Skip to main content
James ChenPortfolio by TerraScale

Illustrative reconstruction · Method demonstration

RedTrack S2S troubleshooting: trace the conversion discrepancy

Follow the event. Find the discrepancy. A way to reason through an S2S mismatch before treating it as a campaign performance problem.

This is a workflow demonstration. It does not represent a completed client investigation, a production trace or a measured improvement. No live identifiers or report data are shown.

Start with a testable question

Suppose the tracker and the destination report different conversion totals. Before changing bids, establish whether they refer to the same event, time window and population.

Ad clickTracker IDLanding pageDefined eventPostback

A structured investigation

  1. Write down the event contract

    Registration, deposit and first deposit are different events. Record the event name, inclusion rules, attribution window, timezone and deduplication basis.

  2. Trace a controlled click

    In an authorized test environment, follow an example click ID through the link and landing page. Check whether redirects or parameter handling lose the identifier.

  3. Inspect the postback path

    Check field mapping, event matching, request status and available delivery logs. A successful HTTP response alone does not establish a correctly attributed conversion.

  4. Check duplicates and delay

    Compare event timestamps and unique event identifiers. Distinguish repeated delivery from a new event and allow the agreed reporting window to mature.

  5. Reconcile the same population

    Align date boundaries and filters; separate received, accepted, attributed and reported events. Record the remaining difference and the evidence needed to explain it.

  6. Verify the correction

    Change one identified failure, rerun the controlled check and confirm the downstream report. Keep the deployment and permission steps separate from this demonstration.

The decision gate

If the event definition or identifier path is unresolved, I would not use the apparent CPA change to justify more spend. First establish what is being counted and which comparison is meaningful.

What a useful investigation note contains

Expected event; observed discrepancy; date and timezone; test identifier placeholder; evidence checked; hypothesis; change made; validation outcome; unresolved questions. Use anonymized references rather than account secrets or customer records.

Related technical notes

Continue the investigation

For repeated deliveries and changing event states, define the event key before adjusting duplicate handling.

Looking for a hands-on Media Buyer?

Share the role, the team and where the work happens.

Discuss a role