Synthetic Industry

Troubleshooting guide · updated 2026-10-11

A GA4 key event shows zero: is it not sent, mis-named or not marked?

Separate the three faults that make a GA4 key event count stay at zero, with a DebugView check for each and a note on what reports can and cannot repair.

Three faults that look the same in a report

A key event is an ordinary GA4 event that someone has marked as important. Google says any event the property collects can be marked, which means a zero in the key events column can come from three different places, and each needs a different repair.

The first is that the action never produces an event: a trigger stopped matching after a theme or form change, the tag was paused, or the page now loads the code differently. The second is that an event is produced but under a different name from the one that was marked. Treat a starred lead and a sent Lead as different events until you have compared the two names character by character. The third is that the event is sent correctly but was never marked, or was marked recently. Google states that marking applies from the time it is marked and does not change historic data, so a recent star will not repopulate old days.

  • Name length matters: Google documents a 40-character limit on event names, and a key event whose name is longer is not reported as a key event.
  • A standard property can hold up to 30 key events, so check the count before adding one.

Investigate in an order that tells them apart

Start with the live behaviour, not the report. Switch on debug mode for your own browser with Tag Assistant or Tag Manager preview, perform the action once, and open DebugView in the property. Google describes DebugView as showing events for the selected device as they are collected, in a seconds stream for the last minute and a minutes stream for the last half hour.

If nothing appears, the fault is upstream of GA4: look at the tag and the trigger in preview. If an event appears under a different name, compare it character by character with the starred name. If the right name appears once, open the property events list and check whether it is marked. Only after those three checks is it worth looking at report filters.

  • Perform the action once, then twice, so you can tell zero from doubling.
  • Copy the event name from DebugView into your notes rather than retyping it.
  • Remember that the debug parameter, once set in a configuration, is removed rather than set to false.

Why a fixed event can still look empty for a day

DebugView and the Realtime report move within minutes. Google says standard reports can take up to 24 hours to show a key event, and its data freshness page says many reports and explorations can take 24 to 48 hours. A repair checked only in the standard report on the same afternoon can look unsuccessful when it worked.

Two other effects are worth knowing. Google notes that debug traffic can skew reports and advises filtering developer traffic. And where consent mode is active, DebugView will not show events from a visitor who has not consented to analytics cookies, so run the test on the accept path.

  • Check DebugView and Realtime first; check the standard report the next day.
  • Keep a note of the time of each test so it can be found later.

What is not covered, and how the paid outcome is accepted

This guide does not repair past counts, add several new events or fix advertising conversions. If more than a couple of events are wrong, the fault is wider than one star and a measurement rebuild fits better. If it is one named action, the fixed outcome Fix one GA4 key event is quoted at £195 as an untested proposal, with payment only after you sign off.

Acceptance is a test you can watch. The agreed action performed once produces exactly one event with the agreed name and parameters in DebugView; performed twice it produces two; the event is marked as a key event; and, once Google has processed the data, an exploration filtered to a test marker on the test visit lists the key event once. If a developer-traffic data filter is Active on the property, the test visit is made without debug mode. Send the event name and the date the count changed. Do not send logins, customer data or form contents.

  • If the action can only be performed with a real payment or a real customer, the fixed test cannot be run.

Sources and limits