Why 40% of Your GA4 Conversion Data Is Wrong — And How to Fix It

GA4 is powerful. It's also easy to misconfigure in ways that make your data completely unreliable. Most marketers trust their numbers because conversions are appearing in the dashboard — but appearing and being accurate are two different things. These are the five issues we find in the majority of accounts we audit, and exactly what to do about each one.

Issue 1: Double-Counting Conversions

If you have both GA4 native events and GTM-fired events tracking the same action — a purchase, a form submission, a lead — you're counting every conversion twice. Your reported ROAS looks strong. Your actual revenue doesn't match. And your bid strategies learn from signals that are twice as large as reality, causing them to overspend chasing phantom conversions.

To fix it: go to GA4 → Admin → Events and review which events are marked as conversions. Then open GTM and check every tag firing on the same triggers. If the same action is tracked in both places, remove the duplicate. Pick one source of truth — either GA4 native events or GTM — and don't mix them for the same conversion type.

Issue 2: Misleading Engagement Rate Data

GA4 replaced bounce rate with engagement rate. A session is "engaged" if it lasts longer than 10 seconds, includes 2+ page views, or has a conversion event. The 10-second threshold is the problem: someone who lands, scrolls briefly, and leaves in 12 seconds counts as engaged. This can make a poor-performing landing page look healthy.

If your engagement rate looks unusually high (above 65–70% for a typical e-commerce or lead-gen site), audit it. Check whether your site fires scroll events that register as interactions on load. A misleadingly high engagement rate can mask a real UX or copy problem that's quietly costing you conversions.

Issue 3: Cross-Domain Tracking Not Configured

If your checkout lives on a different domain — Shopify Payments, Stripe, PayPal, or any external payment processor — you're losing attribution on every purchase. When the user crosses the domain boundary, GA4 creates a new session and loses all source/medium data. Your purchases appear as "Direct" traffic. Your paid channels get zero credit for revenue they generated.

Fix this in GA4 → Admin → Data Streams → Configure tag settings → List unwanted referrals. Add your payment processor domains (e.g., checkout.stripe.com, paypal.com) to this list. If your checkout uses a subdomain, also configure cross-domain measurement to maintain the session across domains.

Issue 4: Consent Mode Not Implemented

Consent Mode v2 has been required in the EU since March 2024. Without it, users who decline cookie consent are untracked — but the more serious issue is what happens to Google Ads. When consent is declined, Google Ads relies on Consent Mode to run conversion modeling (estimating conversions from non-consenting users). Without Consent Mode, modeling doesn't work, and your bid strategies operate on a fraction of the actual signal.

Even outside Europe, US state privacy laws are tightening quickly. Implementing Consent Mode now protects your attribution accuracy before it becomes both a compliance issue and a performance problem.

Issue 5: Wrong Attribution Model

GA4 defaults to Data-Driven Attribution. It sounds right — and usually is — but DDA requires a minimum of around 400 conversions per month to populate correctly. Below that threshold, it silently reverts to Last Click without notifying you. Your reports still say "Data-Driven Attribution" in the settings, but the model isn't actually running.

Check your current model in GA4 → Advertising → Attribution settings. If DDA shows "Insufficient data," switch to Last Click explicitly and acknowledge it — that's more honest than believing you have machine learning attribution when you don't. Once you scale past the threshold, switch back to DDA.

How to Audit Your GA4 in 90 Minutes

The fastest diagnostic: open GTM Preview mode, complete a test conversion on your site, then compare the conversion count in GA4 vs Google Ads. A gap above 5% means there's a tracking issue worth investigating. Do this before any budget decisions — inaccurate data costs far more than the time to fix it.