Most marketers assume their GA4 data is accurate until they have an obvious reason not to. A metric looks odd. A number doesn't match another platform. A campaign result seems too good or too bad to believe. Only then do they start digging — and usually find problems that have been compounding quietly for months.
The ten signs below cover the most common and most impactful data quality failures we find across property audits. Some are visible in your reports if you know where to look. Others require checking Admin settings you may have never opened. None of them will tell you there's a problem unless you specifically look for it.
Your conversion rate seems lower than it should be
A persistently low conversion rate — especially one that doesn't respond to changes you know should improve it — is one of the most common symptoms of data contamination. When staging traffic, development environments, or internal team browsing enters your production property, it adds sessions without adding conversions. The denominator grows. The conversion rate shrinks. Every percentage point looks worse than it actually is.
A property with 15% of sessions coming from a staging environment doesn't have a 2% conversion rate. It has a conversion rate that's been artificially suppressed by roughly 15%. If your real users convert at 2.4%, what your reports show is around 2.0% — and every optimisation decision you make is calibrated against the wrong baseline.
Where to check: Reports → Tech → Tech Overview → change primary dimension to Hostname. Any hostname that isn't your production domain is contaminating your data. How to remove it →
PayPal or Stripe appears in your top traffic sources
If you open your acquisition report and see paypal.com, checkout.stripe.com, or any payment processor listed as a significant source of sessions or revenue — your attribution is broken. What you're seeing is the symptom of a missing referral exclusion: when a customer leaves your site to pay and returns to your confirmation page, GA4 treats the return as a new session originating from the payment processor's domain.
The consequence is not just a messy acquisition report. It's that every purchase attributed to the payment processor is a purchase being stolen from the marketing channel that actually drove it. Your Google Ads ROAS is lower than it should be. Your email revenue is lower. Your organic conversion rate looks weaker. Every channel is underperforming on paper — because PayPal is getting the credit.
Where to check: Reports → Acquisition → Traffic Acquisition → change primary dimension to Session source / medium. Look for any payment processor domain in the top sources list. How to fix referral exclusions →
You can't pull reports older than 60 days
If you open Explorations and find that user-level data beyond two months is unavailable, your property is set to GA4's default data retention of 2 months. This setting causes all event-level and user-level data to be permanently and silently deleted on a rolling 60-day window. Standard aggregate reports are unaffected — but Explorations, custom funnels, path analysis, and cohort reports all require the event-level data that's being deleted.
This is the only data quality issue on this list that cannot be corrected retroactively. Data already deleted is gone. Changing the setting today stops further loss but does not restore what's been purged. If your property has been running for more than two months at the default setting, some of your history is already unrecoverable.
Where to check: Admin → Data Settings → Data Retention. If it reads "2 months," change it to 14 months immediately. Full guide to data retention →
Your email channel is split across multiple rows
Open your acquisition report and filter by email-related mediums. If you see both email and Email in the Session medium column — or newsletter, e-mail, EMAIL, and other variants — your email performance is being fragmented across multiple rows, and some of it is almost certainly landing in Unassigned. GA4's channel attribution is case-sensitive. Different medium values that represent the same channel are treated as entirely separate sources of traffic.
The same problem affects paid, social, and any other channel where UTM conventions haven't been enforced consistently. Each fragmented channel appears smaller than it is. Benchmarks and trend analysis are wrong. The total performance of any channel is impossible to see in a single number without manually summing its variants — which most reports will never do.
Where to check: Reports → Acquisition → Traffic Acquisition → sort by Session medium alphabetically. Scan for duplicate channel names with different casing. How to fix UTM inconsistency →
Unassigned traffic is more than 8% of sessions
Some Unassigned traffic is normal — GA4 can't always determine a source for every session. But when Unassigned climbs above 8–10% of total sessions, it's a signal that something specific is misconfigured. Common culprits: non-standard UTM medium values that don't match GA4's channel grouping rules, sessions originating from the payment processor issue in Sign 2, missing UTMs on campaigns that should have them, or internal link UTM tagging that's resetting session sources mid-journey.
Unassigned traffic is not just an aesthetic problem in your reports. It's attribution credit that belongs to a real channel — email, paid, organic, or referral — that isn't being properly assigned. Every campaign evaluated against this backdrop is being measured against an incomplete picture of its own performance.
Where to check: Reports → Acquisition → Traffic Acquisition. Check the percentage of sessions attributed to the Unassigned channel group. Anything above 8% warrants investigation.
