Evidence
Research and the bar it has to clear
Research here means a dated run with a written method, a fixed sample and limits printed beside the finding. Nothing has cleared that bar yet. When a study does, it appears on this page with its method attached. Until then the page describes the standard instead of results that are not ready.
What counts as research here
A run counts when the method was written before the data was collected, the sample was fixed in advance, and another reader could repeat the steps and see whether the same answer comes back.
An interesting observation from day-to-day work is a note. It does not get presented as a study.
The publication bar
| Method | Written before the run, published with the result. |
|---|---|
| Sample | Fixed and stated, never extended after seeing the data. |
| Date | The run date travels with every figure. |
| Limits | Printed next to the finding, not buried at the foot. |
| Review | Rows a machine produced are read by a person before publication. |
What is in progress
A restaurant bookability sample is being prepared. It stays unpublished while the website-resolution step is unreliable, because a run that mostly measures its own resolver would describe our tooling, not the market.
No figures from that work appear anywhere on this site.
What is not research
Output from an internal pilot, vendor marketing figures, a single screenshot, and anything we cannot date. These can inform an article. They do not carry a study's authority.
Sources
- Local record: bookability check specification
When a study is ready, the method arrives with it.