How these pages work
Every page follows the same order. It begins by asking whether the problem is genuine or a reading error, because many "broken" numbers turn out to be a partial day, a filter left on, a preview session mistaken for the live container, or two reports using different definitions. Next comes a diagnosis table: the symptom as it appears in Tag Manager (GTMTag management system.), GA4Google's event-based analytics platform. or a dashboard, the check that confirms or rules out a cause, the cause, and the fix. Each cause then gets its own short section with the fix, a way to verify it worked, and the habit that stops it coming back, followed by a routine that catches the whole family of faults early.
Pick the page from the symptom you can see, not the cause you suspect.
| What you are seeing | Start with | Then, if it is not that |
|---|---|---|
| A Tag Manager tag never fires, or fires only in preview | GTM tag not firing | Data layer variable undefined |
| A tag fires but a variable it sends is empty or undefined | Data layer variable undefined | GTM tag not firing |
| A property shows nothing at all, not even in Realtime | GA4 not showing any data | GTM tag not firing |
| Numbers fell sharply on the day a cookie banner went live | Data drop after a cookie banner | GA4 not showing any data |
| The Unassigned channel holds a large share of sessions | GA4 unassigned traffic | Direct traffic too high |
| Direct traffic is far above what the business expects | Direct traffic too high | Sudden traffic spike |
| Sessions jumped overnight with no campaign behind it | Sudden traffic spike | Common GA4 mistakes for duplicate tags |
| A Looker Studio chart and the GA4 interface disagree | Looker Studio not matching GA4 | BigQuery export not matching GA4 |
| Queries on the raw export disagree with the interface | BigQuery export not matching GA4 | Looker Studio not matching GA4 |
| A Looker Studio chart, page or report shows an error or no data at all | Looker Studio chart showing an error or no data | Looker Studio not matching GA4 if it shows numbers that disagree |
| A Looker Studio report shows the same figures day after day, or its latest days never appear | Looker Studio report not updating | Looker Studio not matching GA4 if closed days disagree rather than stand still |
| An audience stays empty, or is too small to use in Google Ads | GA4 audience not populating | GA4 not showing any data |
| A server-side container receives no requests or forwards nothing | Server-side GTM not receiving data | GTM tag not firing |
| The shop takes orders but GA4 shows no purchases or zero revenue | GA4 purchase event not tracking | Data layer variable undefined |
| GA4 records purchases but its revenue or order count differs from the order system | GA4 revenue not matching orders | GA4 purchase event not tracking if purchases are missing rather than different |
| Leads arrive by email or in the CRM but GA4 shows fewer form submissions, or none | GA4 not tracking form submissions | GTM tag not firing |
| Cookies or vendor requests appear before the visitor answers the banner | GTM tags firing before consent | Data drop after a cookie banner for the opposite problem |
| Conversions claimed by every channel add up to more than the orders taken | Channel conversions exceed sales | Retargeting taking all the credit if the issue is where credit lands |
| Retargeting and brand search claim most sales at returns nothing else matches | Retargeting and brand search take all the credit | Channel conversions exceed sales if the total is inflated too |
The pages beneath this introduction are listed automatically, so a new diagnosis appears here once it is published.
Broad GA4 data problems live in Common GA4 Mistakes
Some questions are wider than one symptom: "our numbers look wrong" with no single report to point at. Double counting from two installations, "(not set)" values, self-referrals from a payment provider, the "(other)" row and data thresholds are all covered, with their own diagnosis table, in common GA4 mistakes. The pages in this section never repeat that material; where one of those causes is part of a narrower symptom, the page links to the relevant section instead. When the doubt is about the container rather than the property, the preview and debugging tutorial is the tool every page here relies on, and the GA4 hub explains the event model the reports are built from.
Platform-versus-analytics gaps are reconciliation, not troubleshooting
A difference between an ad platform's conversions and the analytics property is usually not a fault. The two systems count different units, credit different touchpoints, apply different windows and see different consent states, so they are expected to disagree. That question belongs to the guide on why platforms and analytics disagree, which gives a reconciliation method rather than a fix. The pages here cover disagreements inside one property: the interface against a dashboard built on it, or against its own raw export. Campaign tagging problems sit in between; when UTMURL tags identifying traffic source. values are the cause, the pages diagnose the symptom and hand the convention itself to the naming guide. For a standing routine that catches all of these before anyone searches for them, the marketing data quality guide sets out the weekly checks.
