Sales forecasting for SMEs: a simple practical method

Pipeline value is the sum of every open deal. It matters for coverage — is there enough to make the target — but it is not what will close: most of it is at early stages and will not close this period, and some of it will never close. The forecast asks a narrower question: of the deals with a close date in this period, weighted by how far each has really got, how much revenue is expected? A pipeline of one crore and a forecast of twenty lakh for the month are both correct; confusing the two is how owners plan cash on hopes.

Keep the two numbers on the same page, labelled, and never let the pipeline number stand in for the forecast in a bank conversation or a hiring decision.

Stage probability comes from your history: if, over the last year, a third of deals that reached “quote sent” closed and three-quarters of those at “sample approved” closed, those are the probabilities for those stages — not the round numbers the CRM shipped with. Recompute quarterly. Evidence checks the stage: a deal at “negotiation” with no written quote and no named decision-maker is not at negotiation, whatever the salesperson feels; the review moves it back. Close dates are the buyer’s: a date the salesperson chose because it was month end is not a date, and a deal whose date has passed twice is either stuck or gone.

Then each deal contributes its value times its stage probability, if its close date falls in the period. That weighted sum is the arithmetic forecast. The judgement layer sits on top: the two or three large deals that would swing the number are reviewed individually, with the specific next step and the specific risk, because no probability describes a single large deal well.

Commit is the revenue you would stake the month on: deals at late stages, with the buyer’s date in the period and the evidence in place, taken at close to full value. Best case is commit plus the deals that could reasonably land — mid-stage, dated in the period, with a real next step. Pipeline is everything else dated in the period, weighted. Reported weekly as three lines against the target, with the change from last week and the reason for it: “commit down three lakh — the Ramesh order slipped to next month, buyer confirmed”.

The three numbers answer three different questions the owner has: can we make payroll, could we make target, and is next month covered. One number cannot answer all three, which is why a single forecast figure is always argued about.

The mistakes: forecasting from pipeline value; using the CRM’s default probabilities; close dates set to month end by habit; letting a big deal sit in commit for three months; sandbagging so the forecast is always beaten, or inflating it so it never is; and never comparing the forecast with what happened. A safeguard: keep last month’s forecast and score it against actuals — the gap, and the reasons, are the method improving.

Measure accuracy monthly: commit against actual, best case against actual, and which stages’ probabilities were wrong. After a quarter, an SME forecast built this way is usually within a fifth, and the owner plans cash, hiring and purchasing on it. This is the forecasting support we provide — the probabilities learned from your deals, the stage evidence rules, the three-number weekly page, and a quarter of reviewing it with you until it is trusted — and the free audit starts by comparing what your pipeline said last quarter with what closed.