This article was originally published here.

We’ve been working with a supplement brand in the US that launched in Q1 2020 for nearly 3 years now. Over that time they’ve relied heavily on paid Facebook ads for new customer acquisition. I was curious how their Facebook CACs & AOVs have trended over time, as this brand has been active through all of COVID & the iOS14.5 / ATT turbulence (as an aside, I was really trying to popularize the moniker IDFApocalypse but that didn’t stick, sadly).

My methodology:

  • I pulled Facebook spend, purchase conversions, and purchase conversion value by month using a normalized 1-day click attribution setting.
  • I included both prospecting and retargeting campaigns, but it’s worth noting that in this case retargeting accounts for less than 10% of total spend. This brand generally does not run remarketing to existing customers via paid channels due to the results of incrementality tests over the years.
  • I also pulled Google Analytics data so that I’d have a relatively constant dataset for the purposes of comparison (I was worried that the pre and post-iOS14.5 Facebook data might be unreliable/incomparable).
  • By default, Google Analytics uses a Last Non-Direct Click attribution setting (which is not quite last click, which is what I think a lot of people assume is the default). This GA data is from a Universal Analytics account, and I believe therefore that the Lookback Window is 6 months by default. I just spent 60 minutes+ trying to answer that lookback window question and I’m still not entirely sure that I’m right. It used to be the case that the lookback window was configurable for an entire GA view, but that is no longer the case (though it is in GA4!).
  • I pulled as many source / medium combinations as I could possibly find and summed them. I was also careful to grab odd source/medium combinations like and / referral to account for Link Shims etc.

The results:

As you can see in the table below, as the spend has increased the Facebook attributed 1-day click ROAS has dropped to around 1.0. You can also see a pretty clear drop off in attributed ROAS in Q2 of 2021 which coincides very neatly with the full rollout of iOS 14.5.

Facebook spend and Facebook attributed ROAS graph
Facebook spend and Facebook-attributed ROAS

Meanwhile, the Google Analytics data tells a more positive story. The ROAS appears much more stable over time, particularly after the iOS 14.5 rollout. Recent months are still materially lower than when the brand launched, but when you consider the spend increase the ROAS decrease strikes me as expected performance degradation.

Facebook spend and Google Analytics attributed ROAS
Facebook spend and Google Analytics-attributed ROAS

This table below shows the FB & GA ROAS side by side by month (again I would focus on Q2 2021 to see what I believe are the iOS 14.5 impacts):

Facebook-attributed ROAS and Google Analytics-attributed ROAS
Facebook-attributed ROAS and Google Analytics-attributed ROAS


1) I know there are many out there in Linkedin/Twitter world that will be horrified by 1.0 1-day click ROAS. Whether or not that's objectively "good", it's important to note that this brand's blended ROAS/MER is consistently 2.5+ and thanks to very high margins it's a very profitable business.

2) I don't mean to suggest that this is particularly revolutionary data. Obviously, FB has a much harder time with attribution, and many people in the ecosystem have made the point that blended ROAS/MER for many of these brands have not fallen off a cliff.

3) Without a doubt it's exponentially more difficult to scale a D2C brand (or any consumer product) on FB today than it was 3 years ago. I don't want to obfuscate that reality. That caveat aside, this is a really positive case study for FB. This brand has spent $5m over the last 3 years, is increasing investment over time, sees healthy ROAS, and is profitable as a business.