GA4 Conversion Tracking Issues: How to Catch the Silent Failures

Conversion tracking breaks quietly. There's no error message, no warning, no red flag in GA4. Campaigns keep running, reports keep populating — but the conversions being counted either never happened or stopped happening three days ago. The dashboard looks operational. The data inside it is wrong.

This is the specific danger of conversion tracking issues. They don't fail loudly. They drift, and downstream of the drift, every decision being made — bid strategies, budget allocations, campaign launches, A/B test conclusions — is being made on a slightly wrong picture of reality. By the time anyone notices, the cost of being wrong has accumulated for two weeks.

This guide covers the patterns that actually break GA4 conversion tracking, how to figure out which one you're in, and what fixing it the right way looks like.

Why conversion tracking specifically is high-stakes

A session tracking failure affects your analytics data. A conversion tracking failure affects the systems that act on your analytics data.

When conversion data is wrong, Smart Bidding strategies optimize toward the wrong signal. Target CPA and Target ROAS get trained on incomplete or duplicated conversion events, and the bidding algorithm responds to that signal in ways that take days or weeks to unwind even after the underlying issue is fixed. Attribution reports misallocate credit between channels — which means the team that "performed" looks worse than the team that "didn't" because the latter's conversions silently kept being recorded while the former's didn't. ROI calculations get baked into next quarter's planning on data that wasn't true.

The longer a conversion tracking issue runs, the more downstream systems learn from it. Two weeks of broken conversion tracking isn't just two weeks of bad reporting. It's two weeks of decisions made against the bad reporting, and the cleanup window is longer than the original gap.

That asymmetry — short failure window, long downstream consequences — is the reason conversion tracking deserves more monitoring attention than session tracking does, even though session failures look more dramatic.

The patterns that actually break it

Conversion tracking fails in a small number of recognizable ways, and recognizing the pattern is most of the diagnostic work.

The most common pattern is an event that was working stops firing. A site update changed a template, a checkout flow was redesigned, a form platform was migrated, a thank-you page URL got renamed. The GTM trigger that fired the conversion event was tied to the old structure, and the new structure no longer matches the trigger condition. The signature is a conversion count that drops to zero or near-zero on a specific date, with sessions and other events looking normal.

The second pattern is the opposite: duplicate firing. A single purchase fires the purchase event two or three times because GTM has multiple triggers active for the same action, or because a thank-you page reloads, or because two GA4 tags ended up live on the same page. The signature is conversion counts that look unusually high while the average value per conversion drops — the same purchase counted multiple times, but its revenue counted once.

The third pattern is event-marked-but-not-key-event. The event is firing correctly in GA4's events report, but it was never toggled on as a Key Event in Admin → Events. It shows up as a regular event but doesn't surface in conversion metrics. The signature is event volume looking normal while conversion volume reads zero — which is the diagnostic clue that the event itself is fine, but the configuration around it isn't.

The fourth pattern is cross-domain attribution failure. Users cross from your marketing site to a checkout on a different domain (or subdomain) without the GA4 session ID being carried across. The purchase event fires correctly on the checkout domain, but it gets attributed to a new direct session rather than the originating campaign. The signature is Direct traffic that appears to convert at a disproportionately high rate, while paid and organic conversions look mysteriously low.

A few less common patterns exist — wrong event mappings, incorrect parameter passing, consent banners filtering out conversion events specifically — but the four above cover the large majority of what actually goes wrong in practice.

Figuring out which one you're in

The diagnostic question is straightforward but the answer requires looking at two things together, not separately.

Start with conversion trends and session trends on the same chart over the past 60 days. The relative shape is more diagnostic than either trend individually. If sessions are stable and conversions dropped sharply on a specific date, you're in pattern one (event stopped firing) — the date will correlate with a site change. If conversions look unusually high but average conversion value dropped, you're in pattern two (duplicate firing). If the events report shows your conversion event arriving normally but the conversions report shows zero, you're in pattern three (not marked as Key Event). If Direct traffic conversions look disproportionate to Direct sessions, you're in pattern four (cross-domain breakage).

Conversion tracking is the analytics signal most worth comparing against itself across time. The pattern of the drop tells you the cause faster than any individual data point.

Once you've identified the pattern, the verification step is GA4 DebugView. Enable debug mode, complete a test conversion in your own browser, and watch what shows up. For pattern one, no event fires — the trigger condition is broken. For pattern two, the event fires multiple times — duplicate triggers active. For pattern three, the event fires correctly — the issue is the Key Event toggle in Admin. For pattern four, the event fires but the session ID doesn't persist across the domain transition — cross-domain config is missing.

DebugView is the diagnostic tool that catches the most issues fastest. Most teams underuse it because it requires actively testing the conversion path rather than passively reading reports.

Fixing without making it worse

The fixes for each pattern are mechanically straightforward. The thing that goes wrong in fixing them is not testing the fix in DebugView before declaring victory.

For pattern one (event stopped firing), update the GTM trigger to match the new page or element conditions, publish the container, and verify in DebugView. The pitfall is fixing the trigger without testing it on the actual user flow — and discovering a week later that the trigger now fires but with the wrong parameters.

For pattern two (duplicate firing), identify and remove the duplicate trigger in GTM. If the duplicate comes from a thank-you page reload, add a sessionStorage check that prevents the event from firing again within the same session. Verify in DebugView that the event fires exactly once per conversion.

For pattern three (not Key Event), toggle the event on under Admin → Events → Manage Key Events. This one fix is the closest to instant in the entire category — no GTM publish required.

For pattern four (cross-domain), add all involved domains under GA4 → Admin → Data Streams → Configure Tag Settings → Configure Domains. Verify in DebugView by completing a cross-domain journey and confirming the session ID persists.

The harder honest truth: the data you missed during the gap doesn't come back. GA4 doesn't backfill conversions. The window where conversion tracking was broken is a permanent hole in your conversion history, and any downstream system that depended on that data — Smart Bidding, attribution analysis, ROAS dashboards — needs to be understood as having learned from incomplete information during that period.

Catching the next one earlier

The reason conversion tracking failures get expensive is the gap between the failure starting and the failure being noticed. Cut that gap from days to hours and the cost shrinks proportionally. The math is simple — and the operational answer is monitoring that watches conversion volume against its own expected pattern, not against a fixed percentage threshold that gets ignored after the third false weekend alert.

Conversion-event monitoring is the single highest-leverage alert most teams aren't running. Sessions get watched because sessions are visible; conversions get watched after the first time the team gets surprised in a client meeting. Skipping the surprise and setting up the monitoring first is the cheaper sequence.