What Remains When Minds Depart?

When employees leave a team or company, they leave behind more than just a vacancy to be filled: they take with them much of the knowledge they have accumulated and created over the years. Depending on which statistics you consult, the estimated cost to rebuild what is lost ranges between one and five annual salaries – in addition, of course, to the onboarding effort for the new hire.

One of many knowledge classifications distinguishes between three types of knowledge: The most obvious of them, explicit knowledge, is that kind of knowledge that can be easily documented in process manuals, work instructions, or project reports. These are usually well-maintained enough to fill parts of the technical knowledge gap left by a departing employee and to train a successor. If documentation of explicit knowledge is incomplete, it might be supplemented by those who remain. While people seldom enjoy this task, it is at least feasible.

Documentation is significantly more challenging with implicit knowledge that is used when generalized explicit knowledge is applied to specific situations. It arises from years of experience, merging technical expertise with the specifics of a company, its projects, and its clients. For example, sales processes are easy to document, but they can only be reproduced successfully when combined with information about the human chemistry between the involved parties. This type of knowledge grows unconsciously and over long periods of time.

A third form of knowledge is intuitive, interpersonal, socio-political relationship knowledge. How does a client “tick”? Which remarks are helpful in which kinds of meeting, and which should be avoided when certain individuals are present? This type of knowledge is fundamentally impossible to formalise and therefore impossible to directly document. It contains judgment-based decision-making and intuitive problem-solving approaches that have developed over a long time, often in close (and rarely conscious) exchange with colleagues or client-side teams. The loss of this knowledge is what hurts most when colleagues leave. Suddenly, teamwork gears start to grind, and no one quite knows what’s happening.

Explicit knowledge can be conveyed through documentation; implicit knowledge might be gleaned from meeting minutes, debriefings, or project reports. Relationship knowledge, however, is accessed so automatically and unconsciously that direct documentation, let alone transfer, seems impossible.

Bernd Waterkamp: Opening the Treasure Chest

I previously interviewed Bernd for the article Profit-Center Aftersales in the 04/24 issue of the magazine technische kommunikation. For Bernd, aftersales is a complex system that is deeply ingrained in a company. Therefore, cultivation and preservation of knowledge are particularly relevant to him.

Mathias Maul: You’ve managed large technical departments. Did you ever experience negative consequences after a departure or staff change because knowledge was not preserved?

Bernd Waterkamp: Of course. That’s the case in many companies. Sometimes efforts are made to regain lost knowledge, but mostly it’s just tolerated, accepted. There’s no outcry.

Mathias: Why is that? Real value is being lost.

Bernd: Most people are simply not aware that knowledge walks out the door with the employees. Turnover is a normal occurrence, as are layoffs and poaching. The problems arise due to the lack of processes for documenting this knowledge during the work cycle.

Mathias: Is there a typical part of the work cycle where this knowledge is generated?

Bernd: Knowledge is often shared informally during breaks, coffee chats, meetings. Or during business trips where several colleagues would sit in a car for hours discussing many different topics. This type of communication has become lost in today’s organisation of digital meetings unless the organisation consciously emphasises it. Many companies cannot successfully replicate these important informal aspects in the digital space. Online meetings are often only about getting through the agenda; important topics that happen “between the lines” are not discussed.

Mathias: So it’s not just about technical expertise, but also about some kind of in-between knowledge? What is its importance?

Bernd: Everyone has their treasure chest, their secret sauce. Some are reluctant to share those however. I experienced this back in the 80s during my apprenticeship … the old master craftsmen had special tools in their toolboxes that they had made themselves. These were sacred, and got locked up after work. They spent years building this. That’s similar to knowledge “between the lines.”

Mathias: And this is presumably not something people consciously think about?

Bernd: Yes, it’s mostly unconscious. Once you had gained the masters’ trust, they were proud to tell you and teach you. It’s not much different in current times.

Mathias: Let’s say that Fritz is about to retire. When I knock on his office door and say,”hey Fritz, tell me what you know,” he would probably just stare at me, puzzled. How can you motivate people to share their knowledge?

Bernd: I think Fritz would participate if he were proud of his accomplishments. People share knowledge when they sense a genuine interest in their work, not when someone just wants to extract their knowledge. Experienced colleagues really want to pass on knowledge so that it is used in a way that aligns with their own thinking. If you just go and say, “Fritz, you’re retiring soon, HR sent me to enter your secret knowledge into a computer,” you certainly won’t get what you need. You have to prepare, have a guide, and develop a feel for which information is valuable. Remember that we are talking about the kind of knowledge that isn’t written down anywhere.

Mathias: Does the company culture also play a role in whether people participate?

