1
Set the forecast period and method
Decide whether you are forecasting monthly, quarterly, or annually. Choose your primary method: pipeline-weighted (best for deal-based B2B sales), historical trend (best for high-volume transactional businesses), or top-down from a market-share assumption.
π‘ Pipeline-weighted forecasting requires an accurate, up-to-date CRM. If your pipeline data is unreliable, start with historical trend analysis and layer in pipeline as data quality improves.
2
Pull and clean the historical baseline
Gather actual revenue for the prior 2β3 years, broken down by month. Identify seasonal patterns, one-time spikes, and the underlying growth rate. Use these as the anchor for your projections.
π‘ Remove one-time deals (large single contracts that are not repeatable) from the baseline before calculating your growth rate β they distort the trend.
3
Export and stage the pipeline
Pull the active pipeline from your CRM, grouped by deal stage. Assign a probability to each stage based on your historical win rate per stage, then calculate the weighted pipeline value.
π‘ Use 90-day rolling win rates rather than all-time averages β recent conversion rates reflect your current market and team, not legacy performance.
4
Build the monthly projection table
Translate the weighted pipeline into monthly revenue projections. Add a renewal or expansion revenue line if applicable. Sum to a quarterly and annual total.
π‘ Flag any month where projected revenue relies on more than 30% from a single deal. That concentration is a risk that should be visible in the forecast, not hidden in the total.
5
Break down projections by rep, territory, or product
Disaggregate the total by the dimension most useful for accountability β typically by sales rep or territory for a B2B team. Check that each rep's forecast sums correctly to the total.
π‘ Ask each rep to submit their own commit number before you build the top-down view. Comparing bottoms-up rep commits to your top-down model reveals alignment gaps early.
6
Document all key assumptions explicitly
List every variable the forecast depends on β win rate, average deal size, sales cycle, ramp time, churn rate β in the assumptions section. State the source and time period for each.
π‘ If a single assumption swings the forecast by more than 10%, flag it as a key risk and build a sensitivity row showing the impact of a 20% change in that variable.
7
Build base, upside, and downside scenarios
Create three versions of the total projection: base (most likely), upside (2β3 favorable outcomes), and downside (2β3 adverse outcomes). The range between downside and upside is your planning band.
π‘ The downside scenario should reflect a realistic bad quarter, not a catastrophic one. If hitting downside requires layoffs or a pivot, it belongs in a separate contingency plan.
8
Set a cadence to update actuals and track variance
Lock the forecast at the start of each period, then update actuals weekly or monthly as results come in. Record the original forecast number before overwriting it β you need the variance data to calibrate future forecasts.
π‘ A forecast that is never wrong is a forecast that is always being revised to match actuals. Measure forecasting accuracy over 4β6 quarters to hold the process accountable.