein (93.4%), On Running (85.1%), Obvi (82.1%), and Fabletics (82.5%) are running heavily image-weighted accounts. These are not small brands hedging because they cannot afford video production. These are the largest Meta spenders in DTC.

This does not mean video is less important. Glossier runs 57% video. Honeylove runs 84% video. Both are successful. The point is that static is not a secondary format. If you haven’t cracked static ads or think they could perform better, that is why I’m writing this.

Quick note from Vinay. Curtis is about to hand you the system: 12 formats, 5 axes, one SKU turned into 20+ distinct ads. A healthy account runs 6 to 8 formats and ships 12 to 15 genuinely new statics a week.

Now look at your team and ask who’s actually producing that.

That’s the gap. The strategy above is right. The problem is a person cutting statics at 30 to 60 minutes each can’t hold 6 to 8 formats across 5 axes and still refresh weekly. So accounts quietly collapse back to 2 to 3 formats and one angle, and the algorithm reads it as sameness.

That’s what Notch fixes. Not a clip generator. An autonomous marketer that ships finished, Meta-ready ads across the formats Curtis just mapped. Paste a product URL → Product Hero, Social Proof, Before & After, UGC-Native, statics and video. Point it at a winner → it rebuilds the concept across new formats and personas, not 20 lazy iterations. Feed it your top ad → it branches genuinely distinct variations before fatigue.

Diversity, not just volume. The exact bar Curtis set.

And it pairs with a human, it doesn’t replace one. Your team directs and edits. Notch carries the production load they physically can’t.

This newsletter was brought to you by Notch, the autonomous marketer that ships winning ads. Built by ex-Meta folks behind Facebook Lead Ads and Instagram ads, raised $8M, now running creative for 4,000+ businesses. Readers get 30% off any annual plan → Sign up or book a custom onboarding call with founder Vinay.

Why the algorithm now rewards static diversity

Meta’s Andromeda update changed how creative gets evaluated and served. The algorithm now uses visual recognition to assess creative diversity across your account.

Real creative diversity now means genuine visual, format, and messaging variation. And static images are where you produce that diversity fastest and cheapest. PLUS those placements on Meta are cheaper on a CPM basis.

A static ad takes 30-60 minutes to produce. A video takes 4-8 hours. A small team can produce 15-25 static variations per week. The same team produces 3-5 videos.

AI changes this, but most people I talk to are still seeing better performance from strategically using AI to augment ad production, not to fully outsource it. End-to-end AI production creates volume. Proper AI usage should create efficient quality output.

Statics are cheaper to make, cheaper to place on Meta, and can test diverse angles quickly. Despite having less information density than videos, they still consistently perform in top DTC accounts.

The 12 static formats every DTC brand needs

Here is the system. Twelve distinct formats, mapped to the funnel stage where each does its best work. A healthy account has active ads in at least 6-8 of these at any given time.

01. Product Hero (BOF)

Clean product photography on a neutral or branded background. The product is the entire message.

Works because it eliminates cognitive load for warm audiences. They already know what you sell. The fastest path to conversion is removing everything between them and the product.

Best for fashion, accessories, beauty, home goods. Especially effective on retargeting.

What to test: solid brand color background vs. lifestyle context vs. pure white.

Dr. Squatch | High Impression Ad

02. Social Proof / Testimonial (MOF, BOF)

Customer quotes, star ratings, review counts, expert endorsements paired with product imagery.

Works because when a prospect sees thousands of five-star reviews or a quote from someone who sounds like them, it short-circuits evaluation. The brain defaults to “if that many people trust it, I probably can too.”

Best for supplements, skincare, health, subscriptions. Anything with a trust gap.

What to test: customer review vs. expert endorsement vs. press mention.

Ridge Wallet | High Impression Ad

03. Before & After (MOF)

Split-frame visual showing the problem state versus the result after using the product.

Works because it makes the transformation tangible. The “before” triggers problem recognition. The “after” triggers aspiration. The product becomes the bridge.

Best for skincare, haircare, supplements, fitness, cleaning products. Obvi has built their entire static strategy around this.

What to test: horizontal split vs. vertical split vs. slider-style overlay.

Obvi | High Impression Ad

04. Feature Callout (MOF)

Product image with annotated benefit callouts pointing to specific features.

Works because annotation feels informational, not promotional. The viewer processes it like content. The arrows and callout boxes create a natural reading path.

Best for tech accessories, supplements with ingredient stories, apparel with functional features. Huel excels at this with clean ingredient breakdowns.

What to test: 3 callouts vs. 5 vs. full ingredient breakdown.

