Guide
How to Forecast Revenue (Without Lying to Your Board)
To forecast revenue, multiply your qualified pipeline by your historical win rate, subtract the deals that will slip past the quarter, then discount for variance. The result is your bankable number, the revenue you can defend at 90% confidence. Everything above it is hope.
Most teams never run that math. The average sales organization closes only 48.2% of deals as originally forecast, and wins just 46.9% of forecasted deals (CSO Insights, World-Class Sales Practices Study). Half of a typical commit is weather.
This page covers the method, the coverage math behind it, and a free calculator that runs the whole thing in about sixty seconds.
The diagnosis
Why revenue forecasts miss (it's never the spreadsheet)
Forecasts don't fail in the spreadsheet. They fail upstream, in the revenue system the spreadsheet is trying to describe. There are five layers where the miss gets built, usually months before anyone sees it in the number.
Market Physics. Who you sell to sets your win rate before a rep ever makes a call. Sell into a market that doesn't feel the problem, and no forecasting method survives contact with it.
Foundation. Thin pipeline can't be forecast. It can only be prayed over. If coverage is short, the forecast conversation is already a fiction conversation, and everything below this layer is rearranging it.
Inspection. Slip is invisible without deal discipline. Deals don't announce they're drifting, they just stop matching their close dates. No inspection rhythm, no early warning. The miss shows up in week thirteen instead of week four.
Execution. Win rates drift down quietly. A point here, a point there, and the same pipeline produces less revenue every quarter while the forecast model still assumes last year's conversion.
Trust & Proof. Deals stall where belief is missing. A buyer who isn't convinced doesn't say no. They say next quarter. That pattern lives in your slip rate, and it compounds.
Name the layer that's leaking, and the forecast fixes itself downstream. That's the order of operations most teams run in reverse.
The method
The bankable number method (four steps)
Qualified pipeline only, and only deals that can close inside the quarter. Your sales cycle decides what qualifies. With a 90-day cycle, a deal created in week six belongs to next quarter, not this one. Count deals, not just dollars. You'll need both.
Not the CRM stage percentages. Your trailing four quarters of qualified-pipeline-to-closed-won, measured. If you don't have the data, start at the market reality of roughly one win in four or five and adjust as your history accumulates.
Some share of the deals you'll win will close late. They're not lost, but for this quarter's number they're a miss. Most teams run 20% or higher and have never measured it.
An average is a promise about many quarters. You only get this one. With a finite number of deals, luck alone swings the outcome by hundreds of thousands of dollars. Subtract 1.28 standard deviations (√(deals × win rate × (1 minus win rate))) from your expected wins and you have the number you'll clear nine quarters in ten.
$9M of qualified pipeline at a $50K average deal is 180 deals. At a 25% win rate, the averages expect 45 wins, or $2.25M. Subtract 20% slip and the promise is $1.8M. Discount for variance and the bankable number is roughly $1.5M. If the target was $2.5M, a million dollars of it was never real. Better to know in week one than week thirteen.
Run your own numbers
Stop forecasting hope. Forecast a number you can defend.
The Forecast Reliability Calculator runs this entire method on your numbers: target, pipeline, and commit in, bankable number and a verdict out. It also names which of the five layers is driving your gap.
Expected value says $1.8M. The 90%-confidence number says $1.5M. Find out how much of your forecast is luck.
Run the Forecast Reliability CalculatorCoverage math
How much pipeline coverage do you actually need?
The 3x rule says three dollars of pipeline for every dollar of quota. It assumed a 33% win rate, and the market stopped paying that rate years ago. ICONIQ Growth's State of Go-to-Market 2025 put median coverage near 3.6x and flagged that at current win rates, it no longer carries the forecast.
Your real requirement isn't folklore. It's arithmetic: 1 ÷ (win rate × the share of wins that land in-quarter).
| Your win rate | Coverage you actually need |
|---|---|
| 15% | 8.3x |
| 20% | 6.3x |
| 25% | 5.0x |
| 33% | 3.8x |
Even at the legacy 33% win rate, the 3x rule fails on its own terms once slip is priced in. Add a variance buffer on top of these and the requirement rises further. The calculator computes your exact number.
One timing note. Coverage is a day-one condition, not a quarter-long average. With a 90-day cycle, the pipeline that carries Q3 has to exist before Q3 starts, which means the creation work happens a quarter earlier. Most coverage problems are calendar problems wearing a disguise.
Comparing methods
Four forecasting methods, and when each one lies
Stage-weighted pipeline. Multiply each deal by its stage probability and sum. Lies when stage definitions are vibes, which is most CRMs. A 60% stage is a feeling unless someone inspected the deal against exit criteria.
Rep commit roll-up. Ask each rep what they'll close, add it up. Lies in both directions at once: sandbagging from reps protecting their number, heroics from reps protecting their job. The errors don't cancel. They hide each other.
Historical run-rate. Last quarter times growth rate. Lies at every inflection point, which is exactly when the board most needs the forecast to be right.
The bankable number. Pipeline × win rate × in-quarter share, discounted for variance. Lies only when the inputs do, and the inputs are measurable. That's the difference. The other three methods can't tell you they're wrong. This one can.
FAQ
Frequently asked questions
How do you forecast revenue for a startup with no sales history?
Use deal-count logic with stated assumptions: pipeline, an honest win-rate guess (start near one in five), and a slip estimate. Then revisit monthly as actuals accumulate. A forecast built on named assumptions beats one built on confidence. You can correct an assumption. You can't correct a mood.
What is a good forecast accuracy?
The average organization closes 48.2% of deals as forecast, so beating that bar is easy. The better question is defensibility: can you clear your committed number nine quarters in ten? Commit at or below your bankable number and the answer is yes by construction.
What is pipeline coverage, and what ratio do I need?
Coverage is qualified pipeline divided by target. The ratio you need is 1 ÷ (win rate × in-quarter close share), not the 3x folklore. At a 25% win rate and 20% slip, that's 5x before the variance buffer.
Why do deals slip, and how do I forecast around it?
Deals slip when no one inspects close dates against actual buyer commitments. The forecast fix is to measure your slip rate and subtract it. The system fix is a deal inspection rhythm that catches drift in week two instead of week twelve.
How accurate are weighted pipeline forecasts?
As accurate as the stage definitions behind them, which is usually not very. Stage probabilities imported from CRM defaults describe someone else's funnel. Measure your own stage-to-close rates or use the bankable number method instead.
What's the difference between a forecast and a target?
A target is ambition. A forecast is math. The gap between them is the work. Teams that conflate the two end up negotiating with arithmetic, and arithmetic doesn't negotiate.
Find the layer your forecast is leaking from.
The diagnostic takes five minutes. It reads your revenue system across all five layers and tells you where the miss is being built.