Playbooks
Practitioner playbooks on knowledge management. Twenty years building and reviving KM at scale — written for the ops manager who inherited the wiki side-of-desk, the team lead trying to make documentation actually stick, and the founder who keeps answering the same question in Slack.
“Poke around the wiki” is not onboarding — it’s abdication with a link attached. Fifteen articles in deliberate order, checkpoints that catch drift early, and a new hire whose confusion is the cheapest audit your knowledge base will ever get.
The knowledge base holds the how-tos; nowhere holds the what-we-chose-and-why. So decisions get relitigated, reversed by accident, and inherited as rules without reasons. One line per decision — what, why, who, revisit-when — captured in thirty seconds at the moment it lands.
Every question answered in chat instead of in a document is a small loan against your future attention — and the interest compounds as repeat questions, slow onboarding, and bus-factor risk. Size the debt with one week of logging, pay down the 20% that matters, and stop the new borrowing at the moment of answer.
A knowledge base with no owners rots — every article is everyone's job and therefore no one's, so it drifts out of date until people stop trusting it. Assign a named owner to every article that matters, set a review cadence, and make ownership visible so staleness has someone accountable.
"I sent it to everyone" never proved anyone read it — and for safety, compliance, and conduct policies, that gap is the one that bites. Reserve attestations for the documents with a real consequence, write them so the acknowledgment is honest, and track completion so it produces evidence, not a spreadsheet.
When your KB chatbot fails on a real question, that miss usually evaporates — and it was the best signal you had. Treat unanswered questions as a demand-ranked backlog of the articles you're missing, and hold a boring weekly hour to turn the top of the list into content.
The person who knows why the Henderson account invoices differently just gave notice, and you have two weeks to extract a decade. The emergency protocol that actually works — and the zero-dollar system that means the next departure isn't a fire drill.
Your team has the docs and still can't find them — so they ask you or rebuild the answer. The problem isn't your content, it's findability. The non-librarian's system for organizing, titling, and tagging so answers surface in two searches — and where AI actually helps.
You wrote the SOP, it's in a folder, and nobody follows it — and it's not your team's discipline, it's the writing. The six reasons SOPs get ignored, the principles of one people actually use, and the only test that tells you it's done.
Your wiki isn't dying because your team doesn't care. It's dying because the workflow you inherited was always going to produce a graveyard. The fix: audit by usage, kill duplicates, capture from real activity, let AI watch so humans can decide.