The Decision Has a Lead Time, and It Sets the Horizon
Why the horizon is read off the decision's lead time rather than off the calendar, how to measure a lead time honestly, and what to do when it is longer than the data supports.
The horizon of a forecast should be read off the lead time of the decision it feeds, not off the calendar and not off what the system happens to produce. If recruitment takes eleven weeks, the recruitment forecast is a twelve-week forecast, whatever the financial year does.
The horizon in “The Decision Has a Lead Time, and It Sets the Horizon” should determine which current records are relevant and how much detail the forecast can honestly support. A planning team evaluating open this Monitask resource for time reports for invoicing can connect recent time and project context with short-term capacity questions, while keeping longer-range assumptions in a separately governed model.
This sounds obvious and is almost universally ignored, because reporting cycles are monthly and quarterly and the forecast inherits their shape without anybody choosing it.
Additional context for “The Decision Has a Lead Time, and It Sets the Horizon” is available from Harvard Business Review forecasting resources. External material is most useful when it prompts a specific validation check rather than replacing local judgement.
Measuring a lead time honestly
End to end, from the moment the decision could be taken to the moment the capacity is actually available. For recruitment that is approval, advertising, sitting, offer, notice period, induction and the ramp to competence.
Most organisations quote the advertising-to-offer part, which is usually a third of the real figure.
A worked lead time
| Stage | Typical | Range seen |
|---|---|---|
| Approval to advertise | 2 weeks | 1–6 |
| Advert to shortlist | 2 weeks | 2–3 |
| Interview to offer | 1 week | 1–4 |
| Notice period | 4 weeks | 2–12 |
| Induction to useful | 3 weeks | 2–8 |
| End to end | 12 weeks | 8–33 |
One employer's own figures for one role family. The point is not the total but the range: the same decision has a lead time between two and eight months depending on the role, and one horizon will not serve both.
The horizon is the lead time plus the review interval
If the decision is reviewed monthly and the lead time is twelve weeks, the forecast has to reach sixteen weeks: by the time the next review comes round, four weeks have already gone.
Forgetting the review interval is the commonest way a horizon comes out a month too short, and the symptom is a decision that is always taken slightly too late.
Different lead times, different horizons, same series
Overtime: one to two weeks. Agency: two to four. Training starts: six to eight. Recruitment: eight to thirty-three. Premises: a year or more.
Five horizons from one history, each published separately, each with its own band. A single number is not a simplification of this; it is a refusal to do it.
When the lead time is longer than the data supports
It happens, particularly with premises and with roles that take half a year to fill. The honest response is to say what the forecast can and cannot carry at that distance, and to let the decision maker choose between a wide band and a shorter commitment.
What is dishonest is producing a point estimate for a horizon the data cannot reach, because the point will survive and the caveat will not.
Shortening the lead time is usually cheaper than forecasting further
Nobody says this often enough. Taking four weeks out of a recruitment process improves the decision more than any amount of modelling at sixteen weeks, because the band at twelve weeks is materially narrower than the band at sixteen.
That makes the lead-time measurement a joint exercise with whoever owns the process, and it is usually the most productive hour in the whole project.
The decision that has no lead time
Some do not: a staffing level that can be changed the same day, a task that can be deferred without consequence. These need monitoring, not forecasting, and a forecast made for them is effort spent on a question that answers itself.
Separating these out early removes perhaps a quarter of what organisations ask their planners to forecast.
Lead times change, and nobody tells the planner
A new approval step, a tighter labour market, a change to notice periods. The lead time moves by a month and the horizon does not, and the forecast quietly stops matching the decision.
An annual check with the process owner is the whole maintenance requirement, and it takes a conversation.
Writing it down next to the number
"Horizon: 16 weeks. Recruitment lead time 12 weeks end to end, measured March; monthly review adds 4." Two sentences at the top of the forecast.
They tell the next planner why the horizon is what it is, and they are the first thing to check when the forecast starts feeling off.
The lead time that is mostly waiting
Worth separating when it is measured: the part of the lead time spent doing something and the part spent waiting for an approval or a calendar. The second is usually larger and is usually the cheaper to shorten.
An approval step that adds three weeks to a twelve-week lead time is a quarter of the horizon, and nobody has ever priced it that way.
What it changes in practice
The decision stops being taken late, which is the single largest avoidable cost in workforce planning and the one least often attributed to the forecast.
A late decision with a perfect number and a timely decision with a rough one produce very different quarters, and only one of them is recoverable.