GA4 attribution is one of the most common sources of confusion in our client meetings. The data is there — but interpreting it requires knowing what assumptions GA4 automatically builds into its model.
The first problem: default channel grouping. In many cases GA4 classifies traffic in ways that don't match how the business actually operates. For example, branded search can get mixed with unbranded, or partner platform traffic can appear as "direct". All of this distorts channel comparisons.
The second problem: data-driven attribution. A good model in principle, but it requires sufficient data volume to work correctly. For smaller stores with low conversion volume it can produce unpredictable results. This article shows which custom models work better in specific situations and how to build a Looker Studio view that reflects the true picture.
Full article coming soon.