Virtually every growth plan of recent years contains a variant of the same sentence: we can grow as fast as we can hire and train people. That assumption was not naive. It came from a period in which vacancies stood open for months, in which onboarding time effectively determined growth speed, and in which capacity was almost one-to-one equal to the number of heads on the payroll. Anyone who wanted to grow faster than the labor market allowed got stuck on waiting lists for people, not on demand for the product.
The assumption was correct because capacity and FTEs were the same thing. There was no third variable. If the work increased, the number of people had to increase along with it, in a fixed ratio. That made the labor market a hard ceiling: however good your plan was, you could not grow beyond what you could staff.
The ratio between work and headcount is no longer fixed. For part of the tasks that used to automatically mean a new FTE, AI can now take over the work, in part with human oversight that assesses and corrects, and for another part it remains unchanged human work. This is not a future scenario: it is already happening, unevenly. A company where reporting, first-line customer contact, or document processing has largely been taken over experiences a different relationship between revenue growth and staffing needs than a company where this has not yet been figured out.
The difference between those two companies rarely lies in the sector or the size. It lies in whether someone has already determined which part of the work falls into the three categories: takeable-over, takeable-over-with-oversight, or unchanged human work. Companies that have figured this out per task see their growth ceiling shift. Companies that have not yet done so are still planning with the old ratio, without anyone having consciously made that choice.
The first signal is usually not AI news, but a budget discrepancy: next year's staffing budget is still being drawn up as a linear function of revenue growth, while in execution tasks have already been shifted to systems that require approval rather than execution. A gap arises between what the budget assumes and what the work floor is already doing.
A second signal is that recruitment pressure and revenue growth diverge without anyone being able to explain why. Vacancies that used to be opened as a matter of course sometimes remain unfilled without service suffering as a result. This is not a coincidence; it is an indication that the fixed ratio between work and FTEs has already been broken in places, while the growth model still assumes it is intact.
A third signal is strategic in nature: competitors with fewer people grow just as fast or faster. This is often attributed to chance, capital, or a favorable market. More often it is a difference in how far one has progressed in figuring out which work can be transferred to AI.
This page addresses an assumption about capacity, not a decision about people. Whether, and how, an organization adjusts staffing in response to changed capacity is up to the employer, and its own legal requirements apply, which are not addressed here. What can be tested here is: does the rule of thumb still hold with which you convert growth into staffing needs? That is a question about the assumption, not about the people on the payroll.
The underlying question — which work in this company can genuinely be taken over by AI — is answered per task with the work scan from FTE TO AI, separate from what an organization subsequently does with that outcome.
The brake-on-growth assumption rarely stands alone. It is often linked to the assumption that this work cannot be standardized, to the idea that the margin lies in execution and not in the design of the process, and to the notion that a new entrant needs years to build up the same capacity. Anyone who wants to understand how such a test works in general can find the method at how you check whether your strategy is still based on reality, and what a strategic assumption is exactly and when it counts as expired is described at the explanation of assumptions and their validity period.
To know whether the assumption still holds, a fixed set of measurements is needed, repeated on a fixed rhythm rather than established once:
A first step is not the full stress test, but a free assumption check: a short conversation in which you name your key assumptions and see, per assumption, when it was last confirmed. The full stress test, with sector data, a self-plot by the management team, and an ongoing assumptions list, 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.