Why bother forecasting at all? Because every big decision in a growing business — hiring a rep, signing a lease, buying stock — is a bet on future revenue. A sales forecast turns that bet from a feeling into a number with reasoning behind it. Even a rough forecast, honestly built, beats confident guessing.
Why sales projections go wrong
Almost every blown forecast fails the same way: deal stages that mean nothing. If "proposal sent" includes deals the buyer has ghosted for two months, any percentage you apply to that stage is fiction. The fix isn’t a better formula — it’s stage definitions based on what the buyer did, not what the rep did. A deal is in "decision" because the buyer agreed to a decision date, not because you emailed a quote.
Rule of thumb: if a deal has had no buyer action in 30 days, it’s not late — it’s dead. Forecast it at zero and be pleasantly surprised.
Three sales forecasting methods that fit SMBs
Stage-weighted pipeline
Multiply each open deal by your measured historical conversion rate for its stage, and sum. The default method once you have six months of clean stage data.
Run-rate with seasonality
Average your last 3–6 months of closed revenue, adjusted for known seasonal swings. Crude but honest — the right method while your stage data is still messy.
Bottom-up capacity
Reps × conversations per week × conversion rate × average deal size. Best for planning: it shows whether the target is even possible with the team you have.
Run two methods side by side. When they disagree sharply, the gap is telling you something — usually that the pipeline is padded. For a deeper tour of methods, HBR’s classic guide to choosing a forecasting technique still holds up.
Clean data is the whole game
Every method above is only as good as what’s in the CRM. That’s a process problem, not a discipline problem: when pipeline stages update through automation instead of memory, and every conversation follows the same Blueprint stages, the forecast starts writing itself.
Key takeaways
- A sales forecast is arithmetic on the pipeline — dishonest pipeline, dishonest forecast.
- Define stages by buyer actions, and zero out deals with no buyer movement in 30 days.
- Use stage-weighted, run-rate and bottom-up methods; let disagreements between them expose padding.
- Automate data capture so the forecast reflects reality, not memory.