Dr. Squatch | High Impression Ad

05. Us vs. Them / Comparison (TOF, MOF)

Side-by-side or chart-style comparison between your product and the competition, the old way, or the category default.

Works by anchoring your product against something the viewer already dislikes. They do not need to evaluate yours in isolation. They only need to agree that yours is better than what they are currently using.

Best for any product displacing an incumbent. Ridge built their brand on this format and has never stopped running it.

What to test: generic category (”old way”) vs. implied competitor vs. direct name.

Ridge Wallet | High Impression Ad

06. UGC-Native / Screenshot (TOF)

Designed to look like organic social content. Selfies, text message screenshots, comment replies, casual phone photos.

Works by bypassing the “ad filter.” After years of scrolling past polished brand creative, people have developed unconscious pattern recognition for ads. UGC-Native earns the half-second of extra attention that separates scroll from click.

Best for beauty, skincare, fashion, food. Glossier has mastered this.

What to test: selfie + text overlay vs. screenshot of texts vs. comment reply.

Glossier | High Impression Ad

07. Grid / Collage / Mood Board (TOF, MOF)

Multiple images arranged in a grid. Pinterest-board energy combining product, lifestyle, UGC, and texture.

Works because grids trigger completeness. The brain wants to scan every cell, which increases dwell time. Each cell can target a different objection or desire.

Best for fashion, jewelry, home goods, beauty with diverse product lines.

What to test: all product vs. product + lifestyle vs. product + UGC.

Glossier | High Impression Ad

08. Offer / Promo (BOF)

Bold typography, discount codes, urgency language, clear CTAs. The conversion closer.

Works through loss aversion. “30% off ends tonight” triggers a fear of missing the deal that is often stronger than the desire for the product itself.

Best for any ecommerce brand running promotions. Especially powerful for the “Four Peaks” strategy of four major sales events per year.

Obvious: Black Friday / New Year / Memorial Day

Less Obvious: Spring Sale / Brand Birthday / Founder Birthday / BIS Sale / Cultural Event Sale

What to test: countdown urgency vs. scarcity vs. exclusivity.

Ridge Wallet | High Impression Ad

09. Ugly / Lo-Fi / MS Paint (TOF)

Deliberately unpolished creative. Handwritten text, crude drawings, mismatched fonts, screenshot energy. The anti-ad.

Works because it violates expectations. In a feed full of polished brand creative, something that looks like it was made in MS Paint creates genuine pattern interruption. The brain registers it as “not an ad.”

Best for brands with a strong established identity who can afford to break it. Also excellent for early-stage brands where authenticity matters more than polish.

What to test: MS Paint text vs. handwritten on paper vs. phone notes screenshot.

Dirty Dog Farm | High Impression Ad

10. Data / Stat Callout (MOF)

One powerful number or data point as the hero of the image. The statistic IS the creative.

Works because specificity feels more trustworthy than qualitative claims. “11,000+ five-star reviews” or “2 million wallets sold” creates instant credibility for the analytical buyer.

Best for supplements, review-heavy brands, subscriptions, products with clinical backing.

What to test: social proof number vs. product spec vs. outcome number.

Obvi | High Impression Ad

11. Meme / Culture (TOF)

Trending meme formats, pop culture references, humor-forward creative with brand messaging layered in.

Works because the brain processes it as entertainment first, advertising second. This earns not just attention but engagement: likes, comments, shares. Those social signals feed back into the algorithm and earn cheaper distribution.

Best for brands with humor-forward identity and a social team that moves fast. Dr. Squatch uses this consistently as a TOF entry point.

What to test: trending template vs. brand-specific recurring joke vs. reaction image.

Dr. Squatch | High Impression Ad

12. Listicle / “Reasons Why” (MOF)

Text-heavy format listing 3-5 benefits, reasons, or ingredients. A mini landing page in a single image.

Works because lists are scannable. The numbered structure promises a specific amount of information and creates a completeness loop. Stacks multiple proof points into one impression.

Best for supplements, meal replacements, tech products, subscriptions. Huel uses this to stack nutritional benefits.

What to test: 3 reasons (cleaner) vs. 5 reasons (more comprehensive).

Huel | High Impression Ad

The 5-axis creative diversity framework

Format diversity is necessary but not sufficient. Genuine creative diversity on Meta lives across five axes. Some formats require some consistent overlap but this is my general framework:

Axis 1: Format. Which of the 12 you are running. Most brands default to 2-3. Top performers maintain 6-8 in rotation.

