When we audit prospective new client accounts, I always look for tell-tale signs that they might have an incrementality problem. I look for the following characteristics:
If an ad account fits all of the above criteria, EVERY SINGLE TIME we've run an incrementality test the results have been devastating (in my mind, devastating means the incremental CAC is 5-10x the pixel reported CAC).
I wanted to better quantify my last criteria about view-through conversions accounting for a "large chunk" of attributed conversions, so I pulled purchase conversion data for prospecting campaigns from the month of February for 10 US, D2C brands using 1-day view, 1-day click, 7-day click, and 28-day click attribution windows (note: I looked at February so that I'd have the benefit of using a fully baked 28-day post click attribution window). This is what the purchase counts look like side-by-side:
These days, most brands are using the new default attribution window of 7-day click + 1-day view. With that in mind, I calculated default Purchases by adding the 1-day view + 7-day click numbers. Finally, to establish my benchmark I calculated the percentage of attributed purchases using the default attribution window that are view-through conversions. This is what that data looks like:
So for example, Brand 8's prospecting campaigns drove 764 conversions when measured using the default 7-day click + 1-day view attribution window. Of those 764 conversions, ~75% were 1-day view conversions. This is a really great example as we had the opportunity to run an incrementality test on those same prospecting campaigns and found that the pixel reported CAC was ~$150 during the test but the incremental CAC was $1,000+!
My prescription: if your ad account fits the criteria I described above and view-through conversions are at 40%+ of default conversions you ought to run an incrementality test ASAP. In those instances, I'd bet heavily that your real, incremental CAC from prospecting is at least 5x+ what the pixel is reporting.
PS: I don't mean to suggest that view-through conversions are implicitly bad or not accretive. In many instances, we've seen view-through conversions be very much validated by incrementality testing.
Another attempt to look at platform-wide dynamics post iOS14.5 through the lens of Thesis clients' data.
A feeble attempt to try to understand how Facebook's performance as a user acquisition channel has changed since the iOS14.5 rollout.