Google Ads, GA4 and your lead center may show different numbers for the same campaign. That alone does not prove an error. First check whether they count the same action, over the same interval, with the same attribution rule.
Start with the conversion definition
Make a short list of events each system records. Clicking Send, a successful form response, a record received by the lead center and sales qualification are different stages. Calling them all leads prevents useful reconciliation.
| Stage | Required evidence |
|---|---|
| Submit click | Browser interaction |
| Form accepted | Successful service response |
| Lead received | Record persisted at destination |
| Conversion sent | Receiving platform response |
| Conversion attributed | Platform report, after processing |
Also compare action sources in Ads. The same stage may appear in a native tag and an Analytics import. Adding these actions without checking their purpose may double-count results. Record which actions your analysis includes.
Align period, time zone and attribution
Ad interaction and conversion may happen on different days. Someone clicks on Monday and converts on Friday. Check whether the selected column groups by interaction or conversion time. Google recommends conversion-time columns when that matches the system being compared. See the discrepancy investigation guide .
Check the account and property time zones, filters and currency. A sale near midnight can fall on different days. Comparing yesterday across dashboards without checking these settings can create a difference that disappears with aligned intervals.
Models, eligible channels and windows also change credit distribution. Document the GA4 attribution settings used in the comparison. Do not expect reports with different criteria to match exactly.
Test the journey to the lead destination
Use a test environment or a contact marked as a test and follow every stage. Do not trigger a sales conversion to test a button. Check:
- Whether the form completes without errors on mobile and desktop.
- Whether the receiving service confirms storage before the interface shows success.
- Whether the destination records the event once, even when the network requires retries.
- Whether the reconciliation ID stays the same throughout the journey, without personal data in reports.
- Whether consent and browser restrictions are respected in the observed scenario.
An HTTP response accepted by the advertising platform proves technical receipt of that request, not final attribution to a campaign. Record these two types of evidence separately.
Keep a record that explains the difference
Create a record with period, time zone, account, event, source, deduplication rule, collection date and processing status. Preserve each source’s number. Adding reports does not automatically produce a unique lead total.
If a tag or integration changed recently, record the date and analyse before and after separately. Adjust campaigns only when the outcome definition is clear enough to support the decision.