1. Is the event sent at all?
Open Tag Assistant or Tag Manager preview, perform one action and watch DebugView. If nothing arrives, the fault is before GA4: the trigger, the tag or the page. If an event arrives, note its exact name and parameters. Do not open a report yet.
- Zero events: look at the tag and trigger in preview.
- Wrong name: compare it with the one marked as a key event.
- Key event at zero: the key event guide.
2. Is the container something you can trust?
If nobody can say what the tags are, you cannot say which one sent the event. Inventory the container, preview the key actions, and change only in a workspace with a version to return to. A clean container turns later steps from guesses into checks.
- Unexplained tags: the container audit guide and the clean-up outcome.
- No written event list: the Search Console and measurement plan guide.
3. Does the data layer carry the right values?
For shops and forms, check what the page pushes. Stale items point to a missing clear-first push. Form counts that include failures point to the trigger choice. The synthetic test-event matrix shows what to compare.
- Ecommerce items wrong: the ecommerce clearing guide.
- Form counts wrong: the form trigger guide.
4. Is each action counted once?
Perform an order or a submission once, then reload. Two events for one action means two senders or a trigger that fires twice; GA4 counts repeats of the same ID once on web streams, so an over-count in reports needs different IDs for one order, while blank or reused IDs make GA4 merge different orders and undercount.
- Purchases above orders: the duplicate purchase guide and £295 repair.
5. Do consent and hand-offs lose the visitor?
If consent defaults arrive late, or a hand-off to another domain starts a new session, otherwise correct events can still vanish or be misattributed. Test on the accept path when using DebugView.
- Consent order: the consent mode guide (technical, not legal advice).
- Checkout shown as referral: the cross-domain guide.
6. Only then read the report
Google says many reports can take 24 to 48 hours. Compare the report row with your test time once that delay has passed. A repair that passed in DebugView and is not in the report yet is not a failure. Choose the smallest priced outcome that matches the first broken step; every price is an untested proposal and payment follows sign-off. Send the symptom and date, not logins or customer data.
Sources and limits
- Google Analytics Help: DebugView Checked 2026-10-11.
- DebugView shows events from a debug device as they are collected.
- Google Analytics Help: Data freshness Checked 2026-10-11.
- Many reports and explorations can take 24 to 48 hours to process data.
- Tag Manager Help: Preview and debug containers Checked 2026-10-11.
- Preview shows which tags fired and what triggered them.