A sale in the CRM proves that the CRM contains a sale record. Establishing what reached an ad platform requires more evidence: the existing trigger, the delivery attempt, the platform's processing result and the reporting view.
Use this guide to review an existing connection. It is not an integration setup tutorial, and it does not assume a missing reported conversion means a sale was lost.
In this guide
- Define the outcome and the route
- Collect evidence at each handoff
- Describe the gap without inventing its cause
- Check campaign use after receipt
- Keep any test bounded
- Prepare a record level handoff without sharing customer details
- Avoid using a replay as the first diagnostic step
- Turn the evidence into the next decision
Define the outcome and the route
Start with one business milestone. “Qualified lead,” “signed agreement” and “payment received” can happen at different times. Record which milestone the integration is supposed to send and what status change or action starts the handoff.
Google Ads distinguishes qualified and converted lead goals, and supports importing offline outcomes. A CRM status alone does not configure that import. The event and connection have to be set up for the intended milestone. Google's offline conversion import guidance
Draw only the systems actually involved: the business record, existing connector or import process, selected destination account and intended conversion action. An extra box in a diagram is not evidence that a working integration exists.
Collect evidence at each handoff
Use a permitted, redacted record or a separately agreed safe test. Record timestamps and non-sensitive reference IDs where appropriate. Keep the original customer details inside approved systems; the worksheet only needs the evidence necessary to follow the path.
| Stage | Evidence to inspect | What remains separate |
|---|---|---|
| Business milestone | The agreed status or outcome and its timestamp | Whether the integration was triggered |
| Trigger | The rule and available execution history for this milestone | Whether a delivery attempt followed |
| Delivery | The existing job or request record and response | Whether the individual record was accepted and processed |
| Platform processing | Available import results, errors and diagnostics for the intended action | Whether the outcome appears in the relevant reporting period |
| Reporting and campaign use | The report, conversion action, date basis and campaign goal settings | Whether advertising caused an incremental sale |
Google Ads' offline data diagnostics provide upload history and errors. Use them alongside the connection's own records. A reassuring overall status is not a substitute for the evidence needed to answer a specific record-level question. About offline data diagnostics
Describe the gap without inventing its cause
Consider this illustrative finding: “The CRM contains the agreed customer milestone. We could not find a corresponding result in the platform records available for review.” This is an example of report wording, not a client finding or a completed platform test.
The next step depends on the last verified stage. If no trigger execution can be found, inspect the trigger rule and history. If a delivery attempt has a documented error, investigate that error. If processing is confirmed but the reporting view differs, check the selected action, date basis and report settings.
Missing access to a connector's logs should be recorded as a limit. It does not establish that the connector failed. Keep three separate notes: what was observed, what might explain it and what still needs verification.
Check campaign use after receipt
Recording an outcome and using it for bidding are separate checks. Review the selected campaign goal and conversion action. Google Ads' primary/secondary settings matter, and custom goals can use secondary actions for bidding. Google's primary and secondary action guidance
Do not change a campaign goal merely because an imported event is visible. Confirm the intended business outcome and a suitable change plan. Attribution credit also does not, by itself, prove the advertising caused the sale.
Keep any test bounded
Before creating a test record, agree the route, synthetic values, timing and possible side effects. A form or CRM update can trigger customer emails, SMS, staff alerts or sales tasks. Stop if those effects cannot be controlled. Do not create a live charge to make a reporting test look complete.
Existing authorized records may be enough for the first review. If a new integration is needed, that is implementation work with its own requirements and acceptance checks.
Prepare a record level handoff without sharing customer details
A useful review packet identifies the milestone, its timestamp, the selected destination action and the authorized evidence locations. Keep a safe reference to the business record inside the approved system. Do not copy an entire customer profile into a worksheet just because the integration contains one.
Distinguish the original event time, the delivery-attempt time and the time you inspected the report. They describe different stages. A later delivery may be intentional, delayed or incomplete; the timestamps help determine which existing configuration and record to inspect. They do not prove a cause on their own.
| Item | Useful record |
|---|---|
| Milestone | The business definition and when it occurred |
| Expected destination | The selected account and conversion action |
| Available history | Authorized trigger, delivery and processing references |
| Current conclusion | The last verified stage and the next unverified stage |
For an agency handoff, identify who can inspect each system. A client may own the CRM while another supplier owns the connector. Missing permissions are a coordination issue until evidence establishes otherwise. Ask for the specific approved record rather than requesting broad administrator access by default.
Avoid using a replay as the first diagnostic step
Resending a record can create another delivery attempt and may have downstream effects. Before any approved replay, inspect the existing deduplication rules, record identifiers and notification behavior. Do not assume that an identifier prevents duplicates across every platform, conversion action or import route.
Where the existing evidence is sufficient, document it first. If a new test is necessary, agree the synthetic data and side effects with the system owner using the safe test worksheet. Keep testing distinct from a decision to backfill historic outcomes. A backfill has its own data, eligibility and duplication checks.
A useful final note might read: “The milestone and delivery attempt are verified. Processing for the selected destination action remains unverified. The integration owner will inspect the specific response and destination diagnostics.” This is illustrative wording, not a finding. It gives the next person a bounded task without presenting an incomplete trail as a confirmed platform fault.
Turn the evidence into the next decision
For unclear milestones, start with lead-stage definitions. For totals that disagree, use the Google Ads and GA4 comparison worksheet.
Our US$195 conversion tracking diagnostic follows one agreed existing journey, one selected ad platform/account, one GA4 property where relevant and up to three agreed events. We deliver written evidence and prioritized recommendations. Book a fit call to confirm whether the existing connection fits; new integrations and repairs are scoped separately.