5 Steps to Manage Turnover and Company Knowledge in SMEs

How knowledge turnover in SMEs hollows out processes

Knowledge turnover in SMEs is the moment when someone leaves and takes with them rules, exceptions and shortcuts that are not written down anywhere. It is not just a headcount replacement: it is a gap in the processes. Those who remain must rebuild from scratch how to write a quote, how to handle a difficult client, how to close an urgent order. Documenting and automating those steps reduces the damage and makes the work repeatable even the day after a resignation.

In an Italian SME, critical roles are few and overlapping. A new colleague, or an automated workflow that is supposed to perform the same work, finds no process sheets, no inventory exceptions, no price lists discussed over the phone. The management system holds records and transactions; people's minds hold the method. When that mind walks out the door, the data remains and the know-how disappears. It is the same mechanism as the invisible costs that erode margins: it does not show up on the balance sheet the day of the resignation, but it becomes visible in the weeks that follow, through errors, delays and endless handover sessions.

Writing down processes before they exist in only one person's head

The first step is to put in writing the steps that today exist only in the memory of those who carry them out every day.

You do not need a library-style manual. You need an honest map: who does what, in what order, with what exceptions, with what documents. In SMEs, tacit know-how is the rule, not the exception. It is learned by shadowing, passed on verbally, adjusted over the phone. It works as long as the same person stays in the role. The day that changes, the handover becomes a treasure hunt through chats, spreadsheets and "the previous person told me so".

Start with the workflows that, if they stop, stop revenue or clients: quotes, order confirmations, invoicing, complaints, reorders. For each one, record inputs, outputs, timelines, who decides, and what happens when a piece of data is missing. That sheet is already knowledge management, even if it lives on a shared page rather than a sophisticated platform.

For those who have little time to document their daily routine

Anyone juggling production, clients and suppliers tends to postpone writing down processes because "we already know how to do it." The postponement is understandable and costly. One hour a week on a single workflow, always the same one until it is clear, is worth more than a documentation project left half-finished. The goal is not to describe the entire company: it is to remove from a single person's head the piece that, if missing, causes visible delays.

Breaking up single-person roles that hold know-how hostage

The second step is to avoid leaving a critical process dependent on one person alone, with no backup and no instructions.

In large organisations, a departure is absorbed because roles are redundant and systems are already full of rules. In an SME, turnover hits harder precisely because that redundancy does not exist. The person who manages orders also knows the difficult clients, the outdated price lists, the slow suppliers. The person who writes quotes knows which extras can be granted and which cannot. That concentration is efficient while it lasts; it becomes fragile the day of the resignation.

A process that lives only in one person's memory is not a process: it is an operational risk that materialises the day they resign.

For those who concentrate quotes, orders and clients in a single person

If quotes, commercial exceptions and complaints all pass through a single workstation, the replacement is not a human resources issue: it is a continuity issue. A brief shadowing period, double verification over two weeks, shared access to pricing rules. This is not bureaucratisation. It is ensuring that the following Monday the work continues without having to rebuild the entire craft from scratch.

Turning invisible costs into a knowledge inventory

The third step is to list where undocumented knowledge is already causing losses in time, margin and quality, before purchasing any new tools.

The costs are not just the salary of the incoming person. They are the weeks when orders go out slowly, quotes are copied incorrectly, suppliers are called twice, and quality control repeats mistakes already seen before. What is usually missing is a simple inventory: which rule lives in a person, which in a spreadsheet, which in the management system, which nowhere at all. Without that list, discussions focus on "training" while ignoring the fact that the method has never been made accessible.

• Pricing rules and discounts that exist only verbally or in a chat.
• Inventory and quality exceptions that no one has entered into the management system.
• Quote templates copied by hand, with fields that get skipped every time.
• Handovers done from memory, with no list of access credentials and deadlines.
• Clients and suppliers known as "special" only to one person.
• Final checks that depend on the eye of the person who is leaving.
• Contractual deadlines kept in a personal calendar.

That list becomes the work queue. Not everything needs to be automated in the first month. It needs to be made visible. Then you decide what deserves a repeatable workflow and what just needs a clear reference sheet. Anyone evaluating where to allocate budget will find here the same criterion used for the budget mistakes that block automation: fund what removes a real bottleneck, not a generic tool.

Having steps executed by a workflow, not by memory

The fourth step is to turn tacit know-how into steps that a colleague, or an automated system, can execute without asking the person who has left.

