An article nobody trusts is worse than none.

Every article has somebody responsible for it and a date it was last checked. A reader can see whether they are looking at this year's answer, and an owner gets told when theirs has gone quiet for too long.

KnowledgeReturns policyLast checked 2024Review
FlaggedGone stalebefore a customer finds out
Owner toldBy namenot a shared inbox
The AI answers from this. Stale is not a tidiness problem.
Eighteen months laterThe policy changed. The article did not

Nobody owned it, nothing dated it, and your chat is still telling customers the old answer at eleven at night.

Why wikis fail

Four reasons an article stops being true.

None of them are about writing. They are about what happens in the eighteen months after somebody wrote it and moved on.

Nobody owns it

So nobody notices when it stops being true

No date on it

A reader cannot tell this year from three years ago

Written once, at launch

When the policy was different

Nobody knows it exists

So the wrong answer never gets corrected

The AI answers from these articles, so an article going stale is not a tidiness problem.

Ownership

Somebody is responsible, by name.

Every article has an owner and a section has one too, so a policy that changes has a person to change it rather than sitting there being quietly wrong until a customer is told the old answer.

An owner gets

  • Their listEverything they are responsible for, in one place
  • A nudgeWhen something has not been checked for too long
  • What changedWhen somebody else edits their article
Visible age

A reader can see how old the answer is.

The date it was last checked sits on the article, so somebody reading it knows whether to trust it or ask. That single line does more for a wiki's credibility than any amount of formatting.

On the page

  • Last checkedNot just last edited by somebody fixing a typo
  • By whomSo a reader knows who to ask
  • Flagged when oldRather than looking identical to yesterday's
Versions

Every change is kept, and any of them can come back.

Somebody rewrites a policy at 6pm and gets it wrong. Every version is kept, you can see exactly what changed, and putting it back is one action rather than an archaeology exercise.

History

  • Every versionWith who wrote it and when
  • What changedShown, rather than two documents to compare by eye
  • Roll it backIn one action

See Wiki on your own data.

Half an hour. Bring the ten questions your team asks most and watch the chat start answering them.