Axis 2: Visual style. Polished brand creative, lo-fi UGC, editorial photography. Running the same visual treatment across every format is still creative sameness.

Axis 3: Messaging angle. Pain point, aspiration, social proof, curiosity, humor, education. Each angle attracts a different segment.

Axis 4: Offer type. Free gift, percentage discount, dollar-off, BOGO, no offer. Dr. Squatch leads promos with free bags. Other brands never mention price. Test both.

Axis 5: Audience persona. Ridge runs the same wallet targeted at minimalists, travelers, tech workers, and gift-givers with completely different creative for each. Same product, different person in the ad, different context. One SKU becomes 20+ distinct ads.

The practical takeaway: if your account has format diversity but the same visual treatment, same angle, same persona across all of them, the algorithm still reads it as sameness.

Brand teardowns: five DTC accounts to study

Five DTC brands. How they build their static systems. What you can steal.

Dr. Squatch: The Volume Machine

Men’s personal care. Estimated $220-280M revenue. Acquired by Unilever in 2025.

Format mix: Heavy on Meme/Culture, Offer/Promo, and Ugly/Lo-Fi for TOF. Product Hero and Social Proof for conversion. Three-way split between humor-driven TOF, education-driven MOF, and urgency-driven BOF.

What to steal: Volume and iteration velocity. They test more creative than almost any brand in DTC and maintain a consistent brand voice that is humor-forward, masculine, and self-aware. They turn one celebrity partnership (Sydney Sweeney, Mike Tyson, Alan Ritchson) into dozens of format variations.

The opportunity if I were running it: They are extremely video-heavy. Scaling static production to match would open real headroom. Specifically more Data Callout formats built around their ingredient story, more Grid/Collage showing the expanding product line, and more UGC-Native since the audience skews young.

Ridge Wallet: The Premium Product Showcase

EDC accessories. ~$200M annual revenue, $300M+ cumulative wallet sales. Expanded from wallets into rings, backpacks, and luggage.

Format mix: Primarily Product Hero, Us vs. Them, and Feature Callout. Premium product photography on clean backgrounds. Side-by-side comparisons (slim Ridge vs. bulky traditional wallet). Material and color showcases. Offer/Promo around new product launches drives incremental ROAS above 4x.

What to steal: Making a simple product category feel premium and diverse. A wallet is a wallet, but Ridge sustains diversity through materials (aluminum, titanium, carbon fiber, Damascus steel), colors (20+ options), and use cases. Their comparison ads are best-in-class because no copy is needed to understand the value gap.

The opportunity if I were running it: They are very product-focused, which is their ceiling. More Social Proof formats built around the 2M+ wallets sold stat, UGC-Native showing real people using the wallet in context, and Listicle formats. The brand is strong enough to support Ugly/Lo-Fi as pattern interrupt.

Glossier: The Community-Driven Approach

Beauty and skincare. The brand that proved community-generated content could outperform traditional beauty advertising. 42.8% image in our dataset (one of the more video-leaning DTC brands).

Format mix: Centers on UGC-Native, Grid/Collage, and Social Proof. Unfiltered selfies, community mood boards, customer review statics that feel indistinguishable from organic Instagram posts. Product Hero exists but is secondary to people-first creative.

What to steal: Their visual identity is so consistent that even “unbranded” UGC immediately reads as Glossier. That is the mark of a brand system working correctly. They build campaigns around customer content, hitting social proof, diversity representation, and feed-native formatting all at once.

The opportunity if I were running it: They are heavy on TOF and MOF but lighter on BOF conversion creative. More direct Offer/Promo formats, Feature Callout ads explaining ingredient science (they have credibility to go deeper), and Comparison ads (natural approach vs. heavy makeup) would all earn a real test.

Obvi: The Bold Transformation Play

Collagen and weight management supplements. 82.1% image in our dataset. One of the most static-heavy supplement brands on Meta. One of the few openly using AI-generated imagery in static ads.

Format mix: Heavy on Before & After, Offer/Promo, and Data Callout. Bold, colorful, emotionally charged transformation ads. Social Proof statics with high-energy testimonials. Seasonal promos with strong urgency. Bright pinks, purples, bold typography.

What to steal: They have made Before & After their own by keeping it fun and aspirational rather than clinical or shame-based. The “before” is treated with empathy, not embarrassment. The “after” is framed as confidence, not just physical change. In a category where transformation messaging easily becomes problematic, this tonal choice is critical.

Huel: The Education-First Native

Nutritionally complete meal replacement. 67.8% image in our dataset. Highly systematized static creative.

