In many companies, the best customer contact runs through one person: the owner, the account manager who has been there for fifteen years, the one who knows in what tone which client wants to be approached. That is not an accident. It is the outcome of years in which personal attention was the shortest route to a satisfied customer, and in which no one took the trouble to document exactly what that person did. The knowledge sits in a head, not in a system.
As long as the company is running, that is not a problem. The moment a buyer looks at it, it becomes a question: what happens to the revenue from these customers if this person leaves. A buyer does not count the number of employees in customer contact, but what still continues without one specific employee. That is a different kind of count, and the outcome is often lower than the owner had expected.
Part of customer contact consists of work that can be taken over: drafting a quote based on earlier quotes, summarizing a phone call, looking up a customer's history before someone calls, formulating a response to a standard question. AI can already do that work today, in parts, with a human approving or rejecting with reason before it reaches the customer.
Another part remains human work, and not because the technology is not ready for it yet: assessing whether a customer does not say something but does mean it, deciding when a correct response is the wrong response, maintaining the tone that suits that one particular customer. That is precisely the part for which that one employee feels irreplaceable.
The difference between companies does not lie in the sector or the size, but in how the contact is organized. Where conversations, appointments and preferences are kept in a system that everyone can consult, the transferable part has already been made visible and AI can pick it up without the customer noticing anything. Where that information only sits in a head, there is nothing to transfer — not even to AI. The backlog then lies not in the technology but in the documentation that never existed.
A buyer who sees customer contact as dependent on one person calculates with a downward risk adjustment or asks for a longer handover period with the departing owner. Both lower the price or the terms. More importantly: the risk cannot be resolved in a single conversation. A buyer does not ask for a promise that it will work out, but for proof that the contact is already recorded somewhere other than in the owner's head.
This is not a question about who works and who does not. It is a question about where the knowledge sits. A company can keep exactly the same number of people employed and still become more transferable, simply by documenting what now only exists verbally.
The customer notices nothing of a well-executed handover. What changes is where the information is stored and who has access to it. A conversation is summarized and logged in a system instead of in one person's memory. A customer's preference is recorded at the moment it is expressed, not reconstructed from memory three years later. AI can do a large part of that recording and summarizing, while the employee keeps conducting the conversation as they always did.
The order in which this happens is not without consequence. First make visible what happens now and by whom, only then determine which part can be made transferable — to a system, to a colleague, to AI with oversight. Whoever reverses that order and starts taking over immediately risks losing precisely the part that kept the customer coming back.
If the topic touches on who retains which role or whose position changes, separate statutory requirements apply to that; that is not a question answered by an AI setup.
Customer contact does not stand alone. A quote that is ready faster affects the calculation that hinges on one calculator, and a schedule that responds to customer preferences affects the scheduling that can be made transferable with AI support. Whoever views customer contact as a separate value driver instead of as part of the broader customer relationships a buyer assesses often misses where the real dependency lies. And to see whether this is indeed the driver that pushes down the price the most, or whether the answer lies elsewhere, there is the question of what depresses the value of a company.
Which part of customer contact in this specific company can truly be taken over by AI, and which part remains human work, differs per task and is mapped out step by step with the work scan from FTE TO AI.
The free value check consists of eight short questions, one per value driver, and provides a picture of which driver depresses this company's price the most today. Customer contact may be that driver, but that only becomes clear after the questions. The full value scan — with a maturity score per driver, the owner-dependency index and a two-year calendar toward the exit moment — is under construction.