The owner of an accounting practice of eight people describes his week this way: "I have two senior collaborators who know everything about the historic clients, an administrative manager who knows everything about internal procedures, and a secretary who knows everything about the service suppliers we use. If one of the three falls ill, the practice slows down. If one of the three leaves, for the first two months we are in trouble".
This description is not an exception, it is the norm for an Italian SME that has grown its knowledge over time. The owner himself, unlike many, sees it with clarity. He knows he is not looking at efficiency; he is looking at dependency. The problem he cannot solve is how to structure that knowledge without blocking operations for six months, without buying an expensive system nobody will use, and without falling into the 1990s "knowledge management" cliché that in Italy is still the reason the phrase sounds bad.
Why the phrase sounds bad
Knowledge management in Italy has a short, troubled history. In the late 1990s and early 2000s, the term entered large companies as an IT project. The typical implementation was an intranet, a company wiki, a system for archiving documents in categorised folders. The HR manager, together with IT, asked collaborators to "document best practices". Collaborators, sensibly, ignored the request or complied with it formally. The wikis filled up with procedures approved by someone and never read by anyone. After a few years, the project quietly stopped.
The method was wrong for three structural reasons. First, it asked people to write in a format that is not natural to their way of working. Second, the incentive was weak: writing procedures for others, without anyone really reading them, brings no personal advantage. Third, the technological systems of the time did not know how to consume that content usefully: they produced an archive, not a cognitive collaborator.
Today the situation has changed radically on all three fronts, but many owners still have that failed version in mind. When they hear "knowledge management" they think of a heavy IT project, destined not to be used. They are right about that version. The new version is a different thing, and deserves a different name.
Where SME knowledge really lives
The knowledge of an Italian SME does not live in documents. It lives in people, in habits, in ways of doing. Think back to the opening accounting practice. The senior who knows the historic clients does not hold a list of information that can be transcribed. He holds a map of relationships: who client Rossi is, how he reacts to an urgent reminder compared to a gentle one, which questions he asks often, which members of his team are key interlocutors. This map is not a list. It is a way of seeing.
Structuring this knowledge does not mean transcribing it into a wiki. It means building a corpus that holds together different levels of information, arranged so that they can be queried and used to decide.
The first level is descriptive. It contains what the company is and does, formally: services, clients, standard procedures, internal structure. This level almost always exists, even if spread across folders, brochures and management systems.
The second level is observational. It contains the patterns the company has learned to recognise on the ground. "Clients in the commercial sector call on average three times in June with the same questions", "seventy-five per cent of complaints arrive in the two weeks following a price increase", "clients who signed in less than six weeks from the first proposal are the ones who then generate most recommendations". This level hardly exists in structured form anywhere. It lives in heads, transmitted through stories.
The third level is predictive. It combines the two previous ones to anticipate behaviour. "This client is in situation X, in the past clients in situation X have done Y, he will probably do Y". This level in SMEs is rare and, when it exists, is intuitive: someone "senses" what will happen, but cannot say it in repeatable terms.
The fourth level is prescriptive. It translates predictions into action suggestions. "Given this situation, it is worth doing Z, with these precautions". This level is the rarest and most valuable.
A knowledge management system that makes sense today makes these four levels work together. It does not produce dead documents. It produces a living corpus, growing with the daily work of the company and replying when needed.
Why the consulting approach fails
An owner decides that structuring is needed. He hires a consultant, sometimes of process, sometimes of technology. The consultant begins interviews, produces flowcharts, documents procedures. After three months he delivers a one-hundred-and-fifty-page operational manual and a configured wiki. The owner pays, the collaborators nod, the manual ends up in a Google Drive nobody opens. Six months later, the practice works as before. Knowledge is still in heads. Twelve thousand euros have been spent on a document nobody uses.
This pattern repeats consistently because it contains two errors that reinforce each other. The first is that the work is done in a mode that does not latch onto daily operations. Separate interviews, after-the-fact documentation, a consultation system disconnected from where work actually happens. The second is that the output format is static. A one-hundred-and-fifty-page manual is a heavy object: out of date after three months, not consultable at a glance, not modifiable without another round of consulting.
The version that works operates in the opposite way. It latches onto daily work: the corpus is built by observing how the company operates, not by interviewing it cold. And it has a dynamic format: it is consultable through natural language, updated continuously, used by the systems the company runs.
What changes in practice
A practice that has structured its knowledge does not have an operational manual in a folder. It has a corpus consulted in real time when someone needs it. A new collaborator, on the first day, does not receive a three-hundred-page PDF. He has access to a system that replies to his questions ("how do we usually handle a commerce client's extension request?") with specific, updated information, in the practice's tone.
A senior going on holiday does not leave operational gaps. The information needed to reply to a client in his name is accessible to whoever replaces him, with the same granularity he would give it. It does not have to be asked by phone from the Maldives.
An owner considering opening a second office does not start from zero in transmitting operational culture. The culture is partly structured, partly still implicit, but the structured part is there, consultable and multipliable.
These scenarios make the advantage evident. It is not an efficiency advantage, it is a continuity advantage. The practice becomes less dependent on single individuals and more capable of handling transitions (exits, arrivals, long holidays, illness, expansions).
The concrete path
Starting this way does not require three months of cold interviews. It requires a different approach: identify a few critical areas where dependency on single individuals is highest and the risk of loss is most concrete, and begin structuring only those. You do not structure everything. It would make no sense. You structure first what would hurt most if lost.
For an accounting practice, typically, it is four areas. The knowledge of historic clients held by the main senior. The reply procedures for recurring queries. The internal escalation rules. The practice writing style. These four areas, well structured, cover most of the continuity risk.
The initial work takes weeks, not months. And it produces a usable asset straight away: a corpus that, even if incomplete, is already consultable and replies to a good part of a new collaborator's questions. Over time it grows, enriches, corrects. It is not a closed project; it is an organism developing alongside the practice.
An asset that accumulates
Structured business knowledge is an asset that grows over time. Every interaction with a client, every decision on a new query, every pattern recognised on the ground adds value to the corpus. After three years, a practice that has started this path has an advantage over an equivalent competitor that has not. Not in the short term, where the difference is marginal. In the medium term, where the difference is cumulative.
Knowledge that stays in heads, by contrast, dissipates. Every departure is a loss not fully recoverable. Every retirement carries away information the new hire will take years to rebuild, imperfectly. The practice does not get dramatically worse, but remains dependent on the fortune of not losing the right people.
The choice between accumulating and dissipating is not a product choice. It is a strategic choice many owners have never put to themselves explicitly. Making it now, with an honest diagnosis conversation about the practice's situation, is worth more than any software purchase.
If you recognise in the practice described at the opening, or some aspect of it, your reality, the first step is not a project. It is a map of where the company's knowledge lives today, and where instead it should start to live outside heads. We do it in a forty-five-minute meeting, with a summary that stays with you in any case.