Bernd: Absolutely. In the past, knowledge loss might not have been as noticeable because people stayed with the company for a long time. Today, turnover is much higher, and hidden know-how gets lost. And if companies only use remote work for ticking off tasks and neglect informal communication, no new “between the lines” knowledge is generated.

Mathias: When does this loss become tangible?

Bernd: People in operations often feel it very quickly when trusted individuals or points of contact are no longer there. It’s an emotional loss. Those who remain may not talk about it, but they accept it. You notice it when processes are no longer running smoothly because you can’t just call Fritz, who always had an answer.

Mathias: How could this implicit knowledge be secured?

Bernd: There’s no one-size-fits-all solution. What always helps however is to emphasise the value of human-to-human communication. Not just see it as a side issue, or as a necessary evil to achieve some goal.

Mathias: Communication as a side issue? Sounds like it’s an uncomfortable topic.

Bernd: It’s often seen that way. Especially for technical people, communication can be a sensitive topic, and few companies really invest into enabling communicative skills. But communication happens anyway when people are working together. So you have to appreciate it.

Mathias: How do you select whose knowledge you want to preserve?

Bernd: One method is to just watch which groups form to have lunch or coffee together. In these groups, there is often one person whom others enjoy listening to. Someone who is not a grump but open, friendly, and polite. You know, one of those people whom you can approach without fear and ask even question that you think of being stupid. So look for the people who like to share knowledge. Those who are respectful and don’t make the questioner feel ridiculous. You can get expert knowledge from everyone, but those people are the ones with the secret sauce.

Mathias: Those are the ones who will be motivated to have their knowledge digitized?

Bernd: There’s a rule of thumb: Someone who cannot do small talk will probably not openly reveal his knowledge. You need to ask yourself: Does this person really have the knowledge that will help the team develop? Often the answer is no. In this case, I may not need to chase after this person’s knowledge, because it’s purely technical and it can be acquired elsewhere.

Mathias: You’ve been a technical writer before. Which role do tech writers play in the knowledge preservation space?

Bernd: A very important role, of course! Tech writers have highly developed skills to collect and process information. They are also used to explain to people the value of their contributions, and do not take “no” as an answer. Being highly structured, curious and persistent in a friendly way are aspects of good technical writers, so they are perfectly suited to this kind of work.

Limits of Automation

Why, however, bother with persuasion when there is a lot of software that, sometimes fully automatically, extracts and processes information from emails, Slack threads, meeting recordings, and other sources?

In many countries (such as the one I’m writing this article in), the mere mention of such programs would probably be enough to receive raised eyebrows at least, or being kicked out of the room by the works council at most – and, I must say, rightly so. Although emails and other business communication “belong” to the company, direct intervention in communication processes is highly intrusive – just imagine the potential effects on one’s communication behaviour if one knew that one’s messages were being analysed.

These caveats apply differently in situations whe information is made available on purpose: meeting minutes, technical reports, project plans, summaries of sales calls in a CRM and the like. Howeber, these information sources contain mostly explicit knowledge. For other knowledge types, a different kind of extraction and processing of knowledge is required. This is where the recent developments in machine learning and artificial intelligence come in handy.

Eric Wei: Electronic Mentors

Eric is a Tokyo-based entrepreneur who was, among other things, responsible for the launch of Amazon’s Kindle products in Japan and led foodpanda, the Japanese branch of the German delivery service Delivery Hero. As founder of the skill matching platform timelyhero, Eric and his team has developed a system to create AI-based “digital knowledge twins” of employees and executives.

Eric Wei: In Japan today, employees often take only one month for a handover when they leave a company. But this is far too short for someone who has worked there for years to pass on their knowledge. This is a risk for companies, because if they cannot keep the best knowledge, it is lost. So there is a big need to secure this knowledge.

Mathias: Please explain – what are you doing to secure it?

Eric: Let’s say you have a sales team with A, B, and C players. For B and C players, it is very difficult to get access to the knowledge of the A players, because the As are busy selling, closing deals faster, and simply often do not have enough time to be mentors for their colleagues. So we extract the knowledge from the As. We find, so to speak, the A-to-Z guide or the secret recipes with which they answer difficult questions or close deals.

Mathias: Can you be more specific? How do you get to the knowledge?

Eric: We conduct interviews with them and ask them about 100 questions about many situations – from lead generation to specific customer inquiries. The question list varies depending on the industry. When we notice that some questions open up further questions, we keep asking. It’s an iterative approach to gradually get into people’s minds and ways of thinking: “What made you do it that way? Why do you think that’s right?” and so on. This is how we branch into areas that neither we nor the employees thought of before.

Mathias: So … you’re picking brains?

