Define the dry-run exposure tuple
Synthetic start check requires one dry-run tuple for each A/B variant, with assigned, rendered and recorded event tags identical. Run TA is A/A/A; TB is B/B/A. Local start is held until both tuples pass. These are provided test records, not actual analytics or ad-platform capabilities.
| Run | Assigned/rendered/tag | Decision |
|---|---|---|
| TA | A/A/A | Pass |
| TB | B/B/A | Fail event tag |
Compare all three variant fields
- Require both dry-run identities
- Compare assigned/rendered/tag
- Correct TB’s recorded variant tag
- Repeat TB before recording start
A successful B screenshot alone cannot verify its recorded event tag.
How many dry runs pass, and can start be recorded?
Show the worked answer
One of two passes; start remains held. TB’s rendered B is wrongly tagged A, which would mix the supplied variant records.
Hold start for the mismatched tag
| Case | Decision or result | Reason |
|---|---|---|
| Use render-only check | Misses TB tag error | Rendered B but recorded A |
| Rerun TB as B/B/B | Can satisfy missing check | Complete coherent tuple |
Keep the dry-run input and observed event tuple with the start record. A planned correction is not a passing rerun.