A forecast is built from three numbers: amount, probability and date
Every open deal in your CRM has an amount, a stage and an expected close date. Those three fields are what a forecast is made of. The amount tells you what you could earn. The probability tells you how likely it is. The date tells you which month the money lands in.
The simplest formula multiplies each deal amount by its stage probability and adds the results up by month: a weighted pipeline. A deal worth AED 40,000 at a stage with a 25% probability contributes AED 10,000 to the forecast. No single deal pays exactly that, but across many deals the errors cancel out.
The weak spot is data quality. A blank amount drops the deal out, and a close date that has said "tomorrow" for three months keeps the forecast sliding. So the data comes first, more on that below.
Stage probability comes from history, not gut feel
In many portals stage probabilities were typed in during setup: 10, 30, 50, 90%. Those are guesses. A real probability comes from closed deals over the last 6 to 12 months: of all deals that reached a stage, how many ended up paying.
| Stage | Deals that reached it | Of those, paid | Probability |
|---|---|---|---|
| New lead | 400 | 48 | 12% |
| Qualified | 200 | 48 | 24% |
| Quote sent | 120 | 48 | 40% |
| Negotiating terms | 70 | 48 | 69% |
| Invoice issued | 55 | 48 | 87% |
A hypothetical example with made-up figures to show the calculation. Your probabilities are calculated from your own deals.
Recalculate probabilities every quarter and separately for each line of business. A repeat order and a first deal with a new client behave differently at the same stage, and so may separate free zone and mainland teams.
The sales cycle moves the forecast between months
Reps set close dates optimistically, so the deal date is compared with the average cycle from each stage to payment. If deals at "Quote sent" typically close 35 days later and a rep has set a date one week out, the forecast uses the calculated date instead.
The second rule covers stalled deals. A deal that has been sitting at a stage two or three times longer than average is very unlikely to close on time. Those deals go onto a separate list for the sales manager and carry less weight in the forecast. You can hand stalled-deal control to the AI sales manager: the agent watches the pipeline, nudges the rep and alerts the manager if nothing happens.
