← Blog

GA4 shipped a fix for messy source data. Here's what it doesn't reach.

Google shipped Source Group in GA4 this June: a new reporting dimension that consolidates the obvious variants of a handful of major platforms: facebook, fb, and m.facebook.com all rolling up under one label, the same for Instagram, TikTok, Pinterest, and Amazon, plus newer sources like ChatGPT and Perplexity. It’s retroactive, needs no tagging changes, and for the specific problem it targets, it works.

What’s worth noticing is what kind of fix it is. Source Group is a derived reporting layer; Google’s own description is explicit that it doesn’t replace the underlying UTM fields. The raw utm_source value you sent is still exactly what you sent, typos included; GA4 now also computes a cleaner label alongside it, for a curated list of platforms Google maintains. Add a source outside that list (an internal partner name, a niche platform, intsagram instead of instagram) and there’s no rule to catch it, because the list isn’t yours to edit.

That’s the actual boundary. Source Group solves “the ten biggest platforms look messy in my reports” without touching the underlying event data, your own naming conventions, or anything downstream of GA4 reading that same hit. A fix at the source, before the event is even recorded, covering whatever vocabulary is actually yours and not Google’s list of the top ten, is a different, earlier layer than a cleanup pass applied afterward to a handful of well-known names.

Both can be true for the same account at the same time: Source Group cleaning up the report view for the platforms it already knows, Inflight correcting the event itself, in real time, for whatever utm_source your team actually uses.