Format mix: Feature Callout, Listicle, and UGC-Native are the core. Annotated ingredient breakdowns are some of the cleanest in DTC. A distinctive “reply to comment” UGC-Native format addresses common objections (”is it actually worth the money?”) in a native, authentic-feeling way.

What to steal: Education that feels native. The reply-to-comment format takes a real customer question and answers it in accessible, non-corporate language. Disarms skepticism by acknowledging the objection rather than pretending it does not exist. Feature Callout ads lead with outcomes (”complete nutrition in 2 minutes”) rather than specs.

The opportunity if I were running it: They play it safe visually. Everything is clean, branded, on-template. More Ugly/Lo-Fi, Meme/Culture (meal prep humor, “what I eat in a day” parodies), and bolder Before & After showing the convenience transformation (messy kitchen + 15 supplements vs. one Huel shake) would push them past their template ceiling.

The 5-minute account audit

Run these ten questions against your current ad account.

1. Are you running ads in at least 6 of the 12 static formats?

2. Do you have both polished and lo-fi visual styles active?

3. Are you testing at least 3 different messaging angles?

4. Do you have ads running with AND without a promotional offer?

5. Are you targeting at least 2 distinct audience personas with different creative?

6. Have you refreshed your oldest running static in the last 4 weeks?

7. Do you have at least one “ugly” or meme-format ad in rotation?

8. Is your static-to-video ratio at least 50/50 by ad count?

9. Are you launching 5+ genuinely new statics per week?

10. Can you name the top-performing format in your account right now?

Score yourself:

8-10 yes: Diversity is strong. Focus on iteration speed.

5-7 yes: Gaps exist. Use the format matrix to identify what is missing.

0-4 yes: Creative diversity is likely your biggest performance bottleneck.

The takeaway

Static images are 55.6% of all ads across the 106 brands we analyzed. Among DTC brands specifically, that number climbs to 64.8%.

There are some brands scaling successfully with 100% or 95% video. But if that’s you, it’s worth occasionally testing new statics to see if you can get one to hit.

The brands winning on Meta in 2026 are running both video and static 90%+ of the time. But the thing that lets them test fast, iterate daily, and maintain genuine creative diversity is a systematic static production process. Twelve formats. Five axes. Six to eight formats active at any time. Twelve to fifteen new statics per week.

When you find a winner, the job is not to crank out 20 more iterations of the same thing. It is to repurpose the winning messaging across genuinely new creative formats. A winning video hook becomes a static headline. A winning static becomes a carousel. A winning testimonial becomes a Data Callout.

Creative diversity beats creative volume. Every time.

Thanks again to Notch for sponsoring the newsletter. I met with them and was very impressed by their products. It's an autonomous marketer that ships winning ads. Built by ex-Meta folks behind Facebook Lead Ads and Instagram ads, raised $8M, now running creative for 4,000+ businesses. Readers get 30% off any annual plan → Sign up or book a custom onboarding call with founder Vinay.

Share

After years of running these accounts, my opinions on what works have hardened. Some tactics I used to defend I now think are a waste of time. Some I dismissed I now consider table stakes.

Im sure ill change my mind on some next year again!

So I built a tier list.

F Tier is “stop wasting time on this.”

S Tier is “this is what compounds.”

If you disagree with where something landed, email me!

F Tier: Stop Doing These

Engagement objectives. You are paying Meta to optimize for the cheapest metric on the platform. Likes don’t buy your product. Unless you are hyper-focused on utilizing these correctly (and using the low-quality ones a good way’s a good way to burn money.

Frequency capping. Every hour you spend tuning frequency caps is an hour you didn’t spend making a new ad. If your audience is fatiguing, the answer is more creative, not less delivery on the creative you have.

D Tier: Mostly a Waste

Lookalike audiences. Meta’s algorithm finds your buyers better than any seed list you’ll ever build. Lookalikes felt like magic in 2018. In 2026 they’re a worse version of adv+ or broad targeting.

Retargeting campaigns. Most retargeting is showing ads to people who were going to buy anyway. The incremental lift on a clean lift test is often very low. If you run these, don’t measure them on click basis (unless they are underperforming on clicks, which we often see, then definitely cut them).

C Tier: Fine, But Stop Obsessing

The CBO vs ABO debate. The honest answer: it depends on whether your media buyer is actually in the account. If they are, ABO gives you more control and lower CPAs. If they aren’t in the account every day, don’t run ABO. Either way, the structure isn’t what’s holding your account back. Its utilizing it well for the product and creative you have that is.

