GA4 and CRM lead counts differ when they count different events, records or reporting windows. Start by comparing GA4's lead event count to website enquiries accepted by your backend. Then compare those accepted enquiries to CRM records. This separates a tracking problem from a delivery problem and tells you which team should investigate.

Imagine opening Monday's report and seeing 80 leads in GA4 and 60 in the CRM. The 20-count gap doesn't tell you how many enquiries were lost. Some events may never have represented an accepted enquiry. Some real enquiries may be awaiting review or delivery. Some records may simply be outside the CRM report's filters.

What does each system count?

System Count to inspect Common mistake
GA4 Count of the event your team uses for accepted lead submissions Comparing event count to users or counting a button click as a lead
Website backend Distinct enquiry IDs durably accepted during the reporting window Counting retries as separate submissions
CRM delivery log Accepted enquiries with a confirmed target record Assuming a successful worker response proves the record is visible
Sales report Enquiry or deal records within the agreed cohort Comparing contacts or qualified opportunities to all website enquiries

Agree on a time zone, date field and exact form scope. A Friday submission can arrive in the CRM on Saturday. If GA4 groups by event date and the CRM groups by record creation date, that one enquiry can appear in different weeks. Report by acceptance time when auditing the journey of a submission cohort.

Worked example with 80 GA4 events

Reconciliation step Change Count
GA4 lead events Starting count 80
Remove repeated events for the same accepted enquiry −9 71
Remove clicks or attempts that the backend rejected −5 66
Add accepted enquiries with no observed GA4 event +4 70
Remove accepted records intentionally held for review −3 67
Remove accepted records without confirmed CRM delivery −4 63
Account for CRM records hidden by the dashboard filter −3 60

This is a hypothetical audit with mutually exclusive buckets. The example contains four enquiries requiring delivery investigation, not twenty. It contains a separate event issue and a dashboard filtering issue. The same raw gap could have entirely different causes for another business.

How to diagnose the gap without guessing

  1. Test one form end to end. Send an invalid submission and verify zero lead events. Send a valid one and verify backend acceptance, one intended analytics event and one CRM enquiry. Test the thank-you page reload and check whether the event fires again.
  2. Compare backend and CRM by stable enquiry ID. Look for accepted IDs that are pending, held, confirmed, failed or delivered with an unknown outcome. Use an external enquiry reference if your CRM supports it. Check CRM permissions and exports before concluding a record is absent.
  3. Check the analytics side separately. Make sure the same action isn't counted by both a tag manager event and an application event. Review whether consent, blockers or lost network responses can explain accepted enquiries without an observed event. Production reports can also differ from exports, so document which report and filters you used.
  4. Fix the stage that produced the difference. Change the trigger for duplicate events, repair field mapping for rejected CRM writes, or correct reporting filters. Keep a small before-and-after sample so you can tell whether the change addressed the problem.

A record-level audit requires an approved way to correlate events. If your analytics setup does not include a permitted correlation key, use controlled test submissions and compare production counts in aggregate. Don't join unrelated records by approximate timestamp and present the result as a confirmed match. Keep email addresses and phone numbers out of analytics event parameters.

What should the weekly reconciliation report show?

Record accepted enquiries, confirmed deliveries, review holds, overdue pending records and delivery failures. Keep a separate analytics event total and annotate changes to tags or consent settings. Show the oldest unresolved delivery as well as the count. Four old failures may need faster attention than ten deliveries in progress.

When delivery is uncertain, search the CRM for the existing reference before replaying the request. An API timeout can happen after the receiving system has stored the record. A blind replay can turn a missing confirmation into a duplicate enquiry.

Frequently asked questions

Should GA4 and CRM totals match exactly?

Expect the test flow to behave as designed. Production totals can differ because of measurement coverage, definitions and time windows. Agree on what differences are explainable and investigate exceptions using backend records.

Which count should sales use?

Use the CRM for follow-up and qualification. Use the backend to verify website acceptance and delivery completeness. Use GA4 to understand observed website behavior and acquisition patterns.

Does marking an event as a GA4 key event make it a qualified lead?

No. The key event designation represents an analytics action. Qualification is a later business decision that belongs in the CRM.