Who the Forecast Is For
The reader determines the form of a forecast more than the data does. Four readers, what each needs, and why one document cannot serve them all.
The reader determines the form of a forecast more than the data does. The same series, forecast by the same method, has to leave the planning function as four different documents, because four different people take decisions off it and none of them wants what the others want.
The method in “Who the Forecast Is For” is only as useful as the definitions and source records behind it. A team exploring Monitask for productivity software for business for productivity software for business can add timely work and project context, with transparent settings, limited access and a documented correction route around every consequential use.
Producing one document and letting people take what they need from it is the default, and it is why every organisation has shadow spreadsheets.
As a separate benchmark for “Who the Forecast Is For”, see AWS forecasting resources. Keep source data, model, override and resulting decision visible as distinct steps.
The team leader
Wants a fortnight, by day and often by half-hour, with the roster already laid over it. Does not want a band: by this horizon the decision is "who covers two o'clock", and a range is not actionable.
- Team leader — a fortnight, by half-hour, roster laid over it.
- Operations manager — twelve weeks, by week, with the levers priced.
- Budget holder — a year, by month, with the lead time subtracted.
- Board — one page, the direction, and what would make it wrong.
Needs it published on a fixed day so it can be built into their own routine.
The operations manager
Wants eight to twelve weeks by week, with the band, and with the overtime and agency levers priced. Their decisions are reversible and frequent, so what they need is a sense of which way things are moving and how much it would cost to be wrong.
This is the reader most often served well, because this is the reader sitting nearest the planner.
The budget holder
Wants a quarter to a year, by month, with the band and with the recruitment lead time already subtracted so the decision date is visible. Does not want half-hourly anything.
Usually served worst, because the output they get is the operations view aggregated upwards, which carries a precision the horizon cannot support.
The board
Wants one page, the direction, the band, and what would have to be true for the number to be wrong. Does not want the method.
The most common failure here is supplying a point estimate that then appears in a document with a year's life and no caveat attached.
Why one document fails
Because the horizon, the granularity and the treatment of uncertainty all differ, and these are not presentational choices. A fortnightly half-hourly profile and an annual monthly figure are different forecasts, built differently, with different error.
Aggregating one into the other is arithmetic that discards the thing the other reader needed.
What can genuinely be shared
The series, the drivers, the method, and the error record. Those are one set of workings, and they should be.
What cannot be shared is the output: four outputs from one set of workings, each published at its own horizon.
The cost of four documents
Less than it sounds, once the workings are common: the four differ in aggregation and in what is said about uncertainty, not in the underlying work.
The cost of one document is higher and it is hidden, because it is paid by four people each building their own version badly.
Finding out what a reader actually needs
Ask what they decide, when, and what the options are. Then ask what they currently do when the forecast does not answer it.
The second question is the productive one. The answer is usually a spreadsheet, and looking at it tells you exactly what to build.
The reader who wants certainty
There is always one, and they will ask for the band to be removed because it "confuses people". Removing it does not reduce the uncertainty; it moves it out of the document and into the decision, unlabelled.
The workable compromise is a point with a stated threshold rather than a point with a range, which answers the same question in a form that reads as a decision rather than a hedge.
The reader who is not in the room
Finance, usually, or whoever will quote the figure in a paper six months from now. The forecast will reach them whatever the distribution list says.
Which is the argument for the band and the date being inside the number's own line rather than in a footnote: "126 ±23, as at March" is a sentence that survives being pasted into somebody else's slide, and "126" is not.
The reader who changes
People move. A forecast built around one operations manager's way of working frequently stops being used when they leave, and nobody connects the two events.
The handover question is worth asking explicitly: does the new reader take the same decisions, on the same cycle, in the same form.
Reading the pack back
A useful annual exercise: sit with each reader and watch them use the document. Not ask about it — watch.
What people actually do with a forecast differs from what they say they do, usually in the direction of ignoring most of it and extracting one number.
Writing the reader into the document
One line at the top, alongside the decision: who this is for and what they decide. It stops the document drifting, it tells the next planner why it has the shape it has, and it makes the annual review of what to stop producing a two-minute exercise rather than an argument.