Sales Forecasting Methods, Step by Step

Sales forecasting comes down to five steps: define your sales process and forecast period, gather your historical deal data, choose a forecasting method that fits your data, build the number, and review it against reality on a fixed cadence. The method you choose matters less than most teams think. The discipline of the five steps matters more. This guide walks through each step and the six most common sales forecasting methods, with a worked example you can copy.

Key takeaways

• Forecasting has five steps: define, gather, choose a method, build, review.
• Pick the method your data can support, not the most sophisticated one.
• Stage-weighted pipeline forecasting is the default for most B2B teams.
• A forecast is never done. Review it weekly against what actually closed.
• Clean CRM data beats a clever method every time.

Step 1: Define your sales process and forecast period

A forecast is a prediction about a process. If the process is fuzzy, the prediction will be too. Before you touch a spreadsheet, agree on two things. First, the stages of your pipeline and what has to be true for a deal to enter each one. If “Proposal” means a document was sent for one rep and a handshake happened for another, your stage data is fiction. Second, the period you are forecasting: this month, this quarter, or both. Most B2B teams forecast the quarter and update monthly or weekly.

Step 2: Gather your historical data

Pull at least four quarters of deal history if you have it: deals won and lost, deal size, how long each deal spent in each stage, and win rates by stage. This is also the moment to be honest about data quality. Deals with no close date, stages that were never updated, and duplicate opportunities will quietly poison every method below. If your CRM history is thin or messy, start collecting clean data now and lean on the simpler methods until you have a year of history you trust.

Step 3: Choose your forecasting method

Here are the six methods most sales teams actually use, and when each one fits.

1. Stage-weighted pipeline forecast

Multiply each open deal by the win probability of its stage, then add them up. If deals in Negotiation close 60% of the time, a $50,000 deal in Negotiation contributes $30,000. This is the default method for B2B teams with a defined pipeline. Its weakness: it treats every deal in a stage the same, whether it is moving fast or has been stuck for months.

2. Length-of-cycle forecast

Score each deal by how far it is through your average sales cycle. A deal 60 days into a typical 90-day cycle gets a higher probability than one that is 10 days in, regardless of stage. Useful when reps are inconsistent about updating stages, because it relies on dates instead.

3. Historical run-rate forecast

Take what you sold in the same period last year, adjust for growth, and use it as a floor or a sanity check. Fast and simple. It ignores your current pipeline entirely, so use it as a cross-check on another method, not on its own.

4. Time series forecast (moving average and smoothing)

Average your sales over recent periods and project the trend forward, giving more weight to recent months if your market moves fast. Fits businesses with steady, repeatable sales patterns and enough history to see a trend. Struggles with lumpy enterprise pipelines where one deal can swing the quarter.

5. Regression forecast

Model how one variable drives another, like how discovery calls booked this month predict revenue next quarter. Powerful when you have a lot of clean history and someone comfortable with the analysis. Overkill for most teams under 50 reps.

6. Committee or intuitive forecast

Managers and reps call the number deal by deal, based on judgment. Every team does some of this, and it works better than its reputation when the people calling the number inspect deals weekly. It breaks when optimism goes unchallenged. Use it to adjust a data-driven method, not to replace one.

The six methods at a glance

Method	Best for	Main weakness
Stage-weighted	B2B teams with a defined pipeline	Treats all deals in a stage the same
Length-of-cycle	Teams with unreliable stage updates	Assumes deals move at an average pace
Historical run-rate	Sanity checks and stable businesses	Ignores the current pipeline
Time series	Steady, high-volume sales patterns	Breaks on lumpy enterprise deals
Regression	Data-rich teams with analyst support	Needs lots of clean history
Committee / intuitive	Adjusting a data-driven number	Optimism goes unchallenged

Step 4: Build the forecast (worked example)

Here is a stage-weighted forecast for an illustrative pipeline. Say your team closes 20% of deals from Discovery, 40% from Proposal, and 60% from Negotiation, based on your own history. Your open pipeline for the quarter:

• Discovery: $400,000 in open deals × 20% = $80,000
• Proposal: $250,000 × 40% = $100,000
• Negotiation: $150,000 × 60% = $90,000

Forecast: $270,000, plus anything already closed this quarter. Then cross-check it against your run rate. If you closed $310,000 in the same quarter last year and nothing big changed, ask why the pipeline says less. Maybe the pipeline is light. Maybe last year had a one-off mega deal. The gap between two methods is where the useful questions live. Finish by having managers adjust deal by deal, on the record: every override should name a reason.

Step 5: Review and update on a fixed cadence

A forecast is a living number. Review it weekly: what closed, what slipped, what changed stage, and how the number moved. Track your forecast error at the end of each period, meaning the gap between the final forecast and what actually closed. If you miss by more than 10% repeatedly, the cause is almost always upstream in the data: stale stages, missing close dates, or deals that should have been marked lost months ago. Fix the inputs before you blame the method.

Where AI fits

Everything above works in a spreadsheet. AI-assisted forecasting automates the heavy parts: it scores each deal individually from its behavior instead of applying one percentage per stage, recalculates continuously, and flags deals where the data disagrees with the rep’s confidence. Think of it as method number one, upgraded deal by deal. If you want the full picture of how that works and how to roll it out, read the complete guide to AI sales forecasting on Coevera: AI Sales Forecasting: The Complete 2026 Guide. The principle stays the same either way: the forecast is only as good as the pipeline data underneath it.

Common forecasting mistakes

• Using list-price stage percentages. Your win rates by stage should come from your own closed deals, not a template.
• Forecasting from a dirty pipeline. Dead deals sitting in Negotiation inflate every method.
• Changing the method every quarter. You can only measure forecast error if the method holds still.
• Treating the forecast call as a reporting ritual. Its real job is deal inspection: the number is a byproduct.
• Ignoring the misses. Every miss is a lesson about your data or your process. Write it down.

FAQ