Ooh thats a good question - it probably just deserves its own analysis, but the short of it is around 80% confidence interval. If no conversions, that's normally around 3x your average CPA, but if you start to get conversions then I might spend a bit more before cutting it.
Thank you for replying! :) I'm only asking this because I've seen very little consistency or least it seems, performance itself is nonlinear which increases the risk of culling winners that look like losers but are in fact winners when given a bit more space and time. Especially since alpha of the power law appears to be quite low (at least from the chart you shared). Appreciate you replying.
Great article. How long do you run an ad for in terms of time and spend before culling it?
Ooh thats a good question - it probably just deserves its own analysis, but the short of it is around 80% confidence interval. If no conversions, that's normally around 3x your average CPA, but if you start to get conversions then I might spend a bit more before cutting it.
Thank you for replying! :) I'm only asking this because I've seen very little consistency or least it seems, performance itself is nonlinear which increases the risk of culling winners that look like losers but are in fact winners when given a bit more space and time. Especially since alpha of the power law appears to be quite low (at least from the chart you shared). Appreciate you replying.