Google Analytics 4 measurement problems: prove collection first, then read the reports

Three layers, three kinds of fault

A GA4 number can be wrong because the site did not send the event, because the property is configured to treat it in a way you did not intend, or because the report you are reading is behind or filtered. Mixing these up wastes days. The reliable order is to prove the event leaves the browser first, then check configuration, then read the report.

DebugView is the proof step. Google describes it as showing events for a debug device as they are collected, with a seconds stream for the last minute and a minutes stream for the last half hour. Reports move more slowly: Google says many reports and explorations can take 24 to 48 hours.

• Collection: is the event sent once with the right name and parameters?
• Configuration: is it marked, grouped and attributed as you intend?
• Reporting: is the report current, and does it include test traffic?

Limits worth knowing before you plan

A standard property can hold up to 30 key events, marking applies from the time of marking and not to history, and event names are limited to 40 characters, with longer key event names not reported as key events. Custom channel groups, cross-domain settings and the Search Console link each have their own documented limits, covered in the linked guides.

None of these is a reason to avoid GA4, but they decide what a repair can and cannot do. A repair can restore correct collection from now; it cannot create events that were never sent.

• Plan the short list of key events before marking anything.
• Do not promise anyone restored history.

Which fault, which route

The linked pages are ordered by symptom. A key event at zero: the key event guide. Purchases above orders: the duplicate-purchase guide. Campaigns split across spellings or filed under Unassigned: the naming and channel-group guides. Checkout shown as a referral: the cross-domain guide. Search data missing from GA4: the Search Console guide. For a diagnosis order across all of them, use the measurement diagnosis map.

Each guide ends with the bounded paid outcome that fits and the test that accepts it. All prices are untested proposals, and payment follows sign-off.

• Start with DebugView on one test action.
• Take the smallest outcome that matches the symptom.

Priced routes and boundaries

Bounded routes: Fix one GA4 key event, £195; Fix duplicate GA4 purchases, £295; Fix campaign tagging and channels, £245; Fix cross-domain hand-off, £345; Shop funnel events from £795. A standing scheduled test with a monthly record is £195 a month, on a cadence agreed in writing. None includes changes to campaigns, legal advice on consent, or any promise about revenue, rankings or ad performance. Send the symptom and the date it began, never logins or customer data.