A buyer does not pay for today's revenue but for the certainty that this revenue will still be there in two years without the current owner at the helm. Every question a buyer asks in the acquisition process comes down to this: what continues if the owner is out for three months. Customer contact, pricing agreements, supplier relationships, the decisions no one else dares to make — if all of these rest with one person, a buyer prices in that risk. Not as a discount on the goodwill, but as a discount on the entire company, because owner dependency works through into the other drivers. Recurring revenue that only runs because the owner calls the customer is not recurring revenue in the sense a buyer counts it; what that distinction precisely entails is set out at what recurring revenue is worth in a sale.
Owner dependency is visible in who picks up the phone when something goes wrong, who is allowed to deviate from the price list, and who is the only one who knows why a process is the way it is. It is also visible in the calendar: how much of the owner's schedule consists of work that can be explained to someone else, and how much consists of work that only he can do because no one else has the context. Companies where that context has been written down, sits in a system, or is carried by a team, show a different dependency than companies where that context exists only in the owner's head.
AI is taking over work, and this does not happen evenly. Part of the work that used to require explanation and experience — putting together a quote, checking a file, giving a first answer to a customer question — can now be done by a system independently. Another part can be done with human oversight: the system proposes, an employee approves or rejects, with reason. And a third part remains human work, because it requires judgment, relationship, or negotiation that cannot be captured in rules.
Those three categories cut right across the question of what in a company hinges on the owner. Work that AI can take over is by definition work that is no longer tied to a person — it sits in a system, and a system is transferable in a way a human is not. Work with oversight shifts the dependency: no longer on the knowledge to perform the task, but on the judgment to assess whether the outcome is good. And work that remains human work keeps the dependency where it already was, unless that work is spread across more than one person.
The difference between companies where this is already visible and companies where it is not yet, rarely lies in the sector. It lies in whether the work was ever written down. A company where processes and systems have been documented can see faster which part of that can be transferred to AI than a company where everything still sits in the heads of a few people; how you make that distinction is set out at what AI does to processes and systems. Where there is no documentation, AI cannot take over anything, simply because there is nothing to transfer — then dependency remains unchanged and high, with or without technology.
The owner dependency index in the value scan looks at eight drivers, and owner dependency cannot be viewed separately from them there. A first indication is obtained by tallying what share of the decisions in an average week is made only by the owner, what share of customer relationships runs outside a system, and what share of the knowledge in the company is not documented anywhere. That last point also touches on what a buyer sees when reviewing the management information: figures that only the owner can interpret weigh differently than a dashboard the team reads itself, as described at what AI does to management information. And because a saving that hinges on one person or one supplier counts differently than a saving embedded in the process, owner dependency also touches on earnings quality; that reasoning is set out at how you substantiate earnings quality.
Improvement starts with making visible what currently rests only with the owner, and then distinguishing which part of that can be documented, transferred to a system, or spread across a team. That last point touches on personnel and succession, and separate statutory requirements apply there for decisions about roles and people — those requirements are not addressed here and are not part of this scan. What can be answered: which part of the work that now rests with the owner is, by its content, of the type a system can take over or that can be done with oversight, and which part remains by definition human work. That is also the question behind team and succession in a broader sense, worked out at what AI does to team and succession.
The underlying question — which work in this company can genuinely be taken over by AI — is answered per task in the work scan of FTE TO AI. For an initial picture of where the pressure on your valuation is greatest today, there is the free value check: eight short questions, one per value driver, resulting in which driver is weighing most heavily on your price today. The full value scan, with maturity scores, evidence per driver and a two-year calendar toward the exit moment, is under development.