THE STORY

Our client (who will remain unnamed due to an NDA) noticed discrepancies in data reporting between Google Analytics and Criteo and wanted us to explore the matter further.

They sent us a report from the Criteo ad platform, which contained information regarding the performances of their ad sets over the last month, last three months, and last six months.

It was up to us to analyze the data, compare it with the Google Analytics data, and determine where the issue was.

Considering that most platforms use different types of data attribution models, we were aware that some discrepancy in the information was mandatory. The only thing left was to determine the reason and the degree by which it occurred.

THE CHALLENGE

First and foremost, we had to determine the type of data reporting utilized in our client’s Criteo setup. We discovered that any purchase made within 30 days of clicking a Criteo ad was reported, regardless of whether it was the final touchpoint.

As a result, we needed to choose the most effective Google Analytics report for the most accurate comparison.

The best way to proceed was with a modified version of the Assisted Conversions report aggregated with Last Click Conversions in Universal Analytics because this number best reflects Criteo’s attribution mechanism.

Once this was completed, we could move on to the actual data comparison and report building.

THE APPROACH

Since Criteo has many different metrics in its reports, we had to choose the ones we could actually compare to the UA-s metrics and dimensions.

In the preliminary raw data report, we compared the following metrics for the last month, last three months, and last six months:

– Total assisted + direct conversions

– Tolal direct conversions

– Sessions

– Conv. Rate assisted + direct

– Conv. Rate direct

– CPA assisted + direct

– CPA direct

– ROAS assisted + direct

– ROAS direct

– Campaign cost

– Revenue assisted + direct

Based on the comparison, we could conclude how significant the discrepancies were and where they occurred. We also decided what data is essential to deliver to our client to answer their questions regarding the differences in reporting.

We developed a universal reporting model to compare the impact of various advertising platforms on our ad performance. It was done by “equalizing” the different attribution models they use.

We created a report using Google Slides to present our analysis and conclusions to the client in a clear and visual way. The report contained only the most crucial information and actionable conclusions we gathered from the initial raw report.

THE RESULTS

To demonstrate the impact of our analysis, we’ll show some of the findings from our report for the date range of the last 90 days (at that time).

For the date range of the last 90 days, matching the attribution, Google Analytics reports 17.46% fewer conversions than Criteo (Criteo being any touchpoint in the buyer’s journey).

It highlights a 21.15% discrepancy in CPA, GA calculating it at $460.82, while Criteo reports $380.36.

ROAS discrepancies are also significant, having a 60.72% discrepancy in calculated ROAS. GA reports ROAS of 2.24, while Criteo reports 5.70.

Comparing clicks and sessions, 34013 clicks on Criteo Ads resulted in 23722 sessions reported by Google Analytics, resulting in an 18.35% better conversion rate in GA, calculated with respective metrics (sessions for GA, clicks for Criteo). 

Including the numbers for the date range of the last 30 and 180 days would be redundant as they present a similar pattern.

It’s important to note that there is a significant discrepancy of about 50% in ROAS when comparing Criteo reports to any touchpoint Google Analytics data. Criteo ads were still involved in purchases that achieved ROAS higher than 2 in the last six months when looking at UA data. 

In our client’s case, this indicates that Criteo is not a great “Bottom of the funnel” traffic source but a useful “Top of the funnel” source.

THE CONCLUSION

The client’s concerns were valid. If they didn’t have the proper analytics implementation, they would have missed out on the additional information to help them make further decisions regarding an ad platform they employ.

It’s important to remember the risks of depending on just one advertising platform to evaluate the performance of your ad set, as demonstrated in this case study. Instead, getting an accurate and more detailed view with analytics data is always better.

We are not saying that the platforms are “lying” to you when reporting data, but each of them simply has a different viewpoint and attribution model. Most of the time, advertising platforms will present themselves in the best possible light through their reports. 

That is why, when making data-driven decisions in your business, we highly recommend you use multiple advertising platforms alongside multiple analytics platforms to get the best possible conclusion.

It will give you a clearer understanding of your advertising efforts and allow you to construct a more objective interpretation of the reported data.

If you want to learn more about GA4 reports and the overall Google Analytics 4 setup steps, check out our GA4 E-book. It’s a great way to discover the true potential of Google’s new analytics platform.