Different totals can be expected or defective
Google Ads and GA4 can report different conversion totals because their collection and reporting questions differ. That does not make every discrepancy harmless. Compare a declared event and cohort, then isolate expected reporting differences from reproducible implementation errors.
Google's Analytics conversion import guidance explains the two measurement routes and differences such as reporting date. Avoid installing another tag merely to make two dashboard totals look closer.
Create one reconciliation row per business event
| Field | Question |
|---|---|
| Business action | Are both systems describing an accepted purchase, or are different actions being compared? |
| Reporting date | Are you comparing interaction-date reporting with event-date reporting? Are time zones aligned? |
| Attribution scope | Which channels and interactions can receive credit in each selected report? |
| Counting and windows | Do counting rules, conversion windows and the selected action match the comparison? |
| Collection | Which native tags, imports, consent conditions and browser/server paths are active? |
| Business records | Which accepted orders, cancellations or adjustments belong in the cohort? |
Separate date differences from missing events
In a synthetic example, a customer clicks an ad on Monday and completes an order on Wednesday. Comparing a Monday interaction-based conversion report with Wednesday event activity can produce different daily totals even when the purchase was collected correctly. An equivalent weekly window may still differ because the attribution scope and settings differ.
Now suppose an accepted order produces no purchase event under the permitted test conditions. That is a collection issue worth investigating, not an attribution explanation. Conversely, one order generating repeated purchase records suggests a duplicate or trigger problem.
Google documents transaction-ID deduplication for web purchases. Check the actual order identity and test behaviour. Do not use a fixed or empty transaction ID across different orders.
Control the change to measurement
- Record current tags, imports, event definitions and goals before editing.
- Reproduce a normal completion and an important failure with permitted test data.
- Identify one discrepancy whose cause can be demonstrated.
- Propose the smallest correction, its approval owner and a recoverable version.
- Verify the changed event and inspect whether bidding or historical comparisons are affected.
Changing a primary conversion can alter campaign optimisation inputs. Review that consequence with the account owner rather than treating measurement as an isolated script change. A test event should not inflate commercial client or order reporting.
Define what a successful investigation delivers
The output should explain expected differences, corrected defects, unresolved uncertainty and the next observation window. It should not promise that every platform will attribute an identical number or that missing historical events can always be recovered.
Commission a tracking reconciliation milestone for this scope. Continue with Google Ads management when the commercial question concerns acquisition decisions after measurement is usable. Our browser/server event guide addresses the related transport-identity problem.