Securing knowledge before retirement: why the last year is not a year

Let us do the arithmetic.

Under Altersteilzeit, Germany's phased retirement scheme, the block model means working normally through the first half and not at all through the second. Four years of Altersteilzeit therefore means your colleague sits at his desk for the last time two years before the official retirement date. That shows up surprisingly rarely in handover plans. Nobody does the maths.

Then there is leave. People who are leaving take their remaining holiday, and they take it at the end, not in February. Flexitime balances and accumulated overtime come on top, and those are rarely paid out.

The last item appears in no spreadsheet at all. In the final months the mind is already half in retirement, and that is entirely fair: after forty years of work you are allowed to look forward to the end. It just means that anticipation competes with every meeting you still want to schedule. In month eleven you get friendly answers and not much else.

Subtract all of that from twelve months. What is left is your window. How large it turns out depends on the company. Large it never is.

And because the topic invites it: this is not legal advice. What applies in an individual case is set out in the employment contract and the collective agreement.

In the case I watched, nobody asked the manager to document his experience. Just as well, that would have gone wrong.

Instead there were interviews, spread over months. After each session somebody worked up what had been said and put the result back in front of him next time. This is exactly where the thing that matters happens: he read what it had turned into and started correcting. That's not quite right. That belongs the other way round. What's missing here is the actual reason the decision went that way back then.

What came out at the end was a graphical overview of how the market had developed and how its parts related, on a single surface instead of in fifty pages of minutes.

The diagram was not the documentation, it was the tool. People politely read past a text; they do not read past a picture. If an arrow points the wrong way, or the connection between two boxes is missing, an expert sees it in seconds and then says so.

Behind this is something familiar to anyone who has ever tried to get an expert to write things down. Facing a blank page, the man does not know what you are missing, because to him his knowledge is not information but common sense. Put a half-finished version in front of him and he becomes precise. Sometimes also irritated, but precise.

Contradicting is easier than recounting.

One story proves nothing. The situation behind it, however, is well documented.

By 2040, according to the Federal Statistical Office, around 13.3 million people in the German labour force will pass the statutory retirement age of 67, which is 30 percent of the entire labour force. Nobody can move up to fill that: the younger cohorts are simply too small.

For mid-sized engineering companies one aspect of this is particularly awkward. It is not just anybody who leaves, it is the most experienced. Processes are usually documented, the reasons behind them are not. In the end the history of a machine or a customer is known by exactly one person, and that person is leaving. Quality management calls this a single point of knowledge.

Nobody can calculate what that costs you with any credibility, but everybody notices it. Successors take longer, repeat mistakes that were already made and fixed ten years ago, and when in doubt call the retiree while he still picks up. That works remarkably well for a while, and then at some point it does not. It gets genuinely expensive at the place nobody looks: in decisions that turn out worse because the context is missing.

Then there is the formal side. ISO 9001:2015 requires in clause 7.1.6 that an organization determines and maintains the knowledge necessary for its processes. In an audit you should therefore be able to explain what happens when the most experienced people leave. Many cannot.

The method needs no software and no budget. It needs someone who actually does the work between the conversations, and that someone is the bottleneck, not the procedure.

• Start with people, not with topics. Who is leaving in the next three years, and whose knowledge does nobody else hold? An honest list is enough; you do not need a knowledge map for this.
• Establish the last actual day of attendance first. Count backwards from there, not from the retirement date.
• Hold conversations instead of handing out documentation assignments.
• After every round, show what it turned into, ideally as a picture. This is where the precision comes from, not from the conversation itself.
• Start earlier than feels necessary.

What a systematic approach looks like is set out on our topic page knowledge transfer at retirement.

For the record: this case predates our founding, hAiner was not involved. It shaped us nonetheless, for two reasons.

The first concerns the method. Knowledge becomes tangible when people have a worked-up version of it in front of them and are allowed to contradict it. hAiner reads a company's existing knowledge sources, meaning documents, file stores and its own website, answers questions about them with source references, and shows where records contradict each other and where something is missing. That does not replace the conversations with colleagues; those remain manual work. You simply know beforehand what you need to talk about, and what comes out of it can still be found by someone afterwards.

The second reason is more mundane. The questions do not stop on the last working day.

They only get more awkward, because now you are calling someone privately and feeling bad about it. hAiner can therefore also be set up so that former employees remain reachable through a separate, tightly limited external access, if both sides want that. The gain here is less the reachability than the location: the answer ends up in the company's knowledge base and not in the inbox of whoever happened to call.

If you want to know where the single points of knowledge sit in your operation, take the knowledge-loss check.