The idea, in plain English
Start here.
Conversion measurement connects defined actions with advertising reporting under the account’s settings. Begin with what actually happened in the business, then check the event implementation and reported action separately. A submit-button click is not an accepted inquiry, and a tag firing does not by itself prove a correctly attributed conversion. Keep duplicate paths, counting rules and missing evidence visible.
A visitor presses Send, but the form rejects a missing field. If the conversion event fires on the button press, the dashboard may receive a success signal for an inquiry that never arrived. The first fix is a clear definition of success, followed by a trace you can inspect.
A few words you’ll meet.
OPEN WHAT YOU NEED ↙Business record
The system’s evidence that the intended action occurred, such as an accepted inquiry record or confirmed order.
Event trigger
The condition that causes measurement information to be sent. It should correspond to the intended action, rather than an earlier convenient interaction.
Primary and secondary
Action settings that affect reporting and optimization in combination with the campaign’s selected goals. Custom goals have a documented exception.
Attribution
The assignment of reporting credit to eligible interactions under a defined model and window. It is not proof that an ad alone caused the outcome.
Define the moment the action becomes true
Write one sentence naming the outcome, the success condition and the system that confirms it. For the fictional repair studio: ‘An inquiry is accepted when the server stores a valid request and assigns its request ID.’ Opening the form or pressing Send is earlier behavior. A paid repair is a later outcome that needs its own definition.
Include the important failure cases. An invalid form, rejected payment or canceled booking should not become the completed outcome merely because the visitor reached a similar screen. Identify the responsible owner and the evidence they can inspect. A success message is useful, but it should agree with the actual accepted record.
Trace how the information reaches measurement
List the possible senders for the same outcome: direct site code, a tag manager, an analytics import, a platform integration or a later business-system import. A missing sender can leave a gap; several uncontrolled senders can duplicate the signal. Choose a documented route and explain any deliberate additional records and duplicate handling.
Keep the event’s identity and meaning consistent. If a supported setup uses an order or event identifier, check that it represents the same real outcome on retries. Do not assume that a new page load means a new inquiry. Work with the site’s existing data-choice and privacy configuration; a server-side route or hashed value does not make all collection automatically appropriate or complete.
Three layers. Three checks.
Check the action settings and selected goal
For standard goals, Google says primary actions are used for bidding when their associated goal is selected, while secondary actions generally appear for observation in All conversions. A custom goal can use an included secondary action for bidding. Review the actual campaign goal as well as the action label; ‘secondary never affects bidding’ is too broad.
The One or Every count setting belongs to each conversion action. Google describes One as one conversion per ad click for that action, rather than one customer forever across the account. Every can count multiple valuable actions after an interaction. These settings do not repair an incorrect trigger or automatically merge separately defined duplicate actions. Record value, currency, window and attribution choices where relevant instead of copying unexplained defaults.
Verify the record, the signal and the report
Use a controlled permitted test path and save the timestamp, expected result and non-sensitive test identifier. Check the business record, inspect the event with appropriate diagnostics and review the intended conversion action after its normal processing interval. Google lists Tag Assistant as a verification tool. One green status alone does not establish all three layers.
Test an invalid attempt, one accepted action and a repeat or refresh path. Distinguish event not sent, event rejected, event not attributed and event not yet reported. Attribution windows, reporting dates, processing and available data can explain differences between systems; they are reasons to investigate, not to force totals to match. Repeat the relevant checks after a material form, tag, goal or import change, and avoid judging the newest incomplete period as if it were fully mature.
One accepted request, three test paths
Original fictional acceptance test. Q17 is a dummy request ID, not personal data or a live conversion. The table defines expected business/event behavior; it does not predict a Google Ads report count.
| Test path | Business evidence | Expected measurement behavior |
|---|---|---|
| Send with a required field missing | No accepted request record | Do not send the completed-inquiry event. |
| Correct the form and submit successfully | One accepted dummy request: Q17 | Send the intended success signal through the documented route. |
| Refresh the confirmation for Q17 | Still the same accepted request | Do not invent a second business outcome; verify repeat handling and settings. |
Put it into practice
Write a three-layer verification plan
The fictional studio fires its success event on every Send-button click. It also imports a similarly named inquiry event from another system. No actual account access or tracking change is required for this exercise.
- Define the accepted inquiry and its confirming record.
- State the expected behavior for an invalid attempt and one successful dummy request.
- List both possible senders and the duplicate question to resolve.
- Record which action, count setting and selected goal an authorized implementer must inspect.
- Explain why an event appearing in a debug tool is not the final report or business check.
Keep your working notes in your own document. This page saves only your completion check, not your answers or customer data.
Check your understanding
Make the call.
Choose an answer for each question. Then compare the reasoning.
You have a useful next step.
- Define the accepted outcome before choosing a trigger.
- Map senders, count settings and campaign goals.
- Verify the business record, event and report separately.
Completion is your own learning record. Quiz scores and practice checks are saved separately, on this browser.
Sources & context
The sources below explain the factual foundations and editorial methods used in this lesson. Worked examples, planning models and exercises are illustrative teaching material.
- Google Ads: About conversion measurement
Primary overview of action sources and verification diagnostics. No fixed learning-period, reporting-delay or identity-document instruction is adopted.
- Google Ads: Primary and secondary conversion actions
Primary standard-goal behavior and the custom-goal exception for secondary actions.
- Google Ads: Conversion counting options
Primary One/Every scope per action and interaction; not a global customer or duplicate-action deduplication rule.