Advantage+ Creative enhancements. Letting Meta auto-crop your ads and overlay random text is often really bad. And many of them just don’t perform well.

Heavy offer-led creative. Promo ads work. They convert. The problem is they train your customers to wait for the next sale. If 70% of your creative is “20% off,” its going to hurt long term growth. I generally target 70 non promo / 30 promo.

B Tier: Solid, Use Them

Interest targeting. (Im going to get hate for this one) Still works on launches when you don’t have data yet. Broad and Advantage+ Audiences outperform it once you have scale. Find an audience of a few million and turn on Advantage+ Audiences, it works like a seed into broad and outperforms on most tests we run.

UGC ads. Table stakes now. Every brand runs UGC. The bar has moved from “do we have any” to “do we have creators who can actually sell on camera.” The format is B Tier because everyone has it. Your creator roster and selling points is what determines whether it actually works.

Static image ads. Fast to produce, easy to iterate, still convert at scale. AI image tools have made these even more efficient. We use designers who use AI to move faster. Most ecom should run 20-30% spend on statics.

Cost caps and bid caps. Great for protecting CAC. The tradeoff is they throttle delivery when you want to scale. I use them most when the performance is volatile.

Whitelisting and partnership ads. Useful format. The lift over standard UGC is real but overhyped, otherwise this would be A tier. If a creator already converts well as a non-whitelisted ad, expect a 10 to 30% lift, not a 2x though. Focus on creators with real names in their handles OR problem specific handles. Worse for generic non person handles.

Long-form video (60s+). 15-30 second ads usually win for DTC. But when a long-form ad rips, it really rips. They tend to be founder stories, documentary-style ads, or product demos where the length is the value. Worth testing!

A Tier: This Is Where Scale Happens

Advantage+ Shopping Campaigns (ASC). Almost every brand should run ASC. Especially low-SKU brands selling to a consistent customer base. Feed it creative, set your budget, and get out of the way. Track performance vs your other campaigns and allocate budget.

Creative diversification. More concepts equals more shots on goal. The brands scaling fastest ship 100+ new ads a month. Not 100 versions of the same concept. 100 different ideas. If your team is shipping 10 ads a month, your ceiling is built in.

Big creative swings. The ads that 10x your account are never the safe ones. Ugly, weird, unexpected. The ad you almost killed in review. Every account I’ve scaled past $1M/month had at least one ad that the brand team initially hated.

Landing page testing. Massively overlooked. Creative testing gets all the attention, but a landing page change can move CVR and doesn’t fatigue. The reason brands ignore it: it’s harder, slower, and requires dev resources. The reason it’s A Tier: the upside is enormous.

S Tier: The Two Things That Compound

After 8 years of running these accounts, only two things consistently compound across every brand I’ve worked with.

A creative testing framework. Not a vague “we test creative.” A system. Specific volume targets, a defined testing window, a kill rule, a scale rule, a rotation schedule. The brands with a real framework outpace the ones without one by 3-5x on creative output and scale much faster.

Founder story ads. Especially when your founder is also your ICP. Nothing I have ever tested outperforms a founder who can sell on camera. The trust signal is impossible to replicate with UGC. The conversion rates are 30-50% higher in my accounts, the ads last longer, and the lift across the whole account is real.

The catch: most founders can’t sell on camera. The ones who can have a permanent advantage.

What did I miss? What did I rank wrong? Reply to this email and tell me where I’m off. Every tier list exists to be argued with.

The theme I saw in brands spending $1m/month+?

Every account I look at is different. Different team structures, different creative pipelines, different attribution stacks, different stages of brand maturity. Across consulting and auditing close to half a billion in total ad spend reviewed, I want to share some themes.

But these aren’t a checklist. Checklists work for the binary, settled stuff. Tracking. Attribution setup. Pixel events. Things where the answer is “this is set up” or “it isn’t.”

They fall apart for the most important pieces.

First, almost every brand I’ve audited has an “affiliate program.” Almost none of them actually have one that works.

What they have is a discount code distribution channel.

The pattern I see over and over: brand hands promo codes to anyone who asks → codes leak to Honey, RetailMeNot, coupon Reddits → existing customers find them at checkout → brand pays commission on revenue that was already coming through.

That’s basically just a tax on your conversion rate.

A real program has per-affiliate trackable codes, fraud detection, multiple commission structures, a creator marketplace, and automated payouts.

Brands doing affiliate right pull 15-40% of total revenue from it. Brands doing it wrong pay 5-10% commission on revenue that was already going to come through the door.

