A buyer does not look at the organigram but at the process itself: does the order-to-invoice flow run through without a specific person performing an action that is nowhere recorded. Processes and systems are the value driver that expresses how predictable, repeatable and transferable the operation is, independent of who happens to be present on a given day. A buyer pays for repeatability, not for improvisation, however well that improvisation turns out.
The question underlying this is not new. What is new is that AI already carries out part of that work today, and that changes what 'documented process' means. A process that runs on a system that itself handles tasks is less dependent on the discipline of employees to follow a procedure. That shifts the weighing: a process with AI in it can in fact be more stable than a process that only exists on paper.
A buyer does not see this in a policy document but in traces. Does a process keep running if the key figure is absent for two weeks. Are work instructions kept somewhere other than in the head of one person. Is there a system that flags deviations without someone having to manually leaf through them. And, specific to this moment: which part of the process is already carried out by software or AI, which part runs through human control with a reason for approval or rejection, and which part is and remains human work because it requires judgment that cannot be formalized.
That threefold division — taken over, under supervision, remaining human work — is precisely what a buyer reads differently today than five years ago. A process that relies entirely on the third category weighs more heavily in risk than a process in which most of the repeatable steps have already been moved to the first two.
One company has already set this up, another has not, and that difference rarely stems from ambition. It stems from how the work is structured. A process that consists of standardized input and clear decision rules is easier to transfer to a system than a process that relies on negotiation, exceptions and customer contact. Companies with many repetitive administrative steps see that AI already takes over those steps today; companies where every case is different keep a larger part with people, with supervision as an intermediate form.
That is not a shortcoming. It is a given about the type of work. What a buyer wants to know is whether that given has been deliberately established — has it been determined per process what is transferable and what is not — or whether it simply grew that way by chance.
Measuring starts with the process itself, not with the organization next to it. For each core process: is it recorded outside the head of the person carrying it out, is there a system that carries or supports the process, and which part of the steps is already handled by AI or software without anyone having to look at it each time. Next: which part runs under supervision — an employee who approves or rejects with a reason, and records that reason somewhere — and which part is designated as remaining human work, with an explanation why.
Evidence for a buyer consists of traces, not intentions: process documentation that has been recently updated, system logs that show that a task was handled automatically, deviation records that have been followed up. A statement that 'everyone here knows how it's done' is not evidence, it is a risk that has not yet been written down.
The question of which work in this specific company can genuinely be taken over by AI is answered per task in the work scan of FTE TO AI, independent of what you subsequently do with that outcome.
Improving starts with recording what already happens before anything is added. A process that no one knows exactly how it runs cannot be assessed on what AI takes over from it. Then follows the classification into the three categories per process step, with a reason attached: why this part is automatic, why that part is under supervision, why another part remains with a person.
This driver is connected to other drivers that a buyer weighs separately. How much of the company runs through the owner personally is addressed in how you measure how many decisions and contacts still run through the owner, and whether a saving on these processes holds up with a buyer depends on what is established in the analysis of profit quality under AI about the origin and repeatability of margin. Process documentation is also not optional from a legal standpoint: which contracts and records belong with it is described in the test of legal hygiene in a company where AI works alongside you. If a process change touches on personnel decisions, separate statutory requirements apply to that, on which this page makes no statement.
When this kind of improvement needs to take place before a sale moment depends on how much time there is; that is worked out in the timeline for preparing for a sale moment.
The value scan puts eight drivers side by side, gives each driver a maturity score with evidence, calculates the owner-dependency index and sets out a two-year calendar toward the moment of sale. That full scan is under construction. Ahead of that, there is the free value check: eight short questions, one per driver, resulting in a picture of which driver is putting the most pressure on your price today.