The default HubSpot deal stages come with default probabilities: 20%, 40%, 60%, 80%. Round numbers. Confident. Almost certainly wrong for your business. A calibrated probability set is the difference between a forecast you trust and one your CFO ignores.
Pull your conversion data first
For each stage, calculate: of deals that ever entered this stage, what percent eventually closed-won? That’s the empirical probability. Use a 12-month rolling window so you don’t include in-flight deals.
Stage: Discovery
Deals entered: 1,847
Deals closed-won: 167
Empirical probability: 9%If your “Discovery” probability is set to 20% and actual is 9%, your weighted pipeline is 2x reality.
Calibrate per pipeline
Enterprise deals close at different rates than SMB deals. Don’t reuse one probability set across pipelines. The motion is different and so are the conversion curves.
Account for stage skipping
Reps skip stages. A deal that jumps Discovery to Proposal skipped two stages of qualification. Your historical data treats it as if it entered every stage; your live data reflects the skip. Adjust by counting only deals that moved sequentially.
Watch for the 50% wall
Most teams’ Proposal stage clusters around 50% historically. Above 50%, sales reps unconsciously sandbag (they want the upside surprise). Below 50%, they over-promise. Sit on the 50% number unless data says otherwise.
Recalibrate quarterly
Pipeline behavior shifts with pricing changes, ICP changes, and economic cycles. Schedule a 30-minute calibration review at the end of each quarter. Update one probability per cycle if needed; don’t change all of them at once or you’ll lose your ability to attribute the shift.
Add a custom probability per deal
The stage probability is a default. Allow reps to override per deal with deal_probability_override. Calculate the working forecast from override when set, stage default otherwise:
effective_probability = COALESCE(deal_probability_override, dealstage_probability)
weighted_pipeline = amount * effective_probabilityDon’t bury bad news
If your calibrated probabilities show a 30% drop in pipeline value, surface it. The number was always wrong; you just see it now. Reframe to leadership as “we now have an honest forecast.”
What to do this week
Pull last 12 months of deals by stage. Calculate empirical conversion. Compare to your current stage probability defaults. Update the worst offender (largest gap) and let the new number run for one quarter before changing more.
