A description of the present cannot be falsified, which is why it is comfortable to sell and hard to learn from.
Why it matters when the plan changes
Most organisational analysis stops at description. That is where the market is and where the methods are mature, along the analytics ladder from descriptive through diagnostic, and it leaves the decision-maker to do the extrapolation themselves. A forecast does that step explicitly and publicly, which transfers the risk of being wrong from the reader to the author. Timing matters here: information delivered while options are still open has value that the same information delivered after commitment does not, which is why a diagnosis produced after the fact reads like an autopsy rather than a warning.
The tension is that forecasts are uncomfortable to produce and to buy. A diagnostic can be accepted or disputed and never settled; a forecast has a date on which someone finds out, and predictions can be scored against outcomes once recorded beforehand. That exposure is precisely what allows a track record to accumulate, and it is why the forecast rung of the ladder stays emptier than the diagnostic one.
In practice
A consulting report describes an organisation's decision-making as slow and its interfaces as unclear, accurately. Nobody can say afterwards whether it was right, because it did not claim anything would happen. The same evidence expressed as a forecast would have named a milestone, a date and a confidence, and would have been checkable in a quarter.
Evidence
Forecasting is the process of making predictions that can later be compared with actual outcomes.
Forecasting, Wikipedia (2026)Scored, feedback-driven forecasting improves judgement, which requires that predictions were recorded beforehand.
Superforecasting and the Good Judgment Project (2026)
What it cannot tell you
The distinction only separates what can be scored from what cannot; it does not tell you whether a given diagnostic is accurate or a given forecast is well calibrated. A forecast stated with unwarranted confidence is not automatically better than an honest diagnostic, and the split says nothing about the skill needed to build either one.
Questions
Yes, as the input. A forecast rests on an accurate reading of the present, so the diagnostic work still has to be done. The argument here is only about where the output stops, and not about whether the underlying description matters.
A stated expectation about what will happen, by when, with a confidence attached, in the sense Wikipedia's 2026 entry on forecasting describes: predictions made from past and present data that can later be compared with outcomes. Remove the date or the confidence and it becomes an unscoreable opinion about the future.
Because diagnosis is safe. A description can be discussed indefinitely and never settled, so no supplier accumulates a public record of being wrong. Forecasting exposes the author to being wrong, and that exposure is exactly what makes the output worth more.
It can, with an honest confidence attached, which is the point of stating one. The Good Judgment Project's approach to superforecasting scores predictions using brier scores once they are recorded beforehand, which works even with thin evidence and simply produces a lower-confidence, still checkable forecast.
For a date and a confidence alongside the findings, and for the reasoning to be recorded so the call can be reviewed. A supplier unwilling to attach either is offering description, which should be priced and relied on as description.