Governance has a bad name here because it usually arrives as a monthly forum with an attendance problem. The work it is meant to do is real and small: somebody has to decide who may publish, what happens when an article goes unreviewed, how a disagreement about content is settled, and when a page is deleted rather than updated. Four decisions, written down once.
Decision one: who may publish
Anyone, anyone with review, or a small set of editors. All three work at different sizes. Open publishing works while the writers all know each other; review works up to the point where the reviewer becomes a queue; editors work but slow contribution to a trickle. Pick deliberately and revisit it when the base doubles.
Decision two: what an overdue review means
If nothing happens when a review date passes, you do not have review dates. The choices are a message to the owner, a visible mark on the page telling readers it is unverified, or unpublishing. Most companies should use the middle one, because it is honest to the reader and does not lose the content.
Decision three: how content disputes end
Two people will disagree about what the article should say, and the argument will be about a real edge case. Name in advance the role that decides, usually the owner of the process the article describes rather than the owner of the base. Without that name, the article ends up saying both things.
Decision four: when a page dies
Articles that are wrong should be deleted, not updated into vagueness. A rule that says a page describing a process the company no longer runs is removed, with a redirect where an obvious successor exists, keeps the base smaller than it would otherwise be, which is the single biggest favor you can do its search.
Who holds these
One person with an hour a week and the standing to chase an owner. The title varies and the requirement does not: they must be able to ask a manager why their team's article is four months overdue and be answered.
What governance is not
It is not a template every article must use, not an approval workflow for typo fixes, and not a quarterly audit producing a spreadsheet. Each of those produces compliance activity and no better answers, and each is the thing people remember when they say governance failed here before.
Questions people ask about knowledge management governance
How often should articles be reviewed?
By class rather than uniformly. Pricing and policy answers need a short cycle, how-to articles about stable internal systems can go a year, and a single global interval guarantees you are wrong in both directions.
Do we need a documented knowledge management policy?
One page, if anything external asks for it. The four decisions above fit on that page. A longer document is a signal that governance is being designed rather than performed.
What should be reported upward?
Overdue articles by team and the count of questions asked that the base did not answer. Both are actionable. Article counts and page views are not, and reporting them invites the wrong incentive.
Does a small company need any of this?
It needs decisions one and two and can defer three and four until the first argument and the first dead process. Deferring is fine; not knowing you deferred is what produces the base nobody trusts.