Documentation Decay Triage (QA)

Documentation Decay Triage (QA)

1. 01

   Set a review clock per document class before decay starts costing delivery time

   Tag every page with a review interval: client-facing process docs at 90 days, internal runbooks at 180 days, evergreen brand assets at 365 days. Slite's drift detection and GitBook's stale-content agent both surface candidates, but the interval is your policy, not the tool's default.

2. 02

   Pull the staleness queue and rank by client exposure, not by age

   A 200-day-old internal note matters less than a 40-day-old onboarding doc a retainer client reads weekly. Score each flagged page on how many active accounts touch it.

3. 03

   Verify the page against the live system it describes

   Open the actual Slack channel, Asana project, or Google Drive folder the doc references. Slab's unified search and Tettra's Slack bot both answer from stored text, so a wrong page propagates wrong answers at machine speed.

4. 04

   Assign one owner by name and give them a 48-hour correction window

   Route the fix to the person who performs the task, not the person who wrote the page. Guru's auto-verification model works the same way: usage and ownership drive accuracy, not seniority.

5. 05

   Rewrite the drifted section and stamp the verification date

   Change only the false portion. Full rewrites reset reviewer familiarity and inflate the next review cycle. Nuclino's Sidekick summarization can draft the replacement paragraph, but a human approves the diff.

6. 06

   Log the correction in a monthly decay report for the account lead

   Track pages flagged, pages fixed, and pages still open past 48 hours. A rising open count is an early signal that a delivery team is over-allocated.

7. 07

   Retire pages that no longer map to a live client or workflow

   Archive rather than delete so historical context survives turnover. A knowledge base that only grows becomes a search liability within two quarters.