GA4 says 40 conversions, Meta says 90, Google Ads says 70. None of them are lying. A plain explanation of why the numbers differ and how to actually use them.
GA4 says the campaign drove 40 conversions. The ad platform says 90. Finance says the bank account noticed neither. Every marketing team we work with hits this, and the first thing to understand is that nobody is lying. The platforms are answering different questions, and the trouble starts when you treat their answers as interchangeable.
Why the numbers differ
Ad platforms answer the question: did someone who saw or clicked our ad convert within our attribution window? They count view-through conversions, model what they cannot observe, and credit themselves generously because their window and their model are built around their own touchpoint. GA4 answers a different question: which source did this session arrive from, according to the attribution model you chose? It cannot see ad views, loses some journeys to consent and browser privacy, and shares credit across channels. Same customer, two honest answers to two different questions.
The mistake is picking a winner
Teams waste quarters arguing about which number is true. The productive move is assigning each number a job. Platform numbers are directional signals for optimisation inside that platform: they tell the algorithm and the buyer which ad, audience and bid is working relative to the others. GA4 is your cross-channel comparison layer: one consistent, conservative ruler across every source. Neither is your source of financial truth. That job belongs to revenue in your CRM or order system, tied back to marketing as cleanly as your measurement architecture allows.
A hierarchy that keeps everyone honest
We set this up the same way on nearly every engagement. Platform metrics optimise tactics daily. GA4 compares channels weekly. CRM revenue judges the strategy monthly. Each layer is allowed to disagree with the one below, and the disagreements are themselves informative: a channel that looks brilliant in its own reporting and invisible in revenue is usually harvesting demand rather than creating it.
When the gaps are worth investigating
Some divergence is structural and permanent. But widening gaps deserve attention: consent banners quietly eating your GA4 data, duplicate conversion tags flattering a platform, an attribution window changed during an account review. In audits we regularly find that the disagreement everyone had learned to live with was actually a broken tag from a launch nobody remembered. For decisions above the tactical level, incrementality beats attribution entirely; our guide to MMM versus attribution covers when each earns its keep.
If your channel meetings are really arguments between reporting systems, ask us for a second opinion. Making the numbers agree is impossible; making them useful is not.
