A forecast can tell you what may happen.

It cannot tell your company who is allowed to respond.

That distinction is easy to lose when the forecast arrives in a dashboard or a planning deck. The number looks more disciplined than last quarter’s judgement. The chart creates the feeling that uncertainty has been converted into a plan.

Then the signal changes and everyone gathers to decide what it means.

That is usually the moment the organisation discovers it has built a better way to see uncertainty, but not a better way to respond to it.

Forecasting is not decision-making

A forecast estimates what may happen. A decision commits people, money, capacity, customer promises, or risk appetite in response.

Those are different jobs.

A demand forecast may say that a product is likely to sell faster next month. It does not itself decide whether to buy more inventory, reserve production capacity, change a customer promise date, or wait for another week of evidence.

A cash forecast may show a tightening position in six weeks. It does not decide whether to accelerate collections, pause discretionary spending, draw on a facility, or accept the risk.

A supply disruption may become more plausible. It does not decide which supplier must be qualified, what stock is worth carrying, or who may approve the cost.

The forecast is allowed to be uncertain. The response cannot be undefined.

When it is undefined, the team has to invent it under pressure. They debate the model, re-litigate assumptions, negotiate authority, and discover that the action affects another department that was not part of the planning process. By then the forecast has stopped being foresight. It has become a late warning.

Decide the response while the signal is calm

For an important forecast, the operating question is not only, “How accurate is this?”

It is also, “What will we do differently if it moves?”

That question should be answered while the signal is still calm.

A useful forecast needs a response contract. It does not have to be elaborate, but it has to make six things visible:

  1. The decision being governed. Name the actual choice: purchase order, staffing level, credit hold, production slot, customer commitment, or investment gate.
  2. The signal and threshold. Define what movement matters, over what horizon, and whether the trigger is a probability, a range, a confidence interval, or a combination of indicators.
  3. The response triggered. Decide in advance whether crossing the threshold triggers an automatic action, a mandatory review, or an escalation. These are not interchangeable.
  4. The decision authority and approval limit. State who may commit resources, who must approve a larger commitment, and what no one may do on the basis of the forecast alone.
  5. The named owner and review deadline. Identify who monitors the signal, who convenes the review, and how quickly the decision must be revisited. A dashboard without an owner is a screen saver.
  6. The exception and escalation path. Define what happens when the signal conflicts with a customer commitment, a regulatory constraint, on-the-ground knowledge, or another fact the forecast did not capture.

In many cases the right result is a review, not an automatic action. But a review is still an operational commitment. It says who brings the evidence, who has authority to decide, and what happens when the decision cannot wait.

A complete contract is more useful than another chart

Consider a six-week cash forecast for a business whose working capital is tightening. The forecast is evidence, not permission to make every response automatically.

A compact response contract might read like this:

  • Decision: protect liquidity without breaking essential customer or regulatory commitments.
  • Signal: projected cash headroom falls below the agreed minimum at any point in the next six weeks.
  • Trigger: mandatory review within one business day, not an automatic stop to spending.
  • Owner: finance monitors the forecast; the CFO convenes the review.
  • Authority: collections acceleration within normal commercial terms is authorised immediately. A discretionary-spend pause requires COO approval. A draw on a facility remains with the authorised finance signatory.
  • Exception: a customer-critical delivery, contractual obligation, or regulatory expenditure goes to the named CFO/COO forum with the reason recorded.

The precise numbers and titles will differ by company. The useful part is the separation.

The person watching the signal is not necessarily the person who can commit cash. The person who convenes the review is not necessarily the person who can override a customer commitment. An exception is not a failure of the contract. It is a condition the contract anticipated and gave to an accountable authority.

That is what prevents a forecast from becoming a meeting about who is allowed to respond.

A threshold is a business boundary

Teams often treat thresholds as a technical parameter. Someone tunes an alert from 60 to 70 percent because there are too many notifications.

The threshold is more consequential than that.

It is a statement about which error the business is prepared to carry.

Order too early and you may tie up cash, create obsolete stock, or pay for capacity you do not need. Order too late and you may miss a sale, break a service commitment, or lose a customer who will not wait. A generic accuracy score does not by itself establish which of those costs the business is willing to accept.

This is why the forecast owner cannot be only the model team. The people who understand supplier lead times, working capital, service levels, contractual penalties, and customer relationships have to help define the threshold.

A forecast can make the trade-off more visible. It does not receive business authority merely because it calculated the probability.

The same discipline applies to scenarios

Scenario work has the same operational test. A scenario does not need to predict the future exactly. It needs to identify which change in evidence should trigger a prepared review, escalation, or action.

The value of a scenario is not that it predicted the future. The value is that it made a future response less improvised.

The record should therefore preserve the signal, the threshold, the authorised decision, and any exception. Afterwards, the organisation can examine whether the forecast was wrong, whether the threshold was wrong, whether an exception made sense, or whether the right people had the authority and evidence they needed.

Once the threshold moves, the organisation should not still be negotiating who has authority to respond.