Conversion Tracking Audit: A Practical Checklist

Start with what the business counts as a result

A conversion tracking audit is for marketing and operations teams that need to explain what their advertising numbers represent. A dashboard may show purchases, inquiries, calls, and downloads together. Before using that total to guide spending, trace each action to an event the business recognizes and a system that can confirm it happened.

This workflow covers definitions, settings, technical checks, and reconciliation. Its scope depends on the implementation and the records available; an audit can identify measurement problems but cannot promise campaign results. Begin with the conversion actions used for active campaigns, then inspect supporting actions and retired implementations that could still send data.

Write an event definition that another team can test

For every action, describe the exact completion condition. A lead form submitted successfully differs from a form opened, a submit button clicked, or a sales conversation accepted. Specify whether a purchase means payment authorized, payment collected, or an order created. Ask the commercial owner to approve that definition before a developer changes the trigger.

In a hypothetical B2B software company, a pricing request creates a CRM contact, while a sales representative later accepts a qualified opportunity. The audit gives those events separate definitions and action names. It keeps a brochure download as a supporting event. The sales owner supplies the qualification rule, and the operations owner identifies the CRM field that records acceptance.

Record the source, business owner, implementation owner, value convention, and evidence location beside each definition. If two actions describe the same event, mark the overlap for investigation. Keep different funnel stages separate so a reader can distinguish one person progressing through a process from several independent commercial outcomes.

Check primary and secondary conversion actions in context

Primary conversion actions are included in the Conversions column and used for bidding when their associated standard goal is used for bidding. Secondary conversion actions are generally observed through All conversions. A secondary action included in a custom goal is also used for bidding. Inspect the goals selected for each campaign alongside the action designation.

Ask the campaign owner which event should guide optimization and why. In the hypothetical software example, the team might select qualified opportunities as its intended bidding outcome once the import is reliable enough for that operating plan. Pricing requests and downloads remain useful context for diagnosing the funnel.

Save the actual campaign goal selection in the audit record. A list of action names alone leaves an important dependency unexplained. When proposing a designation change, describe which campaigns use the goal, how their reported conversion totals may change, and who will review the measurement after the change.

Match counting and windows to the event

Google Ads offers Every and One counting options for applicable conversion actions. Every counts each conversion for that action after an ad interaction; One counts one conversion for that action per ad click. Every commonly fits purchases where each order has value. One commonly fits lead generation where repeated submissions following a click represent the same inquiry.

Counting works per conversion action. Choosing One for a form does not deduplicate a customer across separate actions, advertising interactions, or CRM records. Document the unit you intend to count and inspect duplicate records separately.

A conversion window defines how long after an ad interaction a conversion can be recorded. Compare the configured window with the business's observed time to purchase or qualification. Record the relevant interaction type and source. Distinguish this setting from the date range selected in a report, and label recent periods whose downstream outcomes are still arriving.

Follow the event from the page to the destination

Use a controlled test journey with the implementation owner. Start before the action, complete it once, and capture the event name, action identifier, timestamp, value, currency, and transaction identifier where relevant. Verify the trigger against successful completion. Then repeat the checks for validation failure, page refresh, and a return visit to the confirmation page.

In a hypothetical home goods store, one test order generates a purchase event from both a hardcoded tag and a tag manager rule. The auditor traces both senders, identifies the intended implementation, and assigns removal of the redundant trigger. A second test uses a different order to verify that legitimate new purchases remain distinct.

For the same conversion action, a repeated transaction ID helps Google Ads identify duplicate conversions. Use a distinct ID for each genuine transaction and retain that ID when the same transaction is sent again. Check both missing IDs and a mistakenly hardcoded ID. Include the site's relevant consent states in the test matrix with the person responsible for that implementation.

Reconcile offline and CRM imports by event

Take a bounded sample of CRM events and follow them through the import process. Compare action mapping, event time and time zone, identifiers, value, currency, and the status that made each record eligible. The event timestamp should describe the business event according to the agreed definition, while a separate operational timestamp records when the upload ran.

Track records prepared, submitted, accepted, rejected, and subsequently visible in reporting as separate stages. Review processing feedback before repeating an uncertain upload. Preserve stable identifiers through retries and keep a reconciliation note for corrections, canceled orders, or changed qualification status.

Use separate conversion actions for distinct funnel stages. In the software scenario, an inquiry and an accepted opportunity should remain distinguishable during import and reporting. Compare like periods and event definitions across systems, then classify differences as timing, eligibility, mapping, duplication, or unresolved. Assign each unresolved difference to an owner with the underlying record reference.

Use an audit record that leads to a repair

Keep one row per conversion action, with additional rows for specific defects. The following template is a working record, not a scorecard. Attach the test evidence in the team's authorized environment and summarize the finding in language the business owner can review.

Write findings precisely. Replace tracking broken with purchase fires on a failed payment response, or qualified opportunity import uses the upload time as the event time. Specific observations let the developer reproduce the issue and let the reviewer assess the repair.

Run the audit in a controlled sequence

Schedule the audit around the people who own the website, advertising account, and CRM. Separate discovery from implementation so the team can review changes to signals that guide active campaigns.

1. Inventory active and legacy actions, identify the campaigns that use them, and preserve their current settings.
2. Approve business definitions and document the source record that confirms each event.
3. Check goal selection, primary and secondary designations, counting, windows, values, and currencies.
4. Test successful journeys and failure paths, then investigate duplicate triggers and transaction identifiers.
5. Trace a bounded CRM sample through preparation, processing feedback, and reporting.
6. Prioritize defects by business impact, assign an owner, and describe the proposed correction.
7. Retest the exact failed scenario after implementation and schedule a later reconciliation for delayed outcomes.

Close a defect with fresh evidence from the affected journey. If a CRM stage takes time to occur, record the completed technical test and the date of the remaining business reconciliation separately.

Common mistakes that leave the numbers ambiguous

One mistake is treating every useful event as the same business result. A download, inquiry, and sale can all help explain a journey, but combining them into an unlabeled total makes the next budget discussion harder. Segment by action and identify the outcome that supports the decision.

Another is using One as a substitute for fixing repeated tags, or removing a duplicate sender without testing a new legitimate event. Audit the implementation and the counting rule together. Also check whether a legacy import continues after a replacement goes live.

Finally, a repair needs an effective date. Keep the original finding, the saved change, and the retest together so the next report can explain which periods used the old definition or implementation.

Finish with an agreed measurement record

A useful conversion tracking audit leaves the team with clear event definitions, verified settings, reproducible tests, and an owned repair queue. Start with the actions that guide active decisions. Follow each one from the business event to the advertising report, and close the loop with the commercial owner when the evidence is ready.