When do you actually know enough to decide?


More information can make a decision look better supported without making it any easier to make. The harder question is whether what we know is sufficient for the decision in front of us.
Across several pieces of work this year I’ve seen versions of the same moment. There are customer conversations, research, product notes, market data, decks and opinions from people who know the subject well. The discussion can be intelligent and detailed. Then somebody asks whether we are actually ready to decide, and the answer becomes much less obvious.
Usually, the difficulty is not a lack of information. One assumption may still be carrying more weight than anyone realised. Two credible sources may point in different directions. A customer comment may be genuine but too narrow to support the conclusion now being proposed. Or people may agree on the facts while disagreeing about what those facts mean for the decision.
I’ve started using a simple phrase for this: decision sufficiency.
When more information stops helping
The instinctive response to uncertainty is often to gather more information. Sometimes that is exactly right. A missing customer example, an unresolved technical dependency or a piece of contradictory evidence can materially change what we should do.
But another report, meeting or dataset is only useful if it helps resolve something that matters to the decision.
If another report gives us more detail without helping us choose between the live alternatives, we may know more without being any closer to deciding. If five more customer conversations repeat what the first five already told us, the limiting factor may no longer be the amount of evidence. We may simply not have identified which assumption still matters, what would disprove it, or what consequence we are actually prepared to accept.
That connects directly to the customer-engagement work I wrote about recently. Asking whether a conversation changed anything is useful, but even genuine recognition or new evidence does not automatically justify the next commitment. The next question is whether what changed is enough for the decision we now want to make.
Enough for which decision?
This is the part I think matters most. Sufficiency only makes sense relative to a particular decision.
A small, reversible test does not need the same evidence as a major commercial commitment. Changing a paragraph on a website is different from changing the market proposition behind it. Letting a technical POC continue is different from treating it as proof of commercial value. Asking one more customer question is different from hiring against an assumed demand pattern.
The stronger or less reversible the commitment, the more important it becomes to know what is established, what is still interpretation and what remains open.
That does not mean waiting for certainty. Most consequential decisions are made before certainty is available. The practical issue is whether the uncertainty is visible and whether the size of the action is proportionate to what we actually know.
A decision can still be reasonable when uncertainty remains. The difficulty comes when the presentation looks more complete than the evidence really is and the commitment grows with it.
What is still unresolved?
I find this question more useful than asking whether we have “enough data”.
What are we relying on that is not yet established? Is there a credible alternative explanation? Which contradiction have we not resolved? What would genuinely change our mind? If we are wrong, what consequence are we accepting? And does the person making the decision understand those boundaries?
Those questions do not need to turn into a large governance exercise. In plenty of situations the answer will be simple. The fact base is clear, the consequence is limited and the sensible thing is to act.
They become more important when a weak assumption can travel much further than the evidence behind it.
A technically successful POC, for example, may later be spoken about as commercial validation. Early customer interest may find its way into a forecast. A plausible explanation can influence a roadmap. None of those conclusions is necessarily wrong, but each asks more of the original evidence.
The question is whether the evidence is strong enough for the commitment now being made.
Sometimes the right next decision is smaller
There is another consequence of thinking this way that I like: insufficient evidence does not always mean stopping the work and commissioning more analysis.
Sometimes it means making a smaller decision.
Instead of committing to the whole proposition, test the assumption that matters most. Instead of turning an early pattern into a market claim, keep it as a hypothesis and look for evidence that might prove it wrong. Instead of treating a POC as commercial validation, separate what the technical work established from what still needs to be learned about adoption, value or willingness to buy.
That can keep the work moving without pretending we know more than we do.
It also gives negative evidence somewhere legitimate to go. A contradiction does not have to derail the whole piece of work. It may simply tell us that the next commitment needs to be narrower than we first thought.
Decision sufficiency is not a score
I would be wary of turning this into another maturity model or percentage.
The useful part is being able to see what the decision depends on, what remains uncertain and whether the consequence is appropriate to what is currently known.
Different decisions will have different thresholds. Different organisations will tolerate different levels of uncertainty. Some decisions need speed more than completeness. Others justify a much higher burden of proof.
So I don’t think the useful question is, “Do we have enough information?”
The question I now want to ask is more specific:
What remains unresolved, could it change the decision, and is the commitment we are about to make proportionate to what we actually know?




Comments