Retirement and resignation: how businesses keep knowledge from being lost

Businesses secure knowledge most effectively before retirement and resignation by recording it in documents in good time and making it centrally searchable by question – not only in the last weeks of handover. Classic measures such as handover documentation and mentoring help, but they capture only a fraction of the experience of long-serving employees. The safeguard only becomes truly sustainable when recording knowledge is part of the ongoing routine and what has been collected can also be found later.

In brief

• The departure of long-serving employees costs businesses customer history, installation knowledge and experience all at once.
• Handover documents and mentoring capture only a small part of the knowledge and usually come too late.
• Knowledge is only secured once it is documented and can be found again by asking.
• Voice notes lower the hurdle: talking comes more easily than writing.
• A pragmatic 5-step plan makes securing knowledge achievable without a major project.

Why is the departure of the ‘man who knows everything’ so dangerous?

In almost every long-established business there is this one person: the workshop manager who knows every installation from the last thirty years, the office manager who knows the background of every customer. As long as this person is there, everything works – which is exactly why it rarely becomes apparent how much business-critical knowledge sits exclusively in one single head.

The departure rarely comes as a surprise, but almost always unprepared. A retirement is announced years in advance but gets pushed aside in day-to-day business; a resignation often leaves only a few weeks for handover. In both cases, securing knowledge usually only begins when most of it can no longer be saved.

The consequences show up gradually and over years. The successor suddenly needs hours for tasks that used to take minutes; problems that had been solved are solved a second time at great cost; customers sense that ‘the new person’ does not know their history. This damage can rarely be put into figures – but it is noticeable in almost every business that has experienced it.

Which types of knowledge are at risk when someone leaves?

Not all knowledge is equally at risk: whatever is already in contracts and manuals survives any change of staff. Three types of knowledge that typically exist only in people’s heads are critical.

Type of knowledge	Typical examples	Why it is rarely documented
Customer history	Background, special agreements, peculiarities, sensitive issues	Scattered across emails, phone calls and memory
Installation and site knowledge	Conversions, temporary fixes, ‘special features’ on site	Never recorded during installation – ‘everyone knows that’
Experience	Fault diagnoses, supplier assessments, proven procedures	Taken for granted and hard to describe

Customer history is often both the most valuable and the most fleeting. The fact that a regular customer almost left years ago because of a complaint and has been given special attention ever since is not in any CRM field – if this knowledge is lost, the successor comes across as clumsy without knowing why.

Installation and site knowledge weighs most heavily in technical businesses. The fitter who knows that, in the installation at Hotel Sonnenhof, the valve is fitted in the basement and not where the plan shows it saves the business search time on every job – until he leaves.

Why are handover documentation and mentoring not enough?

Classic handover documentation – a document that the departing employee fills in during the last few weeks – fails because of a fundamental problem: nobody can write down thirty years of experience in a fortnight. What gets recorded is what happens to occur to the person, not what the successor will need in three years’ time.

On top of that comes the retrieval problem: you do not know what you know until someone asks. The crucial details only surface when the specific situation arises – and by then the person who held the knowledge can no longer be reached. Handover documents therefore reliably answer the wrong questions.

Mentoring and overlapping staffing are more effective, because knowledge is passed on in context, but they have their own bottleneck: they need months of overlap, which rarely exists with resignations, and they tie two paid workers to the same tasks. In addition, the knowledge once again ends up in just one single head – the next departure poses the same problem again.

Both measures remain useful, but they do not solve the structural problem: knowledge that exists only verbally or in a single document is not secured. It is only secured once it remains findable independently of individuals.

How is knowledge captured in documents and made searchable?

The sustainable approach has two halves: recording knowledge in documents on an ongoing basis, and opening up these documents so that they can be found again by asking. The first half is easier than expected, because much of the knowledge is already in documents – in emails, quotes, service reports and minutes, just scattered and impossible to find.

For knowledge that has not yet been recorded, low-threshold formats have proven effective. Voice notes are the underestimated tool here: after a job, the long-serving fitter dictates into his smartphone for two minutes what is special about this installation – people find talking much easier than writing documentation. This is transcribed and filed as a searchable document.

The second half – finding things again – is handled by an AI-supported knowledge base: a system that captures all documents centrally, makes even scans and photos readable with text recognition (OCR, the automatic conversion of images into searchable text) and answers questions in plain language, always with a source pointing to the original document. The successor asks ‘What do I need to bear in mind with the installation at Hotel Sonnenhof?’ and gets the answer from the service report from back then – the guide Using ChatGPT with your own data explains in detail how this technology works.

To be honest, it has to be said: even the best system only secures what has been captured. The technology does not replace the decision to take securing knowledge seriously – it ‘only’ ensures that the effort pays off, because what has been recorded is actually found again.

What does a pragmatic 5-step plan look like?

Securing knowledge does not have to be a project with a consultant and a concept phase. For small and medium-sized businesses, a pragmatic approach in five steps has proven effective:

1. Name the knowledge holders and risks: List which people hold which critical knowledge exclusively and how foreseeable their departure is. One hour with the management team is enough for an initial overview.
2. Bring existing documents together: Collect what already exists – service reports, customer correspondence, installation documents – in one central, searchable place instead of leaving it scattered across drives and inboxes.
3. Introduce capture routines: Anchor low-threshold habits in everyday work – the voice note after the customer appointment, the photo of the rating plate, the two-liner on the special agreement. Short and regular beats detailed and never.
4. Capture knowledge retrospectively in a targeted way: Sit down with the most important knowledge holders for short, thematically focused interviews – one hour a week on one group of customers or installations at a time, recorded and transcribed.
5. Make it searchable and test it: Make the collection searchable with AI and use real everyday questions to check whether the answers and their sources are correct. The introduction process of Goma-it’s AI knowledge base shows by way of example how such an introduction runs in a structured way – from the initial conversation through the initial filling to test questions.

The order is crucial: first collect what is there, then build routines, and finally capture retrospectively in a targeted way. Anyone who starts the other way round with weeks of interview marathons overwhelms those involved and usually gives up before the benefit becomes visible.

What changes in the business when knowledge can be searched by question?

The most immediate effect shows in induction: new employees can ask questions instead of waiting until someone has time for them. The system quickly answers the question about a customer’s background or the peculiarities of an installation – including the original document for reading up.

The burden on the remaining workforce also falls. The experienced office manager is no longer interrupted twenty times a day because she is the only one who knows something – the knowledge base answers routine questions, and her time remains free for the cases that really need experience.

In the long term, the business’s risk situation changes. A departure remains a loss – of skill, of relationships, of personality – but it is no longer a question of survival, because the documented knowledge stays in the company. A business that depends on individual heads becomes one that treats its knowledge as shared property.