Safety doesn't come from more planning. It comes from faster learning.
Nobody believes the leadership team that claims to know where the company will be in three years. We build the plan anyway. Three disciplines that keep risk flat, without the firm having to dress up as a startup.
In most leadership team meetings I've seen in the last twelve months, a variant of the same scene played out. Someone presents a three-year plan. Everyone nods. Nobody really believes it. Six months later the same plan is quietly corrected at the offsite.
That's not embarrassing. It's reality.
What would be embarrassing: drawing the conclusion from this experience that you need a better plan. More detailed, with more scenarios, more approval gates. This reflex feels like safety. It isn't. It only delays the moment when the leadership team realises it deceived itself. The loss rarely comes as bankruptcy. Often as two years of stagnation that look unremarkable on the balance sheet.
What "uncertainty" actually means
Uncertainty isn't a storm that passes. It is the condition under which all material investments are made today. Markets shift faster than any plan. Customer behaviour tips within quarters. Competitors emerge from adjacent fields where nobody was a year ago.
The honest answer to that isn't: plan better. It's: learn faster whether the plan holds.
That's the question startups have been operating on for twenty years. Not because founders are smarter. But because they couldn't afford to wait six months for confirmation of a hypothesis. They wouldn't have survived.
Mid-size companies with healthy balance sheets can't afford it either. They just notice it later.
Risk stacks up
The decisive point is mathematical, not philosophical. If you make moves without validation, risk compounds with every move. You invest in a new region on the basis of an assumption. You launch a new product on the basis of two. You build a team on the basis of three. If even one of those assumptions was wrong, everything that follows stands on borrowed ground.
With regular tests, the opposite happens. Each move dissolves an assumption before the next is placed on top. Risk stays flat, even when the environment is unpredictable.
The difference isn't whether you take risk. Both models take risk. The difference is whether the risk becomes visible before it becomes expensive.
Three disciplines that hold in the leadership team
First discipline: every strategic assumption gets a falsification number. Not "we believe more customers in DACH want the product", but "in the next ninety days we see at least 40 qualified enquiries from DACH, or we correct course". This number must be set before the test, otherwise results get rationalised after the fact instead of decided.
Second discipline: tests need a responsible person and a date, not a committee. If a test doesn't deliver a result in four to six weeks, it isn't a test. It's a project with a wish. The most common trap in firms with 30 to 300 employees: a test becomes a steering committee that stretches the learning cycle from four weeks to four quarters.
Third discipline: separate language for assumption and fact. In every meeting you run from next week onward, you could introduce this. Whoever makes a statement says either "we measured" or "we suspect". Both are allowed. But they don't get mixed any more. This one linguistic discipline changes the quality of decisions more than any new method.
What goes wrong if you don't do this
The most common sentence I hear after four or five years in a company isn't "we should have prioritised differently". It's: "we should have noticed much earlier that this one assumption didn't hold."
Noticing early has a name. It's called a test. It isn't young, isn't fashionable, isn't startup-specific. It's the mechanism by which organisations check reality before they book it into the P&L.
What to do this week
Write down the five most important assumptions of your current strategy on one page. Not ten, five.
For each, mark whether you have tested it with the behaviour of real customers in the last ninety days. Not with a workshop. Not with a survey. With behaviour.
For every assumption where the answer is no: who is responsible, by when is there a result, which number decides?
You'll probably find that at least two of the five assumptions have been sitting untested in the plan for months. That's normal. It's also why the question at the start isn't whether you can afford to run tests. It's whether you can afford not to.
