Check Google Ads conversion tracking: find missing and duplicate signals.

Before testing, write one sentence that the business and technical teams understand in the same way. For an inquiry form: “An inquiry counts when the system has successfully accepted it.” A click on the submit button alone does not establish that outcome.

Distinguish an attempt, a successful receipt and a subsequently qualified lead. These states can answer different questions. Assign each state a clear event and a responsible data point. A validation error with no successful submission, for example, should not be reported as a successful receipt.

A website event, a GA4 key event and a Google Ads conversion serve different purposes. An event can be marked as important in Analytics; using it in Google Ads requires a corresponding conversion. Identify the actual source behind each action.

Also record which goals each campaign uses. Primary actions can be used for bidding when their standard goal is selected. Secondary actions usually serve observation, with an exception for custom goals. Check the actual goal configuration rather than relying only on the “primary” label.

Use a test environment or an explicitly agreed test action. Record the time, starting state and expected outcome. Keep settings unchanged during a run so that the resulting observations remain attributable.

Use Tag Assistant or GTM preview to observe triggering. DebugView can help investigate events received in Analytics; debug mode must be enabled. Account for the applicable consent and privacy settings. Missing debug data alone does not prove a fault.

This diagnostic table separates observations from possible causes. It suggests the next test. A suspected explanation becomes a confirmed fault only when the observation supports it.

GA4 can count a key event once per event or once per session. A change applies to future data. It does not repair event logic that fires repeatedly by mistake.

Set the business expectation first. Two genuinely submitted inquiries are different from one inquiry producing two technical signals. Inspect the events themselves, then the reporting count. A lower number after a settings change is not sufficient evidence of correct measurement.

For purchases, a unique transaction ID can prevent Google Ads from counting the same purchase twice within the same conversion action. Different purchases need different IDs; the same order retains its identifier. One fixed ID for all orders can cause undercounting.

The ID does not generally merge different conversion actions. The “One” count setting serves a different purpose: limiting counting after an ad interaction. Establish whether you are examining repeated signals for the same action or multiple distinct actions.

A useful test log connects the business action with technical observations. Record a cause as confirmed only when the test supports it. After a change, repeat the same case and compare the expectation with what you observe.

A clean technical run shows whether the action examined is transmitted as intended. It does not establish lead quality or campaign profitability. Those questions require additional process and business data.

Treat missing and duplicate signals as a diagnosis with a clear scope. When several sources represent the same business action, make an explicit measurement decision and document the goals. Only then does the reporting provide a useful basis for campaign decisions.