The average employee spends close to a fifth of the workweek hunting for information a coworker already has, or tracking down the person who has it. (mckinsey.com) Multiply that across a ten-person team and you are paying for roughly two full salaries a year on a search engine that does not work.
A knowledge management system fixes that, but the phrase gets used loosely. It is not a shared drive full of old files, and it is not a wiki nobody has touched since month three. It is a small, owned system for capturing the decisions and procedures that currently live only in someone’s head, organizing them so a coworker can actually find the answer, and keeping the whole thing from going stale.
What most SMBs have instead
Most small teams already have fragments: a Google Doc of onboarding notes last touched in January, a pinned Slack message with the vendor account numbers, a founder who is the only person who remembers why one client gets a nonstandard discount. None of this is written down badly on purpose. It just never had an owner or a place to live, so it stayed tacit until the one person who knew it left, got sick, or was unreachable during the exact week a client needed the answer.
Layer one: capture what only exists in someone’s head
List the five questions coworkers or new hires ask most often, and write down the real answer to each one, not the polished version. A flexible workspace tool keeps the barrier to starting low: Notion’s free plan covers a solo founder, and its Plus plan runs 10 dollars per member a month for a small team that needs shared, permissioned pages. (notion.com) The goal of this layer is not a beautiful document, it is getting the answer out of one person’s head and into a place a coworker can open without a Slack message.
Layer two: organize it so it survives being searched, not just written
A pile of documents is not a system if nobody can find the right one in under a minute. This is where a tool built specifically for internal knowledge earns its keep over a generic workspace: Slite runs 10 dollars per user a month on its Basic plan and includes a doc-verification workflow that flags pages nobody has confirmed as current, which is the exact failure mode that kills most company wikis. (slite.com) Tettra is a cheaper alternative built for the same job at 8 dollars per user a month, though it carries a 10-user minimum that makes more sense once a team has outgrown a five-person shop. (tettra.com)
Layer three: give it an owner and a review date
A system with no owner decays within a quarter. Assign one person, not a committee, to own the knowledge base part-time, and put a recurring 30-minute slot on their calendar every month to review the five most-viewed pages and flag anything that looks out of date. This is the layer that separates a knowledge base that is still useful in a year from one that quietly turns into digital debris nobody trusts enough to open.
What this does not fix
Writing things down does not replace judgment. A documented refund policy still needs a person to decide when a customer’s situation is the exception the policy did not anticipate. The point of a knowledge management system is not to make anyone’s judgment obsolete, it is to stop burning that judgment on the same repeated question so the people who have it can spend their attention on the calls that actually need a human. A system that frees a manager from answering the same onboarding question for the tenth time is a system that gave that manager an hour back, not a system that made them replaceable.
What it costs to get this wrong
Every unwritten answer is a single point of failure. When a business continuity test reveals that only one person knows how to restart a vendor integration, the fix is not a better disaster plan, it is the knowledge system that should have captured that procedure months earlier. The same discipline that turns a fuzzy process into a documented, testable work instruction is what makes a knowledge base worth maintaining in the first place: write it down once, in a place with an owner, and stop re-explaining it from memory.
Pick your single most-repeated question this week and write the real answer down before you answer it out loud a third time.
