A lead conversion should represent the successful action you agreed to measure. A button click, a validation error and a completed enquiry are not interchangeable. To find duplication, follow one controlled submission from the form to the event and then to the reporting action.
Write the expected behaviour first
For a basic enquiry form, the test expectation might be: one accepted submission creates one lead record and one generate_lead event. A click that fails validation creates no successful-lead event. Refreshing a confirmation page should not create another lead record or event for the same submission.
This expectation is an implementation specification. It is not a claim that every business should use the same event name or counting model. Define qualified enquiries separately if a salesperson must review the lead first.
Run a small test matrix
| Test | Expected lead behaviour |
|---|---|
| Submit invalid required fields | No successful-lead event. |
| Complete one valid test submission | One accepted lead and its intended event. |
| Double-click Submit | The same submission is not recorded twice. |
| Refresh the confirmation screen | No second success event for the original submission. |
| Submit a genuinely new enquiry | A new record, handled according to the agreed counting policy. |
Inspect the event path
Use Tag Manager preview mode to see which tags and triggers run. Then use GA4 DebugView to inspect the received events for the test device. Record the timestamp, page and event name. A tag firing in preview does not by itself prove that a production conversion is counted correctly.
Look for two installations of the same tag, a hard-coded event plus a Tag Manager event, both a form-submit and a confirmation-page trigger, or a success handler attached more than once. These are hypotheses to test; remove only the duplicate path you have identified.
Check the conversion actions too
A single website action can be represented by more than one conversion action. Review whether a native Ads action and an Analytics import describe the same lead, and document which action is intended for bidding. Keep secondary diagnostic actions clearly named. Changing a counting setting can conceal duplicate collection without repairing its cause.
Do not assume purchase deduplication solves lead tracking. Google’s transaction ID guidance concerns duplicate conversions within the same Google Ads conversion action. It is not a universal instruction to send a purchase event for an enquiry.
Verify and annotate the repair
Repeat the test matrix after the change, publish through the site’s normal release process, then check the live installation. Mark test records so they are excluded from operational reporting. Add the repair date to the dashboard; historical duplicate events do not disappear just because future collection is fixed.
If the totals still differ, use the Google Ads versus GA4 reconciliation guide. For implementation help, see GA4 and conversion tracking setup.