How We Deprecated 500+ Articles: Cleaning Up Documentation at Scale
summary
A systematic, data‑driven cleanup retires 500+ articles — about 30% of the knowledge base — over 5–6 months using a three‑phase process: age filters, ticket and traffic signals, and focused audits on top categories. Actions are merging high‑value content, removing duplicates, and enforcing templates; a four‑person team runs the work and tracks results in a Power BI dashboard. Bot deflection rises from 25%→39%, and early tooling experiments use Claude Gemini Google scripts.
AI-written summary · about
chapters (27)
- 0:00Introduction
- 1:24Why deleting matters
- 3:09Help center chaos
- 5:21FAQs overwhelmed users
- 7:51Support team’s claim
- 9:16Why removal felt impossible
- 9:58Three-phase cleanup
- 10:43Phase one: outdated content
- 13:10Phase two: data signals
- 15:38Phase three: cleanup at scale
- 17:44Folder and category cleanup
- 18:28Before and after: merging conflicts
- 19:52Impact of cleanup
- 20:34Challenges and considerations
- 21:55AI tools for audits
- 23:00Key takeaway: kept articles are noise
- 23:22Q&A
- 24:24Q&A understaffed teams
- 25:28Q&A AI response quality
- 25:50Q&A deletion metrics
- 26:52Q&A restricting contributions
- 27:15Q&A project timeline
- 27:58Q&A FAQs and GEO
- 28:40Q&A preventing regression
- 29:01Q&A bot testing
- 29:23Q&A reducing sprawl
- 30:05Q&A ownership and buy-in
related talks
37mDrop the Docs, Get Back to What’s Important!
37mTaking a step back: Analysing your documentation
27mBoost Your API with Effective Documentation Strategies
18mAutomating Product Deprecation
33mDoing Data-Driven Content Maintenance
23mDocumentation in the Age of AI: Why Docs Are a Product
45mChaos to contained How a content project can lead to better content BAU
40mRelease-ready Docs: How a 2-person team keeps a help center consistently accurate