MEASUREMENT DECISION BRIEF
James Chen (Chen Yiming) | Media Buyer & Campaign Operations
Portfolio: https://www.terrascale-media.com/
Method prepared 2 November 2026; published 4 November 2026. Not a client report.

Use this record to distinguish an observation from an explanation.
Do not enter account credentials, personal data, private logs or client IDs.

1. DECISION
What action is being considered, by whom, within what existing authority?
What would change if the evidence were different?

2. OBSERVATION
Observed symptom:
Reporting systems and owner of the business event:
Outcome definition (event / distinct person / FTD / approved FTD / order):
Reporting window, timezone, status cutoff and cohort maturity:
Known missing, delayed or contradictory evidence:

3. BOUNDED CHECK
Symptom: Tracker contains a conversion; traffic source does not.
Request: Event identity, correct attribution ID, endpoint response and source receipt.
Does not prove: HTTP 200 alone does not prove business approval or downstream reporting.
Next check: Connect the same authorized real event across the systems. Do not send a
synthetic event into live optimization. Follow the receiver owner's approved procedure.
Guide: https://www.terrascale-media.com/redtrack-conversion-missing-traffic-source.html

Symptom: A conversion seems to appear twice.
Request: Stable scoped business-event ID, delivery attempts and state history.
Does not prove: The same person or click can contain separate legitimate events.
Next check: Distinguish a retry, an update and a separate action before changing modes.
Detailed method: https://www.terrascale-media.com/postback-retries-duplicate-conversions-event-key.html

Symptom: A creative's ranking changes after switching landing pages.
Request: All intended combinations, assignment unit, comparable traffic, mature outcomes.
Does not prove: A pooled ranking or a time-based comparison does not establish interaction.
Next check: Decide whether a four-combination design is warranted; otherwise keep the
comparison observational and state its limits. Do not change assets without approval.
Detailed method: https://www.terrascale-media.com/creative-landing-page-interaction.html

Symptom: Attributed retail ROAS looks strong during a promotion.
Request: Reconciled contribution and an appropriate control for the defined intervention.
Does not prove: Attribution alone does not show what would have happened without the ads.
Next check: Keep incrementality unknown unless the comparison supports a causal estimate.
Detailed method: https://www.terrascale-media.com/retail-promotion-attribution-incrementality.html

4. REVIEW RECORD
Evidence obtained (public or appropriately redacted references only):
What the evidence supports:
What remains a hypothesis or unknown:
Next action and owner:
Maximum permitted exposure and safety stop, if an experiment is authorized:
Review date and maturity condition:
Result after the check, including inconclusive or failed outcomes:

This brief proposes a workflow. It does not certify a production integration,
completed experiment, measured improvement or guaranteed result.
These are proposed methods, not verified client results.