But here’s what most brands miss: a properly tracked affiliate program doesn’t just generate revenue on its own. It becomes your best creative testing lab. You get real purchase data on which creators, which angles, and which content formats actually convert, before you put a single dollar of ad spend behind them.

Brought to you by UpPromote, the Shopify-native affiliate platform powering 150K+ DTC brands. $1B+ in tracked GMV. [Check out UpPromote →]

Now onto the core themes I see.

1. The Power Law of Ad Spend

People often ask “what structure should I use” and I’m happy to share what we do 90% of the time (ABO for testing & scaling / ASC for scaling, tweaking based on account specific tests).

But a better question is “when I test a structure, what tells me it is working?”

3 things

1. Good ads get through testing (breaking ads out later doesn’t produce winners)

2. Fatigued ads get cut (they don’t keep spending forever)

3. Best ads get scaled.

In every Meta account I audit, 5-10% of the ads do 60-80% of the productive work. That part is consistent across scaling accounts. The goal is getting those ads scaled up and then down once they fatigue.

The above chart is dummy data from our internal dash. An example of an account where good ads were overspent and wasted money (ads deep in the red). And other good ads were never scaled enough (deep green). Breaking out tests creates new winners in this case.

Two symptoms show up in the same account:

Roughly 30% of ads are statistically dead but still spending. Past the point where the math says they’ll hit target. Still running. Still spending. Adset/campaign is still live and forcing spend. No new winners for Meta to rotate to so they just keep chugging along.

Roughly 5% of ads are early winners getting starved. Low CPA, low spend, sitting there doing nothing while breakeven ads scale by inertia.

If you are testing, cutting, and scaling ads within your structure then keep it.

I’ve seen it work well in 100% CBO, 100% ASC, a mix of all three. My point is that outcome is more important than structure. I’ve seen bad accounts with every structure and good accounts with every structure.

It depends on the product, the optimization strategy, and the ads. You need all three to align here to get a strong account outcome.

The goal is to get more winners through learning and pull back on fatigued ads before they waste spend. Not to check the box on the “right structure”

2. The Wall Between Media Buying and Creative

One super common disconnect: Target around “producing more ads” but then creative ships ads. Media buying runs them. Nobody is closing the loop between the two.

When a creative wins, what happens next?

Three things should happen

1. Scaling (new tests within Meta + new budget)

2. Iterations (how can we build on this with new creators/formats/angles)

3. Learning (discussion in the team)

The result is a creative pipeline that produces compound learning.

If you don’t do this, you are just repeating the same process over and over. Your team needs to be improving and learning. That’s what keeps setting the bar higher, and puts you in front of your competitors.

It’s important to note that if you don’t optimize your strongest winners it’s hard to tell what to lean on. (Do you iterate on High ROAS or High spend ads??) Once you have an account where fatigued ads are cut and great ads are scaled, this process becomes much easier.

But the media buying team should be involved. There is nuance to which ads are driving top of funnel, which are scaling the fastest, which are seeing fatigue in the last 2 days. Those things all need to be communicated to the creative team.

The highest scaling accounts have a quick weekly cadence here. But they know when to build off a winner and when to wait for data.

3. Creative Diversity Across 5 Axes

Most accounts I audit have “creative volume” as their KPI. That’s great, but meaningless without Creative Performance and Creative Diversity.

Meta’s algorithm rewards genuine creative variation more than it rewards quantity, and most accounts solve volume before diversity.

5 axes I find are consistently important:

Format. Static, video, carousel, story, collection, catalog, dynamic product ads, UGC, lifestyle, founder POV, problem-solution, testimonial.

Visual style. Polished brand creative, lo-fi UGC, editorial photography, mixed media, meme format. Running the same visual treatment across every format is still a form of creative sameness.

Messaging angle. Pain point, aspiration, social proof, curiosity, humor, education. Each angle attracts a different segment of your potential audience.

Offer type. Free gift with purchase, percentage discount, dollar-off, BOGO, no offer. You should be testing across multiple offer constructions even if you have a primary one.

Audience persona. Same product, different person in the ad, different landing page. One SKU can become 20+ distinct ads if you build it around different personas.

Most high scaling accounts are concentrated on one of these dimensions and diverse on the rest. I rarely see brands that are truly diverse across ALL of them and creating excellent ads. But, if you are concentrated across multiple of these - work hard to start diversifying as much as possible, it reduces your fatigue and increases the volume you can scale.

Don’t force diversity, test diversity and lean into what works. Key difference, it’s fine to have some concentration just not too much.

