A board expresses the ambition: administration must be halved within a year. Everyone nods. And everyone means something different. The finance director thinks of hours in accounts payable. The operations director thinks of reporting burden in the teams. The HR director thinks of less manual work in leave and absence registration. All correct, all incomparable, and at the end of the year no one can determine whether the ambition has been achieved because no one measured the same thing.
The problem is not the ambition. The problem is that 'halving' describes an outcome without saying what the organization needs to be able to do to get there. And that is exactly where it goes wrong, because halving administrative work today usually means something specific: part of that work shifts to AI, another part remains human work with oversight, and a third part remains entirely human work because it requires too much context, exception, or judgment. Without that three-way split, 'halving' is a number without a route.
The ambition 'halve administration' becomes manageable as soon as you break it down. First the vision: why does the organization want this — freeing up capacity for other work, controlling costs, shortening lead time. This determines which administrative processes come first. Next the vision state: what does the organization look like once this has succeeded, in roles and workflows, not in abstract language. Then the target: the concrete, time-bound goal — how much FTE capacity freed up, in which departments, within what timeframe. And finally the target state: the actual setup on the end date, including who still performs which task and who checks which outcome.
These four layers force a management team to translate the word 'halving' into something measurable, well before the year is over.
The shift is not in some future scenario. It is already running through administrative departments today, just unevenly distributed. In some companies, invoice processing, schedule entry, or standard reporting is already largely done by AI, with an employee assessing exceptions and rejecting or approving with reason. In other companies, no one does that yet, with exactly the same software available. The difference is rarely in the technology. It lies in whether the organization has its data in order, whether processes are standardized enough to be handed over, and whether someone has been designated who actually assesses the quality of the AI's work instead of blindly trusting it.
Those three categories — AI takes over the task, AI works with human oversight, or it remains human work — run through every administrative function. Coding invoices can largely be taken over. Assessing a deviating accounts payable claim requires oversight. Resolving a dispute with a supplier remains human work. The ambition to halve is therefore not a single movement, but a sum of hundreds of small decisions about which task falls into which category, and exactly which work is thereby taken over by AI is the question the work scan of FTE TO AI answers per task.
Whether an organization is actually ready to make that shift does not depend on the ambition itself but on eight underlying dimensions: from data quality and process maturity to decision rights, oversight capacity, and the degree to which employees can recognize exceptions. A board that wants to halve without knowing where the organization stands on these dimensions is setting a goal without a route. The question which capabilities does my organization need for this ambition addresses exactly that link: not what you want to achieve, but what must first be organized for it.
It helps here to look at comparable ambitions. The way in which improving margin as AI takes over work requires capabilities largely overlaps with what halving administration requires: standardization, oversight roles, and a decision structure that knows who may reject an AI outcome. The insight into what a target operating model means in plain words is also useful here, because the translation from ambition to roles is precisely that: an operating model, not a wish image.
Halving only becomes measurable once the eight dimensions have been translated back into roles and decision rights. Who assesses an AI outcome before it enters the books? Who is responsible for the quality of the data the system works on? Who decides that a process is not yet ready for handover? These are not technical questions, they are organizational questions, and the answer determines whether the freed-up hours actually become freed-up capacity or resurface elsewhere in the process as delay.
What an employer subsequently does with that freed-up capacity — redeployment, different use, something else — is not something we address. Separate legal requirements apply to that, and it is up to the employer, not a readiness assessment.
Not from a single percentage, but from three things that can be checked side by side: how much FTE capacity has actually been freed up per department, how much of the remaining work under oversight is rejected or approved with reason, and whether the roles carrying that oversight have actually been set up and not implicitly ended up with some random employee. Ambitions that establish this in advance can be tested after a year. Ambitions that do not remain a statement made in the boardroom.
The full ambition test — four layers, eight dimensions, five confidence gates — is under construction. What is already available now is the free readiness check: eight short questions, one per dimension, giving a picture of where your organization is furthest along and where it lags furthest behind. For those who first want to understand how this approach relates to other growth ambitions, growing into a new market as AI takes over work and delivering faster as AI takes over work show that the question behind halving, growing, and accelerating always follows the same pattern: readiness first, then the number.
Vertel wat u wilt bereiken, dan kijken we samen wat daarvoor moet staan.
Answers come from this site’s knowledge base. Not tailored advice, and not a scan of your company.