This page describes a proposed study. It contains no campaign performance findings and is excluded from search indexing.
This is a research protocol, not a published tracking-error study. No prevalence rates or advertiser outcomes are claimed. The purpose is to define a controlled experiment that can later produce inspectable evidence.
Download the tracking protocol ↓Research question and scope
Measure whether a known implementation failure produces the expected browser observation and diagnostic result. Use synthetic pages and business actions, with no customer data. Separate browser instrumentation behavior from platform receipt and advertising attribution.
The study should not estimate how common a problem is across advertisers unless a separate, defensible sampling frame is collected.
Controlled cases
Run a clean baseline and then introduce one condition at a time: duplicate installation, consent denied, consent granted then withdrawn, missing success callback, wrong source, repeated confirmation, and browser/server retries with stable or changing IDs.
For each case, record the fixture version, expected business actions, emitted event calls, IDs, network result where tested, and diagnostic observation. A local mock response must be labeled as simulated.
Publication criteria
Publish results only after the fixtures, test procedure, observations, and limitations are available together. Include negative results and unresolved cases. Have another operator repeat the procedure before using it as evidence of product behavior.
Do not describe an instrumented test as a production attribution study. If platform receipt is tested later, identify the account environment and authorization without exposing credentials.