4. Attribution as a Decision Problem, Not a Tech Problem

The standard attribution conversation goes like this: numbers don’t match between Meta and Triple Whale. Meta says 200 purchases, Triple Whale says 140, Shopify says 175. The team spends weeks trying to figure out which one is “right.” They want to report something clean and easy and set KPIs to know if they “hit” or “miss”.

The key thing to understand is that no single attribution model is perfect and can be relied on. Forcing one is bad for the business.

If you use MTA, it undercounts view-through, long-term attribution impact, and omni channel halo.

If you use an MMM, you can’t optimize quickly enough, and you risk a poorly calibrated model.

The best brands know their North Star metric: total customers acquired per dollar or aMER (acquisition MER). They measure channel performance based on total impact, not just clicks. Within each channel, they use click-based tracking tools to ensure the right ads are getting the right spend, either by setting up the Meta algorithm or optimizing on top of it.

They also have all their tracking set up correctly, and they understand the unit economics and LTV numbers. That means they know how to optimize and what numbers to look for.

Too often, brands focus on which single tool they should trust. The better approach is to understand the web of tools and where each one works and where it does not.

5. Okay, Fine, A Short Checklist

I know I just spent too many words telling you not to audit your account like a checklist.

There are a few things that are worth checking anyway.

• Do you have your Pixel set up correctly with deduped CAPI events?

• Is your Conversions API sending hashed email and phone, plus IP and fbp/fbc cookies?

• Are you testing partnership ads (whitelisted creator content)?

• Are you testing ad copy variations, or are you testing landing pages? (Both work. Pick one as your primary lever.)

• Have you evaluated Advantage+ Enhancements and made an explicit decision about which ones to enable?

If you can answer “yes, intentionally” to all five, you get a gold star. ⭐

If you can’t, fix those first.

The Practical Takeaway

Be careful about following the “right system” the “right number of ads” the “right attribution tool” and work to understand the gaps in each. This is the theme I’m seeing across the biggest accounts. It’s not perfection but reliable systems and strong thinking:

4 things

1. Your structure gets good ads scaling and bad ads paused

2. Your creative team compounds learnings

3. Your creative production is expanding ideas not narrowing in

4. Your attribution is nuanced not mono-focused.

Subscribe now

Credit to Hugo - CMO of sassy saints who inspired this idea - if you want the full podcast from him ill share it at the end!

In this model direct creative cost runs about 3-5% of spend. A comparable agency runs closer to 7-15%, in my experience. That’s a nice benefit but not even the main one.

The interesting part is the moat that it builds (again shout out to Hugo for illustrating this clearly)

The 2026 AI Problem

Anyone can spin up 50 static variants in a Claude window. Anyone can generate decent imagery, decent hooks, decent copy. The barrier to making a passable ad collapsed somewhere around mid-2025. “Decent” is the new minimum bar now.

AI ‘UGC’ is getting better and sometimes outperforms standard UGC. You might be scaling an AI ad right now!

But the moat on that is $0. Any brand with any budget can copy that ad in an hour for their own brand. Meta ad library now sorts by impressions, look-a-likes pop up immediately. That causes quicker ad fatigue.

It means that layer of your account is now a commodity. Your competitor’s intern with a $20 Claude subscription can ship your top performer today if it’s a generic AI winner.

That doesn’t mean it can’t perform (I’ve seen it work) but it isn’t a moat you can build on.

Standard Model:

Brand pays for creative. Creator pulled from a roster, briefed, video edited, delivered.

The creator gets paid and then might repeat the next month with a different brand.

So many brands I know are stuck in this loop of deliver → reset and they are trying to compound instead. The question today is how we get to a compounding process.

But you may say “no my agency is excellent, they take the top performers and pull them in and they get weekly scripts! They analyze data and improve each week”

Perfect: Keep them (really), but also, take that and put it on steroids. That is the creator group.

Take your top creators, pull them into a group (slack, telegram, whatsapp, whatever), and start coaching instead of scripting. The cost stays flat. The output gets weirder, more specific, more aligned with what actually scales in the account.

Cost Breakdown

Most creators charge a flat fee per video of roughly $200-300 which often comes out to around 3% of spend depending on scale.

Or once you get comfortable I recommend a model of 5% of spend.

The first is easier to get started, finance isn’t concerned. Creators get clear payments, etc.

The latter is better aligned. If a creator wants to make 20 ads in a day? Awesome, test them in a group and if they have 3 winners that scale, they get paid!

