2690
← All workflows

Find broken conversion tracking

Ad reports change and nobody knows whether sales or tracking changed.

For teams facing an unexplained conversion-report gap after a site or measurement change.

What good looks like.

A discrepancy is classified as reporting difference or a reproducible measurement fault with verified repair evidence.

Starts when
Reported conversions change sharply or tracking code is released.
Accountable owner
Measurement owner

Where AI could help.

AI can summarize dated discrepancies and release notes; an analyst verifies event semantics and avoids treating attribution differences as lost orders.

How the work could move.

A suggested process to adapt to your team. Each handoff needs a clear owner and a visible check.

  1. Align the reports

    Analyst

    Compare dated orders and platform conversions using stated timezone, attribution window and order status.

    Ready when: The comparison documents expected differences.

  2. Trace a known event

    Measurement owner

    Follow a permitted test checkout through browser and server events where configured.

    Ready when: Expected identifiers and event counts are recorded.

  3. Inspect recent changes

    Developer

    Link releases, consent behavior and integration changes to the first discrepancy.

    Ready when: A reproducible fault or a remaining hypothesis is documented.

  4. Repair and retest

    Developer

    Fix the confirmed missing or duplicate event and repeat affected test paths.

    Ready when: Event evidence matches the intended contract.

  5. Close the interpretation

    Analyst

    Record remaining attribution limits and notify the budget decision owner through the existing review.

    Ready when: Spend interpretation uses a documented measurement status.

Explore the setup, checks, and exceptions

Start with your tools.

Use redacted order exports, ad reporting, an event debugger and your release log. Follow the store’s approved test-order process.

When things go wrong.

  • A purchase can appear in multiple platforms without representing multiple source orders.
  • Consent-dependent behavior must be tested rather than bypassed.
  • Delayed reporting requires a later recheck before declaring a repair unsuccessful.

A useful first step.

Trace one approved test order across source and analytics records before diagnosing aggregate totals.

What to measure

Track reproducible missing or duplicate events and time from fault confirmation to verified repair.

Want this working with your team and tools?

Build it yourself, adapt what you already have, or work with us. We can help you choose where to start and handle the technical build.

Talk about your idea

May we use Google Analytics to understand visits and improve this site? Analytics cookies are optional. Privacy details