A conversion event hasn't fired in weeks
Go to Admin → Events and look at every event that has the "Mark as conversion" toggle enabled. For each one, check when it last fired. If a conversion event on an active site hasn't fired in 14 or more days, you have a zombie conversion — an event that's registered as a conversion in GA4 but is generating no data and no signal. It may have been broken by a site update, renamed without updating the GA4 configuration, or simply never implemented correctly in the first place.
Zombie conversions are quietly damaging in two ways. First, they make your conversion tracking appear more comprehensive than it is — you see a list of conversion events and assume they're all working. Second, if Google Ads is connected, Smart Bidding may be referencing a conversion action that is no longer receiving data, distorting the algorithm's understanding of what a valuable user looks like.
Where to check: Admin → Events. Filter for conversions. Check last-fired dates. Any active site with a conversion event showing no activity for 14+ days needs investigation.
GA4 revenue is significantly higher than your actual orders
Pull your GA4 revenue and conversion count for any recent 30-day period. Then compare it against the actual order count from your e-commerce platform, payment processor, or CRM for the same period. A small discrepancy — 3–5% — is normal and attributable to users with cookies disabled, ad blockers, or privacy browsers. A large discrepancy in either direction is a serious tracking problem.
If GA4 revenue is significantly higher than actual revenue, you most likely have duplicate event firing — a purchase event that's firing twice for every real transaction, typically because both a GTM tag and a hardcoded snippet are active simultaneously. If GA4 revenue is significantly lower, you have tracking gaps: the purchase event isn't firing for all successful transactions, payment processor attribution is stealing credit, or Consent Mode v2 is missing and EU conversions aren't being collected or modelled.
Where to check: Compare GA4 ecommerce conversions and revenue against your platform's order management system for the same date range. A difference of more than 10% in either direction needs a root cause investigation.
Your EU traffic drives almost no conversions
If your site has significant traffic from the EU but your conversion reports show EU visitors converting at dramatically lower rates than equivalent US or UK audiences — not slightly lower, but near-zero — Consent Mode v2 is almost certainly missing or misconfigured. Without Consent Mode v2, GA4 cannot collect data from users who decline cookie consent, and cannot model their behaviour. In markets with high consent decline rates like Germany and France, this can mean 40–50% of your EU user activity is invisible to GA4 entirely.
The result looks like a geographic performance anomaly. EU campaigns appear to have terrible ROAS. EU audiences look like poor converters. Budget flows away from EU markets — when the real problem is that the data infrastructure for measuring them doesn't exist. Full guide to implementing Consent Mode v2 →
Where to check: Admin → Data Display → Consent Settings. If no consent signals are shown, or signals are present for only a fraction of sessions, Consent Mode v2 is not implemented.
Your reports have an orange warning triangle
An orange triangle icon at the top of a GA4 report or exploration means one of two things: either the data has been sampled (GA4 is showing you a subset of events rather than all of them), or reporting thresholds have been applied (rows with fewer users than a privacy threshold have been removed). Either way, what you're looking at is an incomplete picture of your data.
Sampling typically occurs in Explorations with long date ranges or complex query structures. For standard reports, it's less common but still possible on high-traffic properties. Thresholding typically occurs when Google Signals is enabled and demographic breakdowns result in small-sample dimension combinations. The fix for sampling is to reduce the date range or query complexity. The fix for thresholding is to adjust the reporting identity in Admin → Reporting Identity, or to accept that certain granular demographic cuts will always be subject to privacy thresholds.
Where to check: Any active orange icon in the top right of a report or exploration. Hover over it to understand whether you're seeing sampling or thresholding, and how severely the data is affected.
Your numbers look completely normal
This is the most insidious sign of all. Most of the problems above produce numbers that look reasonable on the surface. A 1.8% conversion rate with staging traffic contaminating 15% of sessions looks like a plausible conversion rate. An acquisition report dominated by paypal.com / referral looks like referral traffic behaving normally. Zombie conversion events don't produce error messages — they just produce silence that looks like a campaign that isn't converting.
The properties with the worst data quality problems are often the ones where nobody has looked for them, because nothing in the reports is obviously wrong. The data quality issues described on this page don't usually produce anomalies you'd notice without specifically checking. They produce data that is consistent, coherent, and wrong in ways that compound quietly over months and years.
The only way to know whether your GA4 data is reliable is to systematically check the things that can go wrong. Not because something looks suspicious — but because the absence of obvious problems is not the same as the absence of problems.
How many of these apply to your property?
Based on audits across hundreds of GA4 properties, here's what the distribution looks like:
The checklist above gives you the starting points for manual investigation. For each sign you've recognised, the linked deep-dives explain the root cause and the specific fix in full. Or, if you want a comprehensive picture of every issue across all seven data quality categories in one go, GA4 Health Check runs 47 automated checks and returns a scored PDF report in under 60 seconds.