Either way it should include the following:

1. Lifetime usage of edited ads

2. B roll rights (No % of spend costs)

3. Partnership access for all ads from day 1

What Compounds (And Why It’s a Moat)

We evaluate a new creator (before they join the group). They make 1-2 ads, one of them is decent. We started them on the top product with our top USP. Best shot at success. Not bad, not stellar.

Then they join the group.

After a few weeks they hit their stride and start producing more winners. People recognize their ads in the comment as if they are an influencer (Which is great, many selling points, many angles, that is good for the brand) it builds familiarity and comfort.

After a few more weeks they start getting creative: What if we tried X or Y or a new product?

It’s distributed creative strategy.

That’s the moat.

An AI can give you 50 variants of an ad concept in 60 seconds. It can’t give you the angle a creator developed over three months of paying attention to your brand’s actual customers. The AI doesn’t know what it’s like to use a product for months. It can’t personally try out your newest launch. AI UGC can only replicate authenticity at best. And it can’t execute on an angle that you never would have suggested in the first place.

The Partnership Ad Tailwind

If you were at Meta Performance Marketing Summit this year or if you’ve talked to a Meta rep in 2026 you’ve heard them pushing Partnership ads.

It’s the format where the creator’s handle appears on the ad and you whitelist their handle to run it from their account. Reps have told me these ads are getting actively favored by Meta’s algorithm right now.

Partnership ads feel more organic in the feed, which keeps users on platform longer. Meta wins, the creator wins, the brand wins. Everybody is incentivized to push that format.

In my experience ~70% of the time partnership ads outperform standard ads. It’s just hard to manage when you are using new creators every week.

We request whitelisting from every creator in our groups as a condition of working together. That gives us a stable pool of partnership-eligible accounts at all times.

The Playbook / Set up

Sourcing. Most of my top creators came from Billo, were past customers, or actors on backstage. Work with as many creators as you can, wherever you can find them, find the ones that resonate with your brand and pull them in!

Group size. Scales with spend. A brand spending $50K/month doesn’t need more than 5 to 7 active creators. A brand spending $500K+ should run 20+ as you scale (start small). The constraint is cost and keeping the group updated with feedback and approving new ads quickly or giving feedback.

Comms channel. One channel per brand, managed by our team. Ideas drop weekly. Creators can see each other’s work as we share top performers. This is the powerful part. They all want to get a ping that their ad was a winner.

Review workflow. Slack works fine to start. We eventually built an internal tool we call Ad Review that we use in our groups. Creators upload assets, AI reviews them immediately for a first pass, the team reviews and gives feedback in one place, captions automatically added (music coming soon), approved ads uploaded quickly, payments get tracked for creators easily + analytics on common mistakes in ads. Removes the back-and-forth scroll problem that big Slack channels create. Not required to run a creator group, but worth knowing the bottleneck is real and there are ways to engineer past it.

Pay model. Mix of flat and performance. Flat rate per asset to make the economics predictable for the creator. Performance bonuses tied to ads that scale past a threshold. Or % of spend once everyone is comfortable. Best creators end up making meaningfully more than they would on standard UGC rates, which is the loop you want. That keeps them loyal and you can scale quicker and save money testing bad ads. Win win.

Where This Doesn’t Work

A creator group is not a fix for a brand whose UGC was never working. It amplifies what’s already there. If your account has historically performed on statics and your video UGC has never scaled, this won’t change that. Different problem, different solution.

Four places I would not set this up:

Static-heavy brands. If your hero creative is product-on-white and you can’t remember the last UGC video that scaled, you don’t have a creator group problem. You have a format problem.

Brands under $50K monthly spend. The management overhead doesn’t pencil. You’re better off running standard Billo tests until you’re large enough to support the group. Just keep going back to those top creators!

Brands with no internal owner. Someone has to run the Slack. If it’s not your team, it has to be ours. If neither, don’t start.

Brands that need volume over authenticity. Some categories optimize for sheer asset throughput. If you’re testing 200 ads a week and authenticity is not your differentiator, then you can skip this.

The Layer Underneath

Most of the conversation about AI and DTC right now is about workflow. How fast can you generate concepts. How quickly can you spin up variants. How efficient can you make the production layer.

I often talk about a target of 1 ad for every $3k in spend (or 1 concept for $10k in spend). You need to not over optimize for this metric though. Quality first, volume second.

If you are tempted to AI-ify everything to hit the goal then this is for you.

The brands that own their creator relationships are building something AI can’t take from them and Meta is actively boosting.

If you build this out let me know!

As promised, here is the podcast from Hugo, worth a watch!

If this was useful, send it to your team or share on LinkedIn with your favorite graphic and tag me! Best compliment you can give!

Share