A forecast and a plan are different documents and should not share a number

Two documents that keep getting merged

A forecast is a belief. It says what you think will happen, and it is answerable to the world: if the thing does not happen, the forecast was wrong. A plan is a commitment. It says what you will do, and it is answerable to you: if you do not do it, the plan was not followed. These are different kinds of claim with different failure conditions, and almost every schedule in existence contains both without labelling either.

The merged version is usually a date. The date is simultaneously an expectation and a promise, and because it has to serve both purposes it ends up serving neither. As a forecast it is optimistic, because promises are. As a commitment it is unreliable, because expectations move.

A target wearing the clothes of a prediction

The most common form of the confusion is a target presented as an estimate. Somebody wants a particular outcome, states it as what will happen, and the statement then does none of the things a forecast is for. Nobody can plan around it, because it does not describe the expected case, and nobody can learn from its failure, because failure will be attributed to execution rather than to the estimate.

This is a quiet but severe cost. An organisation that never separates the two loses the ability to keep score on its own predictions, and losing that means the estimates cannot improve. You can only calibrate a number that was intended to be true.

A plan consumes a forecast, and should say which one

Plans are built on top of predictions, which is fine and unavoidable. The problem is when the prediction is buried. A plan that assumes a supplier delivers on time, that nobody critical leaves, and that the approval arrives in the usual window has three forecasts inside it, and if none is written down then nobody can check whether they are jointly plausible.

Extracting them is the single highest-value thing you can do to a plan. Once the assumptions are listed, two useful facts become visible: which of them the plan is most sensitive to, and how many things have to go right at once. A plan resting on six independent-looking assumptions each of which is individually likely is a plan that is quite likely to fail somewhere, and that arithmetic is invisible while the assumptions are implicit.

They should respond to news in opposite ways

When evidence arrives, a forecast should move immediately and without embarrassment. That is what it is for. A forecaster who never revises is not being consistent; they are declining to use information.

A plan should be stickier, because a plan coordinates people and the value of coordination comes partly from its stability. Changing the plan every time the forecast moves imposes costs on everyone downstream who has arranged their work around it. The right pattern is that forecasts update continuously and plans update in deliberate steps, when the accumulated change is large enough to be worth the disruption.

The reverse pattern is the pathology to watch for: a plan that stays fixed while its assumptions are quietly known to be false, and a forecast that gets edited to keep the plan looking achievable. That combination has the two documents in exactly the wrong postures.

Where the confusion costs most

It costs most in personal projects, where there is no one to enforce the distinction. A person who says a piece of work will take three weeks has usually stated a hope, and when it takes six the lesson recorded is about their own discipline rather than about their estimating. The next estimate is therefore no better than the last, which is why people can spend years making the same error.

It also costs in negotiations, where the two documents get deliberately blurred. A date offered as a commitment invites the other party to plan around it; the same date offered as an estimate does not. Being explicit about which one you are giving is not pedantry, it is the whole content of the promise.

Keeping them apart in practice

Write the expectation and the intention as separate lines. What I expect: this, with a range, and here is what it is based on. What I am committing to: this, and here is what I will do if the expectation turns out to be wrong. The second line is where a contingent plan belongs, and a contingent plan is the honest way to hold a commitment while admitting uncertainty.

Then keep score only on the first line. A forecast is a scorable claim and improves with feedback; a plan is judged by whether it was carried out and whether it was a sensible thing to attempt. Mixing the two scores is how people end up believing they are bad at planning when they are actually bad at predicting, which are different problems with different remedies.