We work with a number of companies that look at performance in terms of last click Google Analytics attribution. We also have a handful of clients that have "internal" reporting, which usually means they are capturing and storing the Google Analytics UTM parameters' values in their own database. Those two sets of numbers are usually almost 1-to-1 (my sense is that some people prefer "internal reporting" as it sounds more complex!).

Like every other online advertiser, we see massive differences between what Google Analytics records on a last click basis and what Facebook's pixel reports, even when using a 1-day click attribution window. There are a myriad of reasons for those discrepancies, one of which is Facebook's use of link shims (this is why you might see l.facebook.com in your Google Analytics source/medium reporting).

I was wondering if there might be consistent discrepancies so I pulled Facebook Purchase data (March 1-23) using a 1-day click attribution window for 8 brands and compared that data to Transaction reporting from Google Analytics for the same period (all of these clients are using Google's default 30 day click lookback window). I also totaled Transactions in Google Analytic's reporting from other Facebook referral sources (ex: l.facebook.com, m.facebook.com) for the same period. I put the Facebook numbers over the GA numbers and here is what I found:

To help understand the data above, let's take Brand 4 for example. In that case, FB reports 137% more conversions (using a 1-day click attribution window) than Google Analytics captured using the corresponding utm_source and utm_medium values. When you factor in other Facebook.com referral sources, Facebook still reported 126% more conversions than Google Analytics.

In 7 out of 8 cases, the Facebook 1-day click attribution purchase count is significantly greater than the Google Analytics attributed Transaction count. It seems like the average discrepancy between Facebook and Google Analytics is probably around 200%, but we'd need more data points to say that definitively.

You'll also note that two highly considered products (Brands 8 & 6) have MASSIVE discrepancies. FB reports 561% and 800% more conversions on a 1-day click basis, respectively!). Especially in that case the GA data is effectively useless.

I hope this post is somewhat useful to you or at least thought provoking. Personally, my major takeaways here are:

  • FB vs GA ¬†discrepancies differ dramatically more category to category so it's unwise to apply any baseline or benchmark... in a pinch I'll assume GA * 200% = FB 1-day click.
  • Relying on last click attribution via GA or "internal reporting" is a mistake... it's clearly not reflective of the actual campaign dynamics and these discrepancies are just going to get worse over time (thanks to iOS14 etc).
  • I'll really think that for the purposes of optimization you need to be able to set goals in terms of what the networks' can actually report. That said, relying entirely on any networks' reporting is a mistake. I remain a big proponent of incrementality/hold out testing, despite its impending demise on the FB platform.