Before any method: the data has to be honest
A forecast is arithmetic on your pipeline. If the pipeline contains deals that died in April and were never closed out, no method will help — the arithmetic will be performed faithfully on fiction.
So the first step is not choosing a method. It is going through open deals and closing out anything with no next action and no contact in twice your normal cycle length. This is uncomfortable and it is the single highest-value thing most teams can do to their forecast.
Method 1 — weighted pipeline
Multiply each open deal's value by the historical close rate of the stage it is in, and add them up.
The important word is historical. The percentage must come from your own closed deals — of the last hundred that reached Proposal, how many were won? If you assign the percentages by feel, you have produced a number that looks calculated and is not, which is worse than an honest guess because people trust it.
Good for: steady, repeatable sales with enough volume to compute real rates. Bad for: small numbers of large deals, where one deal swings everything and an average is meaningless.
Method 2 — the three-scenario forecast
Sort open deals into three buckets and report all three totals:
| Bucket | Definition | Report as |
|---|---|---|
| Commit | Verbal yes, or in signature; you would bet on it | The floor |
| Best case | Real, active, could close this period | The ceiling |
| Pipeline | Everything else open | Context, not forecast |
Less precise than weighted pipeline and considerably more useful in practice, because it forces a judgement per deal rather than applying an average to a category. It also communicates uncertainty honestly — a range is a truer answer than a single number, and everyone receiving it knows that.
Method 3 — run rate
Ignore the pipeline. Take what you closed in the last three periods, average it, adjust for anything you know is different, and use that.
This feels crude and frequently beats the other two, particularly in teams whose pipeline hygiene is poor. It is a good sanity check regardless: if your weighted pipeline says double your run rate, one of the two is wrong and it is usually not the run rate.
Run all three occasionally. When they disagree sharply, the disagreement is the finding — it usually points at stage inflation.
The two numbers to track over time
- Forecast accuracy. Forecast divided by actual, recorded every period. After six periods you will know whether your team runs optimistic and by roughly how much — and a known bias can simply be corrected for.
- Sales cycle length. Median days from created to won. Everything else depends on it, and it is the number that quietly changes when your mix of customers changes.
Saying the quiet part
Every forecast is wrong. The purpose is not accuracy for its own sake — it is deciding whether to hire, whether to spend, and whether to worry. A range with stated assumptions serves those decisions better than a precise number that turns out to be fiction, and it is much easier to correct in public.
Forecasting rests entirely on stage discipline: what actually moves a deal.