Key takeaways
- Clean the deal list before weighting it; stale dates and missing next steps distort everything.
- Weight by stage, keep commit separate, and never count the same deal twice.
- Add seasonality and delivery capacity, because demand you cannot serve is not revenue.
- Track forecast error by stage and owner, then recalibrate rather than argue.
On this page
Start with the pipeline you already have
A sales forecast is only as good as the list behind it. Before choosing a method, clean the pipeline: one row per opportunity, a value, a stage, an expected close date, an owner, and a next step with a date. Remove deals with no activity for a defined period, split multi-phase deals, and correct close dates that have slipped more than twice.
Do not forecast from the CRM dashboard alone. Stage names often hide different behaviour between teams, and a deal can sit in a late stage without a buyer, a budget, or a signed proposal. Build the forecast from the deal list, then reconcile it to the CRM total.
Choose a method that matches your data
More sophisticated methods need more history and more discipline. Use the simplest method the data can support, and move down the table only when the current method has been tested against actual closes.
| Method | Data needed | Use when | Main weakness |
|---|---|---|---|
| Run-rate | Revenue by month | Sales are repeatable and stable | Misses pipeline changes |
| Stage-weighted pipeline | Open deals, stages, and values | Deals follow a consistent path | Stage probabilities can be optimistic |
| Cohort conversion | Historical wins and losses by stage | Enough closed deals to compare | Thin data in small teams |
| Driver model | Leads, conversion, deal size, capacity | Marketing and sales inputs are measurable | Easy to overfit assumptions |
Build a stage-weighted forecast
Multiply each open deal by its stage probability, then add the results. If three deals are worth £40,000 at 50%, £25,000 at 20%, and £60,000 at 10%, the weighted forecast is £20,000 plus £5,000 plus £6,000, or £31,000.
Keep three numbers separate: commit, best case, and pipeline. Commit contains deals with agreed terms and a clear purchase order path. Best case adds deals with a credible next step. Pipeline is everything else. A single blended number hides the risk.
- Exclude deals with no next step or no dated activity.
- Cap the close date at the current period only when the buyer has confirmed it.
- Use observed conversion by stage once you have enough closed deals.
- Apply the same rules to every salesperson.
Add seasonality and capacity
A pipeline forecast assumes the business can deliver what it sells. Check capacity before accepting the number: delivery hours, stock, onboarding slots, and support cover. If the peak month is forecast at 12% above average but capacity allows 10%, forecast 10% and record the unmet demand as a risk.
Seasonality should come from at least two full periods, not from memory. Build a monthly index, apply it to the underlying run rate, and label the result as an assumption. Where there is no history, use a conservative range and revisit it after the first full cycle.
Run three scenarios and a weekly review
Present a downside based on commit only, a base case on the weighted pipeline, and an upside that adds the best case. The gap between them is the uncertainty the business is carrying, and it should be visible to whoever plans hiring or stock.
Review deal-level movement weekly and the total monthly. Track forecast versus actual by stage, owner, and deal size. If the error is consistently above 10% for three months, recalibrate the stage weights or the close-date rules rather than asking for more optimism.
- Move commit deals that have slipped twice out of the current period.
- Put new deals added after the mid-month cut into the next forecast.
- Compare the forecast with the finance revenue plan and explain the difference.
- Record why a deal was won or lost, not only its value.
Keep assumptions visible
Put the assumptions on one page: stage weights, average deal size, conversion, seasonality, capacity, currency, and the date of the data cut. Give each assumption an owner and a review date. Version the page so a later reader can see what changed.
A forecast that cannot be explained six weeks later is not a forecast; it is a number that happened to be right or wrong. The assumption log is what turns the exercise into a repeatable process.
Frequently asked questions
Is a CRM required to forecast sales?+
No for a small team. A structured spreadsheet works, provided the fields, definitions, and update rhythm are enforced. The process matters more than the tool.
How many pipeline stages should we use?+
Five to seven that match how buyers actually progress. Fewer if deals are simple and fast; more only if each stage has a distinct exit criterion.
How do we forecast without historical data?+
Use commit value plus a conservative conversion assumption, label it clearly, and replace the assumption with observed conversion after a few cycles.
Should the sales forecast match the finance forecast?+
They should reconcile. Explain differences in timing, probability, scope, and currency rather than forcing one number to match the other.
Need help applying this?
Get help with Automation & Workflows.
Review the service scope or tell us what you are trying to improve. We will reply with a clear next step.
Free KPI definition templateRelated reading
Variance analysis: price, volume, and mix
Price, volume, and mix explain most revenue and margin gaps. A worked example shows how to split the variance and reconcile it to the total.
Reporting calendar: keeping packs on schedule
A reporting calendar is a dependency map. Build it backwards from the review date, set cut-offs, and measure every miss to keep packs on schedule.