Every list of challenges in this area contains culture, adoption and leadership buy-in, none of which tells anybody what to do next week. The five below are ordered by when they arrive, because a company that knows the next one is coming handles it far better than one that meets it as a surprise in month seven.
Month one: nobody will own an article
Ownership is a standing cost with no visible credit, so the first honest conversation is about who carries it. Where a question has no willing owner, that is usually a question about a process nobody owns, and finding that out in month one is a gift rather than a delay.
Month three: people still ask instead of searching
Expected and frequently mistaken for failure. Habits change after searching has personally worked several times, and the only lever that speeds it up is putting the search where the question is already asked: in the support tool, in the chat channel, in the onboarding checklist.
Month six: coverage plateaus
The easy articles are written, the remaining questions are the hard ones, and the writers have gone back to their jobs. This is where a program either gets a standing slot in somebody's week or quietly stops, and the choice is usually made by default rather than deliberately.
Year one: the first confidently wrong article
Somebody relies on a page that stopped being true and it reaches a customer. The damage is to trust rather than to the process, and the recovery is the visible unverified mark on overdue articles, which should have been in place from the start and almost never is.
Year two: the base is bigger than anyone can maintain
Search quality falls because near-misses accumulate, and the instinct is to reorganise. The thing that works is deletion: articles about processes the company no longer runs, duplicates that were never merged, pages that answer nothing. A smaller base searches better, and nobody enjoys the meeting where that is proposed.
Questions people ask about knowledge management challenges
Is culture really the problem?
Rarely, and naming it as the problem is how a program avoids the mechanical fixes that would have worked. Ownership, a consequence for an overdue review and the search being where the question is asked are all mechanical, and they resolve most of what gets called culture.
Which of these kills the most programmes?
The month six plateau, because it looks like completion. Coverage of the easy half feels like success and the questions that remain are exactly the ones whose answers were hardest to extract.
How do we prepare for year one?
Put the unverified mark in from the beginning, even while everything is fresh. It costs nothing when nothing is overdue and it is very hard to introduce after the first incident, because by then it reads as an accusation.