Strategy documents in this area fail in a predictable way: they describe an end state and assign it to nobody. The version that gets acted on has four sections and fits on one page, and everything in it is a decision rather than an aspiration.
What is in scope, and what is somewhere else
Name the bases you are choosing to run and the things that are explicitly not in them: the policies, the process documents, the onboarding material. Half the value of the document is the second list, because it stops the base becoming everything.
Who owns what
Categories assigned to named people. Not roles, names. A strategy that assigns ownership to a team assigns it to nobody, because a team cannot be emailed and asked a question, and the test for this section is whether an email could be sent to a person at the end of reading it.
The review cadence, with its cost stated
Say the cadence and say what it costs in reviews per owner per month. A cadence chosen without that number is a wish. Formal regimes attach the cadence to a rule for exactly this reason: 44 U.S.C. 3507(g) caps an approved federal collection at three years so the review is not a matter of anyone's intent.
What you are not doing this year
The federated taxonomy, the AI assistant, the community of practice. Writing them down as deliberately deferred, with a sentence of why, is what stops them being raised every quarter as though they had been forgotten, and it is usually the section leadership finds most useful.
Who reads it
One page means the owners actually read it, which matters more than any content in it, because they are the people whose behaviour it is trying to change. A twelve-page strategy is read by its author and by whoever approved the budget, and by nobody who will be asked to review an article.
The one page, as it reads
Scope: the customer base at help.example.com and the internal base in the wiki's Reference space; not the policies, the process documents or onboarding, which live elsewhere. Owners: seven categories, seven names, listed. Cadence: customer articles reviewed every three months, internal every six; at the current count that is eleven reviews a month across the seven owners, agreed by each. Deferred this year: the assistant, until two review cycles have run; a second language; the community of practice. Reviewed: annually, by the operations lead, and this page has an owner and a date like everything else. That is the entire document, and every line of it is something a person can be asked about.
Questions people ask about knowledge management strategy
Should the strategy come before the base?
A one-page one, yes, because the ownership decision is genuinely hard to retrofit. A twelve-page one, no: it delays the thing that produces learning.
How often should the strategy itself be reviewed?
Annually, and it should be one of the articles with an owner and a review date like everything else. A strategy exempt from its own cadence is a poor advertisement for it.
What if leadership wants a bigger document?
Write the one page and attach the planner's output. The numbers do more persuading than the prose, because they turn an intention into a monthly commitment somebody has to agree to.
Who signs off the one page?
The owners named in it, by agreeing to their monthly number, and whoever leads the function. A strategy nobody has agreed a number under is a description; a strategy seven people have each said yes to is a commitment with names on it.
What if an owner leaves?
Their categories are reassigned on the day, by name, and the record shows the change. That is the one event a strategy has to survive, and it is the reason ownership is transferable rather than attached to the author.