Describe what was measured
Every statistic must identify its population, unit, cohort, time window, exclusions and outcome definition.
Public research commitment · effective 2 November 2026
SignalCraft Research Data Standard
The minimum information SignalCraft will publish beside any aggregate statistic, benchmark or claim derived from product activity.
Do not upgrade a proxy into an outcome
Tool usage may describe engagement. It must not be labelled customer demand, revenue, retention or startup success.
Eight publication requirements
Population and unit
State whether a number represents people, sessions, ideas, reports, sources or events. Never switch denominators without saying so.
Cohort and time window
Publish the inclusion dates, geography, category filters and whether the cohort is fixed or continuously changing.
Outcome definition
Define exactly what happened. A click, validation rerun, landing page, paid pilot, purchase and retained customer are different outcomes.
Sample and exclusions
Report the sample size, missing records, duplicate handling, test or bot exclusions and any minimum-data threshold.
Method and model version
Describe how fields were classified, which rules or model version were used and whether the analysis changed during the period.
Uncertainty
Show ranges, limitations and plausible alternative explanations. Avoid precision the data cannot support.
Privacy and aggregation
Publish only appropriately aggregated results and review whether small groups could reveal private or commercially sensitive information.
Correction history
Date every release, preserve material definition changes and provide a way to challenge or correct a claim.
Language SignalCraft will reserve
Appropriate
- Associated with
- Observed within this cohort
- Platform action
- Directional pattern
- Model assessment distribution
Requires direct outcome evidence
- ×Predicts startup success
- ×Proves product-market fit
- ×Validated demand
- ×Likely to succeed
- ×Customers will pay
Current status
SignalCraft does not currently publish a success benchmark from its assessments. Product activity is not an observed business outcome, and the available sample is not presented as representative of the wider founder population.