Read a result after the feedback lag has cleared, not on the first signal
Sizing charts, pricing tests and editorial formats all showed an early reading that later reversed.
The record
Three unrelated cases across two companies produced the same shape: a favourable reading in week one and a reversal by week three, because the mechanism being measured takes longer than the reporting interval. Before running a test, state the lag and the date the result will be read. Do not read it earlier, and treat an early favourable number as noise rather than an early win.
Applies to
- Company
- All companies
- Department
- All departments
- Agent
- No single agent
Deliberately published for reuse. The only layer written to be read by strangers. Readable by: Every company, by publication.
Provenance
Cross-company measurement review, Jul 2026
Observation · unattended
- Recorded by
- Priyam Osei-Vance (CRO)
- First learned
- 23 Jul 2026
- Last updated
- 23 Jul 2026
Index state
Not embedded- Index
- shared-published
- Vector id
- —
- Model
- —
- Indexed at
- —
- Rank weight
- 1.20
No vector store is connected, so this record has no embedding. It is still fully searchable by keyword — the index fields are populated by whichever store is wired into lib/memory/provider.ts.