How to Forecast Sales Accurately (Methods and Formulas)

To forecast sales, combine your historical revenue trend with a weighted view of your current pipeline, where each deal's value is multiplied by the win probability of its stage. Add predictable repeat revenue, compare the methods, investigate any gap, and publish a range rather than a single number. Then measure accuracy every period.

Few teams get this right. Gartner found only 45% of sales leaders and sellers had high confidence in their organisation's forecasting accuracy. This guide on how to forecast sales covers the five main methods, their formulas, and a worked example that runs three of them on the same data.

# Key takeaways

• A forecast is only as good as the pipeline data behind it. Fix data first.
• Weighted pipeline is the best starting method for most deal-based businesses.
• Run at least two methods and treat disagreement as a signal to investigate.
• Publish a range (commit and best case), not a single figure.
• Track accuracy with one formula every quarter and recalibrate stage probabilities.

# The 5 main sales forecasting methods

Method	Formula	Best for	Weakness
Historical	Same period last year × (1 + growth rate)	Stable businesses with repeat revenue	Ignores what is actually in the pipeline
Weighted pipeline	Sum of (deal value × stage probability)	Deal-based B2B and service sales	Wrong if stage probabilities are guesses
Forecast categories	Closed + Commit (+ Best case for upside)	Teams with experienced reps	Relies on rep optimism or sandbagging
Sales velocity	(Opportunities × deal value × win rate) ÷ cycle length	Estimating future run-rate	Averages hide big one-off deals
AI or multivariable	Model using deal age, activity, source, history	Teams with a year or more of clean data	Opaque, needs volume and data quality

The weighted method is built into most CRMs. HubSpot, for example, calculates a weighted amount as deal amount multiplied by deal probability, and uses forecast categories named Not forecasted, Pipeline, Best case, Commit and Closed won. The velocity formula comes from Salesforce.

# How to forecast sales in 6 steps

1. Clean the pipeline. Remove dead deals, fix close dates and confirm every deal has a next step. Our guide to sales pipeline management covers the weekly review.
2. Set stage probabilities from history. For each stage, divide deals eventually won by deals that reached it over the last six to twelve months. See the 7 stages of a sales pipeline for stage definitions.
3. Separate predictable revenue. Renewals, retainers and repeat orders are forecast differently from new deals.
4. Run at least two methods. Usually historical plus weighted pipeline, and forecast categories if reps are experienced.
5. Reconcile and publish a range. Where methods disagree by more than 10%, investigate before publishing.
6. Measure accuracy afterwards. Use the formula below and adjust probabilities each quarter.

# Worked example: three methods, one quarter

Take an illustrative Nagpur packaging supplier forecasting Q4 (October to December 2026). Last year's Q4 revenue was ₹80 lakh and the business has been growing about 12% a year. Around ₹25 lakh a quarter comes from repeat orders by existing clients.

Method 1, historical: ₹80 lakh × 1.12 = ₹89.6 lakh.

Method 2, weighted pipeline: open new-business deals expected to close in Q4.

Stage	Open deal value	Stage probability	Weighted value
Demo held	₹50 lakh	20%	₹10 lakh
Proposal sent	₹60 lakh	40%	₹24 lakh
Negotiation	₹40 lakh	70%	₹28 lakh
New business subtotal	₹150 lakh		₹62 lakh

Add ₹25 lakh of repeat orders and the weighted forecast is ₹87 lakh.

Method 3, forecast categories: reps mark ₹48 lakh of deals as Commit and a further ₹22 lakh as Best case. With repeat orders, that gives ₹73 lakh commit and ₹95 lakh best case.

Reconcile. Historical (₹89.6 lakh) and weighted (₹87 lakh) agree closely, while reps' commit (₹73 lakh) is lower. That gap is worth a conversation: either reps are being cautious, or some negotiation-stage deals are weaker than the 70% probability suggests. A sensible published forecast: ₹85 lakh expected, ₹73 lakh commit, ₹95 lakh best case.

