Ask five sales leaders how they forecast and you’ll get five different answers. Some roll up rep-submitted commits. Some run regression models. Some rely on AI-generated deal scores. Some still use a spreadsheet and instinct.
The method you choose — or more accurately, the combination you build — has a direct impact on how accurate your forecasts are, how much time your team spends on them, and how much confidence leadership can actually place in the number.
Here’s a plain-language comparison of the six most common sales forecasting methods: what each one gets right, where it breaks down, and when it’s worth using.
At a glance
1. Stage-based (funnel) forecasting
How it works
Each deal stage in your CRM is assigned a fixed probability of closing — 10% at Stage 1, 50% at Stage 3, 90% at Stage 5. The forecast is the sum of (deal value × stage probability) across all open opportunities.
What it gets right
Simple, consistent, and already built into every CRM. Sales leaders can see the forecast without learning a new tool. A reasonable starting point for teams early in their forecasting maturity.
Where it breaks down
Stage is a proxy for deal progress, not a measure of it. A deal sitting in Stage 3 for 45 days with no buyer engagement carries the same probability as one that moved into Stage 3 this week with three stakeholders actively involved. The model treats them identically. It’s also entirely dependent on reps keeping stage data current — which, in most organizations, they don’t.
2. Historical forecasting
How it works
Uses past performance — historical win rates, average deal size, average sales cycle length — to project what’s likely to close in a future period. If you’ve historically closed 30% of Stage 3 deals, you apply that rate to your current Stage 3 pipeline.
What it gets right
Grounded in actual outcomes, not theoretical stage probabilities. For stable businesses with consistent deal patterns, it provides a reliable baseline.
Where it breaks down
It tells you what happened, not what’s happening. Historical rates can’t account for changes in your market, product, team, or the individual deals currently in flight. Outliers distort it fast — a few unusually large wins or losses can skew the entire baseline.
3. Rep-submitted (bottom-up) forecasting
How it works
Reps categorize their deals into buckets — commit, best case, pipeline — and roll those numbers up through the management chain. Each manager reviews, adjusts based on their own judgment, and passes the number up.
What it gets right
It captures qualitative context that data alone misses. The rep who knows a deal is at risk because of a relationship issue. The manager who’s seen this pattern before. It also creates accountability: reps are on the hook for what they commit.
Where it breaks down
Highly susceptible to human bias in both directions. Reps sandbag to manage expectations or inflate to look good. Managers adjust on gut feel rather than evidence. By the time the number reaches the CRO, it can be five layers of subjectivity stacked on top of each other.
It’s also expensive in time. The weekly forecast call ritual is largely built around gathering and reconciling these submissions. Most organizations spend more time on the forecast process than on fixing the deals inside it.
4. Deal-by-deal (opportunity) forecasting
How it works
A detailed review of each individual deal — evaluating deal-specific factors like qualification completeness, stakeholder engagement, champion strength, and competitive position to assign a close probability.
What it gets right
The most accurate method for large, complex deals where aggregate statistics don’t apply. A $2M enterprise deal has enough at stake to warrant individual inspection. It also forces the coaching conversation — you can’t discuss what to do about a deal unless you understand it.
Where it breaks down
Labor-intensive and doesn’t scale. A team with 50 open opportunities can’t review every deal in depth every week. And it’s still susceptible to the same subjectivity problem as rep-submitted forecasting if the evaluation isn’t grounded in data — deal-by-deal review based on rep narrative is still just rep narrative with more steps.