Every business has tried this. A shared drive, then a wiki, then an intranet, then whatever the newest tool was. Each one starts well, and each one is abandoned within a year.
The failure is never the software. It is that all of them depend on people tidying up.
Why they rot
A knowledge system asks busy people to do work with no immediate payoff — file this correctly, tag it, link it to the related thing, update the page that is now wrong. The benefit accrues to someone else, later. So it does not get done, and within months the system contains a mixture of the true, the stale and the contradictory, with no way to tell which is which.
At that point people stop trusting it, which means they stop using it, which means it decays faster. Nobody decides to abandon it. It just quietly stops being consulted.
A knowledge system that depends on people tidying it will die within a year. Every business has proved this to itself at least once.
The part that actually matters
The fix is not better discipline or a nicer interface. It is that the tidying has to happen without anyone doing it.
Overnight, something reads what came in that day, files it, links it to everything related, and flags where it contradicts a decision made earlier. Not a search index — an actual pass over the material that leaves it more organised than it found it.
That is the unglamorous half, and it is the half that decides whether the thing survives contact with a real company.
And then it has to produce something
The second mistake is treating this as memory. Being able to ask questions of your own material is genuinely useful, and it is not worth building a system for on its own — people ask a handful of questions a week and then forget the tool exists.
What earns its place is production. The knowledge is only valuable when it comes back out as finished work: a quotation with the right prices and the terms this customer actually agreed, a proposal written for this business rather than a template with the name changed, a compliance filing already checked against the rule.
Which is why we describe it as three moves rather than two: know, connect, produce. Most products stop at the first, a few reach the second, and almost none do the third — because the third is where you have to encode what your company thinks good looks like.
What this means practically
- If it needs people to file things, it will fail — assume that and design around it
- If it only answers questions, it will be forgotten — measure it on work produced, not queries served
- If it cannot tell you when two pieces of your own material disagree, it is a search box
- If what comes out still needs a senior person to check every line, nothing has been removed
The test is simple enough. A year from now, is it more organised than the day it was set up, or less? Everything else follows from that answer.