# How to measure forecast accuracy

Use one formula consistently:

Forecast accuracy = 1 − (|actual − forecast| ÷ actual)

If the supplier's actual Q4 revenue is ₹82 lakh against an ₹85 lakh forecast, accuracy is 1 − (3 ÷ 82), or about 96%. Track the direction of misses too. Consistent over-forecasting usually means stage probabilities are too generous or dead deals are not being closed.

Leadership involvement matters. In a 2024 Gartner survey, organisations where the chief sales officer directly led analytics were 2.3 times more likely to achieve higher forecast accuracy.

# Common forecasting mistakes

• Forecasting from dirty data. Gartner's 2020 research found just 47% of respondents believed their organisation maintained high-quality data. The Salesforce State of Sales 2026 found 79% of top-performing teams prioritise data hygiene, against 54% of underperformers. See why messy CRM data breaks forecasts and AI.
• Using CRM default probabilities. They are placeholders, not your win rates.
• Counting deals with no close date. If it has no date, it is not in this quarter's forecast.
• Mixing repeat and new revenue. They behave differently and hide each other's problems.
• Treating the forecast as a target. A forecast predicts; a target motivates. Mixing them encourages sandbagging.

# Where AI fits in sales forecasting in 2026

AI forecasting models look at signals beyond stage, such as days since the last buyer reply, number of stakeholders engaged or lead source. That is the same idea behind predictive lead scoring, applied to deals. An agentic CRM can also keep the inputs clean by logging calls, emails and WhatsApp chats automatically, which is often the bigger win.

The limits are practical. Models need a year or more of consistent history and enough deals to learn from. A 10-person team closing 15 deals a month will usually get more from clean data and a disciplined weighted forecast than from a black-box model. If you are new to pipelines, start with what a sales pipeline is.

Accurate sales forecasting combines weighted pipeline probabilities with historical rep closing velocity. Discount unverified verbal commitments by factoring in real decision-maker access and contractual review timelines. Comparing forecasted numbers against rolling quarterly actuals enables sales leaders to continuously calibrate weighting models, delivering forecasts that executive boards and investors can confidently trust.

# Frequently asked questions

# What is the simplest way to forecast sales?

For a small business, the simplest reliable method is a weighted pipeline forecast. Multiply each open deal's value by the win probability of its stage, add the results and add any predictable repeat revenue. Compare that number against the same period last year to sanity-check it, and you have a usable forecast in under an hour.

# How accurate should a sales forecast be?

There is no universal benchmark, but many teams aim to land within about 10% of actual revenue for a quarter. What matters more is the trend: measure accuracy every period with the same formula and try to narrow the gap. If you consistently miss in one direction, your stage probabilities or rep judgement need recalibrating.

# Which sales forecasting method is best?

No single method is best. Historical forecasting works for stable businesses with repeat revenue. Weighted pipeline suits deal-based sales. Forecast categories capture rep judgement, and AI models help once you have enough clean data. Most accurate forecasts combine two or three methods and investigate when they disagree.

# Can AI forecast sales accurately?

AI forecasting can spot patterns humans miss, such as deals that stall after a certain number of days without a reply. It needs a year or more of clean, consistent CRM data to work well. With messy data or few deals, a simple weighted pipeline plus manager review will usually beat an AI model.

# Conclusion: forecast a range, then learn from the miss

Accurate forecasting is less about clever models and more about clean data, honest stage probabilities and comparing two or three methods. Publish a range, measure the result and recalibrate every quarter.

Want forecasts built on a pipeline that updates itself? Try Autometa CRM free.

# Related reading

• How to Build a Sales Pipeline From Scratch
• Sales Pipeline vs Sales Funnel: What's the Difference?
• Why Use a CRM? 10 Benefits Backed by 2026 Data
• How AI Is Used in CRM: 10 Practical Applications

# Sources

1. Gartner Says Less Than 50% of Sales Leaders and Sellers Have High Confidence in Forecasting Accuracy — Gartner, 2020
2. Gartner Survey Finds Sales Analytics Has Less Influence on Sales Performance Than What Leadership Expected — Gartner, 2024
3. HubSpot's default deal properties — HubSpot Knowledge Base, 2026
4. How to Supercharge Your Sales Velocity for Quicker Wins — Salesforce, 2025
5. State of Sales 2026 — Salesforce, 2026