Not deciding is still a decision —
it's just the most expensive one

There is a deeply ingrained instinct in supply chain management to wait for more information before committing. It feels prudent. It feels professional. In stable environments, it often is. But in the environment we've been operating in for the last three years, one we will continue to operate in for the foreseeable future, this instinct has quietly become one of the most damaging habits in the function.

The core problem is an accounting one. When you make a decision (to fly goods, to place an early order, to lock a contract at current rates) the cost is visible and immediate. It shows up as a line item. Someone questions it. It needs justification, and the person who approved it owns it completely.

When you don't decide, the cost is invisible and distributed. It shows up as a stockout recorded as a lost sale that no one formally attributes to the delay. As a markdown on goods that arrived two weeks late and missed the selling window. As a competitor that filled the shelf while you were waiting for your container. None of these appear in the budget under "cost of indecision." They disappear into the noise of quarterly results. Nobody signs their name on them.

There is a second dimension that becomes even more relevant in volatile contexts. When the environment is chaotic, when tariffs shift, when routes are disrupted, when lead times are unpredictable, waiting for certainty is not a neutral act. Every day of delay is a day in which your options narrow. The container you could have booked last week is no longer available at the same price. The supplier slot you could have reserved is now committed to someone else. The freight forwarder who had capacity on Tuesday is fully booked by Thursday. Optionality decays. And once it's gone, you're not choosing between options anymore. You're managing consequences.

I want to be precise about what I'm not saying: this is not an argument for speed over accuracy, or for ignoring risk. There are decisions that genuinely require more information before they can be made responsibly. The problem I'm describing is different. It's the decision that could be made with the information already available, but doesn't get made because the incentive structure rewards waiting and punishes visible mistakes. That is a structural problem, not a judgment problem. And structural problems require structural solutions.

Let me make this concrete with declared hypothetical numbers, because the principle is more important than any specific figure. Suppose your airfreight cost to expedite a late shipment is €30,000. That number will get questioned in any budget review — it's large, it's unusual, it has a name and an owner. Now suppose that the stockout caused by not expediting generates €90,000 in lost margin over three weeks: missed full-price sales, markdown recovery on residual stock, and partial reorder costs. That €90,000 loss will not appear as a line item anywhere. It will show as slightly lower revenue, absorbed into the quarterly variance, attributed to "market softness" or "lower-than-expected demand." The airfreight approval gets interrogated. The inaction that cost three times as much goes unquestioned.

This is the structural distortion that produces bad decisions at scale. If your finance team can see the cost of action but not the cost of inaction, they will systematically push back on every visible expense while being blind to the larger invisible ones. The solution is not to argue the point in the moment. It's to build the comparison into the decision framework before the situation becomes critical. A simple pre-mortem table: "if we act now, cost is X; if we wait two weeks and the scenario plays out, cost is Y." That table changes the conversation entirely.

In my experience, the teams that handle volatility best are not the ones with the most sophisticated forecasting systems. They're the ones that have pre-agreed decision thresholds. Not "we'll decide when we have more data," but "if stock coverage drops below X days with Y days of lead time remaining, we authorize airfreight up to Z cost without further approval." That pre-authorization removes the delay, removes the political negotiation, and removes the option to wait entirely. I've seen companies build these triggers and cut their reactive airfreight spend by a third within two quarters, not because they flew less, but because they flew earlier and cheaper. The decision didn't change. The timing did.

One concrete place where AI earns its keep in this specific problem: building the comparison table at scale. Most planners have the instinct to run the cost-of-action calculation, but not the bandwidth to run it across 200 SKUs every week. An AI prompt structured around your coverage data, lead time variability, and margin by product can generate a ranked list of "decisions pending": the items where the cost of waiting is already compounding, ordered by financial exposure. It won't tell you what to decide. It will tell you what needs a decision urgently, and what the financial case looks like for each.

The limitation worth naming: this only works if your data is clean enough to trust. If your stock positions are updated weekly rather than daily, or if your sell-through data has a three-day lag, the model will calculate the right answer on the wrong inputs. AI with your numbers is leverage. AI without them is theater. Before you build the prompt, build the data pipeline.