Documenting is not enough if the document sits in a folder no one opens. The method must live where the work happens: in the management system, in the order form, in the quote checklist, in the message that fires when a piece of data is missing. Artificial intelligence automation, in this context, does not "know" things on behalf of people. It captures rules you already have, reapplies them, and flags exceptions. It increases the likelihood that the work stays consistent even with a different team.

Two concrete cases point in this direction. At ERP Costruzioni, in the construction sector, drafting a quote went from 8 hours to 5 clicks in 30 days. At About medically s.r.l., in the parapharmaceutical sector, document generation speed increased by 85% in 90 days.

Automation does not replace the person who leaves: it makes repeatable what that person knew how to do, even the following Monday.

A workshop with few product lines and a great deal of oral knowledge

A workshop or craftsperson selling ceramics, components or kits and managing a limited number of product lines accumulates exceptions: batches, finishes, firing or processing times, clients who always want the same variant. If those exceptions live in a notebook or in one person's head, a replacement wipes out months of fine-tuning. A reference sheet per product line, linked to the order, is enough to prevent the next person from starting from zero. This is not an IT project: it is respect for the work already done.

Starting with a visible process without an IT team

The fifth step is to choose a single painful workflow, describe it, run it in a guided way, and only then expand, without hiring a dedicated IT manager.

Many Italian SMEs do not have an internal systems department and do not need to build one in order to secure their knowledge. What is needed is a partner who translates the method into a concrete workflow, using language that is close to those working on the shop floor, in administration and in sales. Process automation without an internal IT manager works when the scope is narrow and results can be judged quickly in the field, not after a complete overhaul of the management system.

1. Phase 1 (Selection): one workflow that, if it stops, is felt by the client or the margin.
2. Phase 2 (Mapping): who, what, when, exceptions, documents, systems already in use.
3. Phase 3 (Rules): what is fixed, what needs a person to confirm.
4. Phase 4 (Execution): forms, checklists or automations that guide the work.
5. Phase 5 (Testing): two weeks running in parallel, errors noted, rules corrected.
6. Phase 6 (Expansion): only then move on to the next workflow.

Resist the temptation to "put artificial intelligence on everything." An assistant that summarises emails does not preserve warehouse know-how. A workflow that fills in the quote using the actual rules does. That is the difference: company knowledge is retained when the system requires it to be used, not when a document exists that no one consults.

How Leomat makes processes independent of individuals

Leomat helps Italian SMEs transform the method that today lives in people into tailored, measurable workflows, without over-engineering.

Leomat was created to bring frontier technology to businesses: it builds the possibilities of artificial intelligence with method, rather than simply promising them. The work focuses on automations built around real processes, not generic tools forced into shape. Among the services offered is the development of ERP systems tailored to the company: the point is not to add more screens, but to ensure that quotes, orders and documents follow the rules even when the person at the desk changes.

The ERP Costruzioni case, from 8 hours to 5 clicks on quote drafting in 30 days, shows what happens when a manual task moves into a tailored workflow. About medically s.r.l. achieved a 85% increase in document generation speed in 90 days. The method serves this purpose: operational concreteness, no processes inflated beyond what is needed.

How do you document processes in an SME without stopping the work?

You choose a single workflow that touches the client or the margin and describe it while executing it, not in a separate room. Inputs, outputs, exceptions and who decides are enough. One hour a week on the same step, until a colleague can repeat it without asking. Then you decide whether that sheet stays a guide or becomes an automation. Trying to document everything at once is the fastest way to finish nothing.

What happens to company knowledge when someone leaves?

If the rules are not in the process, they leave with that person: exceptions, special clients, ways of closing an urgent order. The incoming person rebuilds through trial and error, with mistakes and delays. The management system remains full of data and empty of method. The damage shows up over weeks, not on the day of the resignation. Putting the rules where the work happens reduces that gap without waiting for a perfect team.

Does artificial intelligence preserve know-how or replace it?

It preserves what you have already made explicit: price lists, steps, checks, templates. It does not invent the craft on your behalf and does not replace judgement on genuine exceptions. Applied to the right workflows, it makes daily work repeatable and reduces dependence on a single person's memory. Used as generic conversation, it protects nothing. The difference lies in the rules you decide to have the system execute.

Do you need an IT manager to start with automation?

No, if the scope is a visible workflow and the partner translates the method into steps usable by those already working in the company. What is needed is clarity on who does what, not an internal systems department. Many SMEs start this way: one painful process, a short trial, rules corrected in the field. Hiring a dedicated technical profile makes sense later, if the number of workflows to manage together becomes large.

This post was created with AI