Do Decisions Have an Expiry Date?

Last week we met again with the partners of one of the leading companies in its sector, people we first sat down with about four years ago. At that first meeting, we had talked about how we might tackle some of their operational and decision processes together. But the company was preparing for a major ERP transformation, and naturally all attention was there. They had politely said, “Not now; let’s finish the ERP transformation first.” Four years have passed. The ERP has changed, after a rather painful period, but some of the fundamental process problems we discussed four years ago were still on the table. That same day I received an email from one of my teammates. Another company we had recently met was postponing talks until next year, again because of an ERP transformation. Reading the email, I smiled. My first thought was, “What a pity, they’ve made a mistake.” Then a new question joined the one I asked myself in the fourth essay: how do I know they really made a mistake?

Whatever we believe, it isn’t hard to find a saying to back it up. If we want to wait, we say “Better late than never”: don’t rush, wait for the right moment. If we want to act, we say “Strike while the iron is hot”: don’t miss the opportunity. Depending on the situation, both offer us a safe harbour. Perhaps this is exactly where one of the hard parts of management begins. We have no shortage of advice; we just don’t know which piece of it is right in a given situation. Just as a decision made too early can have a cost, so can making the right decision too late. And the second is much harder to see. There is no wrong decision, no KPI flashing red, no invoice showing money lost. Someone has simply said, “It’s not the right time yet.”

Anyone who has seen Interstellar will remember a curious problem of time. Cooper and his crew land on Miller’s planet, near a giant black hole. Because of the strong gravity there, roughly one hour on the planet equals seven years for those far away. When the crew return to the ship a few hours later, very little time has passed for them, but for the people they left behind, 23 years have gone by. Cooper watches the messages his children sent over the years in a few minutes. The same span of time has not carried the same cost for the people who lived it. Companies obviously have no black holes, but I sometimes think time flows differently there in a similar way. For a manager, “let’s wait another quarter” may mean three months. In operations, the same three months may mean thousands of manual tasks repeated every day; in sales, lost customers; for an employee, hundreds of hours; and for a competitor, three months to move ahead. The calendar moves at the same pace for everyone, but the time of decisions does not flow with the same value for everyone.

It would be easy to conclude from this, “Then let’s decide faster.” But waiting has value too. A small problem Richard Feynman got hooked on over lunch in the 1970s is about exactly this. Imagine you go to the same restaurant again and again. Should you order the dish you know and like, or keep trying new dishes for the chance of finding a better one? Feynman worked through the problem mathematically in his notes. A study published in 2026, which revisited those notes years later, showed that his approach corresponds to a mathematically optimal strategy. The problem touches a very basic tension in decision-making that we now call “optimal stopping”: how much should we search, and when should we stop searching and start using the best option we have?

What really strikes me about Feynman’s problem is this: when we try a new dish, we may find a better one, but at the same time we use up one of the meals in which we could have eaten that better dish. In other words, more information doesn’t come free. In business too, waiting another month may genuinely bring information that improves our decision. But for that information we also spend a month in which the decision could have been creating value. Besides, in Feynman’s restaurant we can assume how many meals lie ahead. In business, we usually don’t know how many opportunities remain. Will the same customer still be there in two years? When will our competitor move? What will a problem that is solvable today have turned into two years from now? We don’t know. We have to decide when to stop, without knowing how much time we have left.

Having worked inside many different decision problems at Arya for years, I can see that when managers say “let’s wait a little longer”, it is usually not bad management. Let the ERP be finished, the data cleaned, the organisation settled, the budget confirmed, the market a little clearer… Each of these is a perfectly reasonable argument on its own. But I think what makes waiting a good decision is not the waiting itself; it is knowing what we are waiting for. If we can’t say why our mind would change when the thing we are waiting for arrives, perhaps we aren’t waiting for new information at all. We are just moving the moment of decision to a more comfortable future.

AI will not make this problem go away; if anything, it will make it bigger. After all, we no longer need a lack of information as a reason not to decide. We can generate one more scenario, run another forecast, feed the model new data, look at dozens more alternatives. That is why decision quality is not only about what we choose; time is also a variable of decision quality. Beyond a certain point, the value new information would bring falls below the value we lose while waiting for it. In every meeting where we say “let’s wait a little longer”, perhaps we should ask ourselves two questions: what are we waiting for, and what are we willing to lose until then?

So my first thought about the company that pushed the project to next year because of its ERP transformation is still the same: they are making a mistake. I don’t say this because the company from four years ago is still discussing similar problems today. I say it because I can see the cost of waiting much more clearly now. Of course, it isn’t as dramatic as in Interstellar, but in organisations, too, the same three months are not the same three months for everyone. The person who can put a decision on hold can afford to wait; the system that decision affects keeps running. So if the person delaying the decision isn’t the one paying the cost of waiting, how can they decide how long it is right to wait?