Conversion tracking is a decision system, not a collection of tags. The business must define valuable outcomes, each event needs one accountable implementation path and consent and privacy requirements must be respected in every market served.
What this guide covers
Design a goal hierarchy, select one tag owner, test deduplication and consent behavior, and reconcile platform events with source systems before allowing automated bidding to rely on them.
Define the goal and value model
Start from business outcomes before selecting event names or tag tools.
Separate primary outcomes from observations
Phone clicks, page depth and form starts may help diagnosis but should not automatically guide bids.
- Name the commercial decision for each goal.
- Choose an appropriate counting method.
- Assign values only when they represent meaningful differences.
Create a source-of-truth map
Document where each outcome originates and how it can be reconciled.
- Map website, call, CRM and purchase sources.
- Choose stable event identifiers.
- Define the accountable business owner.
Choose one implementation owner
Google tag, Google Tag Manager and connected platforms can coexist, but duplicate ownership of one event causes inflated data.
Document the tag path
Record which tool loads the base tag and which sends each event.
- Inspect page source and tag diagnostics.
- Remove legacy or theme-level duplicates.
- Keep test and production identifiers separate.
Test single-event behavior
Refreshes, back-button navigation and payment returns can resend events without safeguards.
- Use unique transaction or event identifiers.
- Test repeated confirmation-page access.
- Compare browser, platform and source-system counts.
Implement privacy and first-party signals responsibly
Consent requirements and permitted data use vary by market and business context.
Test consent behavior
A consent-mode signal is not a substitute for a real user choice and accurate policy.
- Map categories to tag behavior.
- Test accept, reject and change-choice paths.
- Avoid collecting data before the applicable choice.
Evaluate enhanced and offline conversions
These features require appropriate notices, data controls and reliable matching processes.
- Review legal basis with qualified advisers.
- Normalize and secure permitted first-party data.
- Monitor upload quality and correction procedures.
Validate and monitor the full lifecycle
Tag-assistant success does not prove downstream import, attribution or value quality.
Run controlled test scenarios
Use real browser, device and consent states and record expected results.
- Test success, failure and duplicate submission.
- Confirm goal status and diagnostics.
- Check values, currencies and attribution timestamps.
Reconcile on a recurring basis
Platform data will not always match source systems exactly, but unexplained shifts need investigation.
- Compare daily and weekly totals.
- Set alerts for material changes.
- Retest after site, CMP or tag releases.
Primary sources
Platform features and policies change. Review the current primary documentation before implementation.