Eric: In a way, yes. We start with a structured list of questions, but of course, we cannot know the exact structure of the employee’s knowledge. So we cannot know where we will end up at the beginning of the interview. The AI then helps to recognise patterns. The longer the interview, the more knowledge we can extract. This then becomes a company asset that is not lost, even in case that an employee is poached by a competitor.

Mathias: This does remind me of my work in executive coaching. It helps to go deep. But this requires a lot of trust, so what about the willingness of employees to cooperate when you do this in a company? I can imagine that some might be hesitant.

Eric: We argue that it is in the interest of all employees and of the company if everyone performs well, and that our approach will save top performers’ valuable time. Some companies then create incentives, for example, the interviewed A-class salespeople get a bonus if their colleagues perform better through digital mentoring. We treat them like knowledge influencers, and they know that the more accessible their knowledge is, the more they benefit. The entire team becomes more successful if the top executive opens up to having parts of his valuable knowledge digitised.

Mathias: So, time for money? A-class performers give fewer workshops or mentoring sessions, have more time for their actual work, and are rewarded?

Eric: Exactly!

Mathias: How long does the whole process take?

Eric: About a month, but of course, this depends on the team size. We take breaks in between the interviews, because people continue to think in the background. We also feed back replies of the AI to the people whom we’ve interviewed to make sure that the knowledge was encoded correctly. If possible, we also don’t just interview one person, but look for a suitable sample size. This helps to reduce the bias.

Mathias: Does that mean you are not just replicating one person, but team knowledge?

Eric: If possible, we work with clusters of employees. If there are a hundred employees in the sales team, there might be ten A-class performers. The AI analyses these ten inputs and develops the best approach from them, find the biases that made them successful … their “secret sauce.”

Mathias: Which LLMs and languages do you use?

Eric: At the moment we are using various LLMs, choosing the best fit as they evolve. We work only in Japanese right now, but as soon as the knowledge is trained, it can be applied to many languages. In the end, we arrive at a true single source of truth. And every three to six months, we refresh the system with new interviews, depending on how quickly the company, the market, or its customers change.

Miracles

The process for uncovering implicit knowledge might seem chaotic, but how else would you find out what you don’t know when you don’t know what you don’t know? As mentioned in the interview, there are parallels to coaching and psychotherapy (being experienced in both, I can assure you that there’s not much of a difference). In some methods, it is helpful to inquire about precisely this implicit and relationship knowledge in order to arrive at a suitable intervention. Over time, various questioning models have been developed that make this possible.

• Circular questions from systemic models capture relationships and connections by asking about the perspectives of others, for example, “What would your colleague X think about that?”
• Socratic dialogues bring implicit knowledge to light by leading the interviewee to their own insight – maieutics (roughly translated as “knowledge birth”) – through systematic questioning.
• The NLP Meta Model helps to structurally question concrete behaviours and inner convictions, and makes hidden thought structures visible.
• Solution-focused questioning techniques such as the Miracle Question (Steve de Shazer) leave problems aside and focus on solutions – and on ways to get there.
• Or simply ask ”Whyyyy?” repeatedly until you’ve squeezed out the last bit of knowledge or – a clear indicator of a sufficient number of questions – that the interviewee has had enough.

Often, it is the mix of different models that works. Because our brains are excellent pattern recognisers and (to put it very simply) get bored when patterns repeat. Surprises in questioning techniques help to make the inner search for answers to previously unheard questions varied. This increases the playful aspect of the interview, which in turn increases the motivation to stay on the ball.

Those who do not want or cannot hire consultants such as Eric, Bernd or myself can of course try rolling their own digital twins. All that is needed is some time, a willingness to experiment, a playful spirit, and existing tools such as Claude, Mistral or Gemini. For those with high demands on data security, on-site models from the Qwen or Gemma families can be a workable solutions as well. But even if results of your home-made knowledge bots are not as reliable as you might have hoped for: you’ve had many conversations with your employees, and that alone is a value in itself.

Invisible Values

Maybe this is even the most important value of them all. Both interview partners agreed that without appraising the value of knowledge in the company culture, all initiatives at conserving it will fail. This sounds banal, but in practice, it is anything but established knowledge. To keep parts of people’s value when they leave, first be clear on the value they have created, then use technology as a sidekick to replicate this. Not to replace their successors, but as a vehicle to help them people get even better at their jobs and lifting everyone up. ‌
‌
‌
Interviews were edited for length and clarity.

Bernd Waterkamp is a former VP of Aftersales in the mobility industry. At BWeitblick Consulting, he consults with organizations to improve service, quality and aftersales processes, raise productivity and improve communication at the workplace.

Eric Wei, Founder and CEO of Dimes, Inc., runs the skill matching platform timelyhero to give people around the world the opportunity to teach what they’re good at and learn new skills. In his former career, he held top positions at Amazon, Google, DiDi and foodpanda.