Metrics
Evidence Freshness: Calculation and Review Cycles
Evidence freshness is the share of evidence verified within its review cycle, with cycles set by evidence type.
- Author
- Xingge
- Published
- 2026-11-09
- Updated
- 2026-11-09
- Level
- L2 Reproducible
- Version
- 1.0
Term details
- Name
- Evidence Freshness: Calculation and Review Cycles
- Type
- Metrics
- ID
- evidence-freshness-en
Evidence Freshness: Calculation and Review Cycles
Evidence freshness measures how many evidence entries are still within their review cycle. It is calculated as valid entries divided by total evidence entries, with cycles set by type and counted monthly.
Definition and calculation standard
Add the review cycle to the last verification date to get the due date; if the due date is not earlier than the counting date, the entry is fresh. The denominator is the total number of evidence entries, and the numerator is the number of fresh entries.
| Type | Review cycle | Note |
|---|---|---|
| Methods and steps | 365 days | Review earlier when mechanics change |
| Glossary and metrics | 180 days | Review earlier when standards update |
| Experiment data | 365 days | Review earlier when a new evaluation round is published |
| Templates and checklists | 180 days | Review earlier when templates are revised |
| Q&A entries | 365 days | Review earlier when the basis changes |
Collection and judgment
- Read the last verification date. Use the date in the register, not the page publication date.
- Match the review cycle. Take the cycle for the evidence type, and classify unclassified entries first.
- Calculate the due date. Add the cycle to the last verification date.
- Mark the status. Entries whose due date is before the counting date are marked expired and enter the review queue.
- Aggregate the ratio. Count fresh entries and total entries, calculate the ratio, and keep one decimal place.
Data sources and record fields
Data comes from the evidence register, with fields including number, type, last verification date, review cycle, due date, status, and owner.
Due reminders are generated by a scheduled task, and the last verification date is updated after manual review.
Limitations and misreadings
- Fresh is not correct. An entry verified on schedule but factually wrong still counts as fresh.
- Cycles are set uniformly by type, so individual differences may be averaged away.
- The last verification date depends on manual entry; missing entries distort the ratio.
- This metric does not reflect when the external source itself was published, only this site's verification action.
Relationship to other metrics
The division of labor is clear: the broken evidence rate looks at link usability, and freshness looks at whether review was completed within the cycle.
When both are low, fix verifiability first and then address coverage.
Related sections: metric index, method index, glossary index, research index, checklist index.