A strategy does not become outdated all at once. It ages assumption by assumption, usually without anyone marking the moment. The plan itself does not change; the world beneath it does. That is exactly why management boards fail to notice it for a long time: there is no document that says "this is no longer correct," there is only a growing gap between what is written and what is actually happening.
Every strategy rests on a series of unspoken assumptions: about what competitors are capable of, about what regulation allows, about what the team itself can and cannot handle. As long as those assumptions remain unnamed, no one can check them. The first sign that a plan rests on outdated figures is therefore not a disappointing result. It is the absence of a list stating when each assumption was last tested. Whoever cannot point to that also does not know how old the foundation under the plan is. Who guards the assumptions when no one owns them describes why this ownership is in practice often assigned to no one.
The reason this is more urgent now than five years ago is not that strategies are being made worse. It is that one specific assumption becomes invalid faster than before: the assumption about who does which work. AI is taking over parts of the work, sometimes completely, sometimes with human oversight that approves or rejects with reason, and sometimes not at all because the work is not suited for it. That is already happening today, per task, per department, differently per company — not as an announcement but as a creeping practice. A plan written a year ago based on available capacity, lead times, or cost prices per task can therefore quietly come to rest on figures that no longer hold, without there having been any occasion to revise it. Why an annual plan does not keep up with the AI shift shows why the annual planning cycle structurally lags behind this shift.
The difference between companies that already track this and companies that do not rarely lies in the sector. It lies in whether there is someone who treats the assumption list as a separate object, apart from the strategy document, and who goes through that list at a fixed frequency. Companies where that happens see a shift in the work within weeks. Companies where the assumptions are only implicit in the plan see it only once the result already deviates — and by then the question of why can no longer easily be traced back to a single cause.
The method central here places three things side by side: an outside view (sector data, regulatory clocks, public signals about what competitors are doing), a self-assessment by the management team itself, and a list of assumptions with a status per assumption: still holds, is uncertain, or has lapsed. The value lies not in a prediction but in the comparison: where does the outside view diverge from the view the management team has of itself, and on which assumption does that difference sit.
That also means what the method does not deliver. It does not say with certainty when an assumption lapses — some outside signals are themselves coarse, published with delay, or based on a sample that does not map one-to-one onto the company itself. An assumption marked as "uncertain" is not a prediction of lapse; it is an indication that it has not been confirmed recently and therefore deserves attention. And an outcome where inside and outside happen to agree says nothing if the underlying assumption is not formulated sharply enough to be able to fail. How do you reduce an assumption to something you can measure addresses that step: an assumption that is not formulated in measurable terms cannot be meaningfully tested by the stress test.
Another scenario the method explicitly deals with: two assumptions that contradict each other, for example an assumption about growth in capacity alongside an assumption about a decline in tasks due to automation. Both may seem plausible separately and yet cannot be true at the same time. What if two assumptions contradict each other describes how that conflict is made visible instead of quietly disappearing into the average.
The stress test says nothing about what an employer should do with this information regarding personnel. For decisions that affect the employment relationship, separate legal requirements apply, to which this method adds nothing and from which it takes nothing away. What the method does do is provide facts about which work can today be taken over by AI, which work is partly done that way under oversight, and which work remains human work — expressed in freed-up hours and FTE capacity, not in assumptions about who will still do that work in the future.
The first step is not setting up a full stress test, but making the assumptions that already exist visible. How do you recognize that a strategy is outdated and how do you measure whether a strategy works without waiting for the annual figures offer two angles to start with before any figures are collected.
In addition, there is the free assumption check: a short round in which you name your most important assumptions and see, per assumption, when it was last confirmed. Part of that is the question of which work in your company can genuinely be taken over by AI, something the work scan from FTE TO AI answers per task. The full stress test, with the outside view alongside the management team's self-assessment, is under construction.
Stel uw vraag. Vaak zit de echte vraag een laag dieper — daar mag ik naar vragen.
Answers come from this site’s knowledge base. Not tailored advice, and not a scan of your company.