Auditing a GA4 Property: The Data Quality Issues We Find Most Often

Every GA4 property we audit is confident. The reports render, the charts move, the numbers have decimal places. Confidence is the problem: a property can be materially wrong for years without ever looking broken, because analytics fails silently, and every decision built on it inherits the fault. Most audits we run find at least one issue distorting a number somebody relies on.

The faults are remarkably consistent across businesses. Here are the ones we find most often, roughly in the order we check for them, and what each one does to your decisions.

Duplicate and Missing Events

The first check is always the same: an event-count-by-name view, looking for pairs. Duplicate `purchase` or `generate_lead` events (classically from a tag firing on both a button click and the thank-you page, or from GTM and hardcoded gtag running side by side) inflate conversion counts and quietly corrupt every downstream consumer, including Smart Bidding if the events feed Google Ads. The reverse fault hides in single-page-app sections and payment redirects, where key events never fire at all. DebugView plus a full test transaction finds both in an hour, and that hour is the best-value hour in analytics. A naming mess makes this check harder, which is one more argument for the taxonomy discipline we covered separately.

Self-Referrals and Payment Gateway Attribution Theft

Open the traffic acquisition report and look for your own domain, your subdomains, and payment providers as referral sources. Every conversion credited to a payment gateway referral is a conversion stolen from the channel that actually earned it, because the user bounced out to pay and returned as a "referral". The fix takes minutes: list unwanted referrals for gateways, and configure cross-domain measurement for your own domains. Accounts running paid media with this fault are systematically under-crediting their campaigns and over-crediting nothing.

Unfiltered Internal and Developer Traffic

Your own staff, your agency and your staging environment are polluting production data unless someone explicitly excluded them. The symptoms are subtle: implausibly engaged direct traffic, conversion tests appearing as conversions, dev-site URLs in page reports. GA4's data filters handle internal IPs and developer traffic, with a testing mode so you can preview before activating; the audit step is checking the filters exist and are set to Active, because a filter left in Testing for two years is a fault we find often enough to check for it specifically.

Attribution and Configuration Defaults Nobody Chose

A cluster of settings silently shape your numbers, and most properties run whatever was set on day one. Data retention defaults to two months, which cripples year-on-year exploration analysis unless someone extended it (and is another reason to switch on the BigQuery export regardless). Session timeout, enhanced measurement toggles and the Google Ads link each carry defaults that may not match your business. Key events are the big one: audit which events are marked as key events, because we regularly find newsletter signups counted alongside purchases, inflating "conversions" in every report and, where linked, feeding Google Ads bidding a corrupted value signal.

Consent and Tag Coverage Gaps

Since consent banners became universal, the most common new fault is structural undercounting: the tag fires only for consenting users, no consent mode signalling is in place, and nobody adjusted expectations, so channel numbers drift below platform numbers and both get distrusted. Auditing this means checking what actually fires under each consent state, which we covered in depth in Consent Mode v2. The related coverage fault is simple page absence: new templates, landing page builders and checkout upgrades that shipped without the tag. A crawl comparison against your tag coverage catches it.

Running Your Own Audit

A serviceable self-audit is a repeating checklist run quarterly: event counts by name (duplicates, disappearances, rogue names), referral sources (self, gateways), a full conversion walk-through in DebugView, filter status, key event list review, retention and link settings, and a reconciliation of GA4 conversions against your CRM or order system with an agreed tolerance. Document the property's intended state in your tracking specification so the audit is comparison against a standard rather than archaeology.

The pattern across everything above: none of these faults announce themselves, and all of them are cheap to fix once found. If you would rather have the full inspection done professionally, with the fixes implemented rather than listed, a GA4 audit is a standard component of our analytics engagements and GA4 configuration work. Get in touch.