Attribution in GA4: Features and Recommendations

In previous articles of our GA4 series, we’ve focused on the major differences in the way GA4 vs Universal Analytics works. The missing piece to our GA4 puzzle, so to speak, is attribution within GA4, as we are now working with a whole new system of campaign performance attribution models. In this article, we’ll explore these new models, how they differ from what we’ve used before in Universal Analytics, and make recommendations on how to use these new capabilities.

In Universal Analytics, the campaign attribution model was focused on the last touch. Back when it was first designed, this was more or less the only option for proper campaign attribution, so it was not a disadvantage at first. What was somewhat disadvantageous though, was that the attribution system was designed so that each campaign click would open a brand-new site session - even if the previous session was still active. Though this function made it simpler to identify the last click, it also had the side effect of inflating session numbers.

Over time, however, that last-touch system became outdated. What originally was a simple system that returned consistent results became an obstacle that obscured full customer journey touchpoints. As businesses realized there was much more to their campaign performance than just the last click, they needed better solutions. Where GA4 comes in is completely overhauling this system. Now in GA4, if a session is active, new campaign sources to that session will be included in that session. This solves the issue of session inflation, but introduces new questions around which of those campaign touchpoints should receive the default conversion attribution. Where GA4 has landed is, in the main Traffic Acquisition report, the default will be the first campaign touch in that session, meaning the initial campaign touchpoint that began the final, converting session.

GA4 also offers a User Acquisition model, which works much like a first-touch model. This shows businesses where their most valuable customers are initially acquired, and helps enable some campaign lifetime value analysis, which is especially valuable for businesses with repeat customers or user bases.

With these new attribution models the numbers we see in our reports are quite different from Universal Analytics, and during this transition period to GA4, it makes year-over-year comparisons to Universal Analytics data difficult. Fortunately, GA4 provides tools for further exploration. Data-driven attribution modeling (DDA) is now available for all GA4 accounts, which was previously a 360-only beta program. Instead of statically determining conversion attribution based on order of clicks, data-driven attribution modeling uses a machine learning algorithm to analyze patterns in your converting traffic and identify influential touchpoints - even if it’s not the last touch. It also updates itself dynamically over time, enabling it to incorporate the latest trends in business marketing activities.

This reveals contributions to success that a last touch model might obscure, and adds another tool in the box for an expert analyst to use to optimize campaign performance. Importantly, data-driven attribution only works with converted traffic (users who have already completed a purchase or submitted a lead, etc.), which offers insights into successful channels, but does not work well for channels or campaigns that generate traffic but not as many conversions.  

Because of that limitation, MERGE does not recommend using the DDA model for regular reporting or for campaign budget analysis. The Traffic Acquisition report is still the recommended method for regular reporting and campaign performance evaluation, but MERGE also highly recommends regular deeper-dive analyses using the DDA model. If we’ve learned anything from the long-overdue overhaul of the last-touch models, it’s that ignorance is not bliss!

Seeking a partner for all things data-related? Connect with the experts at MERGE, as we offer services for deeper-dive analysis, as well as actionable dashboards that ensure you know exactly how well your digital marketing campaigns are performing.