On this page
The short version
Sales forecasting is the practice of estimating the bookings a B2B sales team will close in a period, from rep judgment, stage-weighted pipeline, historical trends or a predictive model, and then scoring each estimate against what closed. RevSure, the agentic GTM platform built on the Full Funnel Data Graph, runs that work weekly through its Digital RevOps Analyst, while the revenue team makes the final forecast call.
A growth marketing leader at a compliance software company that serves financial institutions stopped reading his pipeline projection every day. The number moved between mornings, and the sales leaders it had to agree with set their number once a week: "they run their forecasting on Fridays," he told us on a September call. So he compared the two on Fridays. Asked how forecasting was going, he said: "It was pretty, it was pretty close in August too. Like it was, it lined up pretty well to theirs."
That habit holds most of what forecast accuracy measurement needs. A forecast can only be scored if someone records it at a fixed moment, compares it with the same definition of bookings, and repeats the comparison every period. This guide covers the main sales forecasting methods for B2B teams, a forecast accuracy formula with a worked example, and the checks a Digital RevOps Analyst runs each week.
What is sales forecasting?
Sales forecasting is estimating the value and timing of the deals a sales team will close in a future period, usually a month or a quarter, so leadership can set targets, plan headcount and commit a number to the board. In B2B, most of the forecast comes from open pipeline: the opportunities already in the CRM, with their stages, amounts and expected close dates.
Sales forecast. Expected bookings for a period from the current sales pipeline, usually owned by sales leadership and rolled up from reps and managers.
Revenue forecast. Expected revenue the company will recognize, which adds renewals, expansion, churn and the timing rules finance applies to bookings. Revenue forecasting usually sits with finance and covers a year or more.
Pipeline projection. An estimate of the pipeline still to be created and how it will convert. It matters because next quarter's bookings often depend on deals that do not exist yet. Our guide to what a sales pipeline is covers the stages and metrics underneath all three.
What are the main sales forecasting methods for B2B?
Most B2B teams run two or three methods at once and compare them, because each one reads different data and fails in a different way.
The rep roll-up is the method most teams already run, and our post on the Forecast Variance agent covers why a commit category is a rep's belief rather than a probability. The other methods fail in ways that are easier to miss, and three of them showed up on recent customer calls.
Stage-weighted forecasts are only as good as the stage map
At an autonomous security testing company, the marketing operations team found that the proof-of-value stage sat in the early bucket of one executive dashboard, while a dashboard built for the CMO treated it as late stage. The team's operations lead put the question directly on a July call: "why do we keep POV in early stage if our win rate from POV is like 70%?"
A stage-weighted forecast built on that map would count every proof-of-value deal at an early-stage probability and understate the quarter. Stage weights hold up when each one comes from the company's own stage-to-close history, by segment, and gets recomputed every quarter. Our piece on why traditional forecasts fail before the quarter starts covers the wider system those stage rules sit in.
Time-series models miss what the history does not contain
A data and analytics leader at a cybersecurity company described the ARIMA-style time-series models his team had used to forecast pipeline. When pipeline started falling, "the models weren't right because it didn't have context," he said on a June call with us.
A trend line reads past bookings. It cannot see that a segment's new pipeline dried up eight weeks ago, or that a large campaign moved into next quarter. Time-series methods work as a planning baseline and fail as the only number in a quarter where something changed.
Marketing forecasts need the plan, not only the history
When a new revenue leader arrived at a security awareness training company, the marketing performance lead got a direct request, which he described on a September call: "we've got a new CRO or Chief Growth Officer who's come in and quite rightly said, marketing, I need a bit more from you here. And one of the things I want to see is a marketing forecast."
Working through how to build it, he named the gap in a history-only approach: "the data that's missing though from this is knowing which campaigns we're running in future." His team already mapped its campaigns, with budgets and expected lead counts, in a project management tool. A driver-based forecast takes exactly those inputs, campaign dates, budgets and expected volume, and runs them through the company's own conversion rates to produce pipeline and bookings by week.
How do you measure forecast accuracy?
Forecast accuracy measures how close a forecast, recorded at a fixed point in time, came to the actual result for the same period. It is usually expressed as a percentage:
Forecast accuracy (%) = (1 - |Actual - Forecast| / Actual) x 100
The vertical bars mean the absolute value, so a forecast that misses high and one that misses low by the same amount score the same. Two companion measures keep that from hiding a pattern:
Forecast bias = Forecast - Actual. A positive number means the team over-forecast; a negative number means it under-forecast. Accuracy tells a team how far off it was, and bias tells it which way it tends to miss.
Mean absolute percentage error (MAPE) = the average of each period's absolute percentage error. It scores a series of months or segments together, so one good quarter does not cover for three bad months.
A worked forecast accuracy example
Take an illustrative company that records its forecast for each month of the third quarter on the first Friday of that month, then compares it with closed-won bookings once the month ends.
July shows the formula at work: |$1.00M - $1.20M| / $1.00M = 20.0 percent error, so accuracy is 80.0 percent, and the bias of +$0.20M says the miss was high.
The quarter row is the trap. At the quarterly level the forecast was 97.2 percent accurate, because July's overcall and August's undercall cancel each other out. Measured month by month, the same forecasts averaged 11.9 percent error. A board sees the 97 percent. The sales team, which had to staff and pace against each month, lived with the 12 percent. Score accuracy at the level where decisions are made, and keep bias next to it so a team that always calls high can see that it does.
Rules that keep the score honest
Freeze the forecast at a fixed point. Record the number at the same moment every period, such as the Friday forecast call or the first Friday of the month, and never compare a live number with a final one. Snapshots make this possible. A marketing operations lead at a healthcare identity security company explained on a September call that "the revenue operations team does primarily their reporting based on snapshot reporting," with a daily snapshot of opportunities, which lets any forecast be scored against the pipeline as it stood on the day it was made.
Use one definition of bookings on both sides. New business and renewals, booked date and close date, currency and segment have to match between the forecast and the actual, or the error measures the definitions instead of the forecast.
Compare every method against a simple baseline. Score each method against something naive, such as last quarter's bookings or the plain stage-weighted number. A method that cannot beat the baseline over several periods has not earned its place in the call.
Keep the misses next to the method. Over a few quarters the record shows which method to trust in which segment and month, which is worth more than any single accurate quarter.
What does a Digital RevOps Analyst check every week?
A Digital RevOps Analyst is one of our Digital Workers: an AI-driven RevOps role that our GTM engineers run on top of the context layer. Its forecasting job is to model, predict and continuously monitor the forecast, and to identify risk and upside across the quarter. The deliverables for that job are a forecast model, a recommended forecast call and a scenario view.
A weekly forecast review covers seven checks, prepared before the forecast call and reviewed by a RevOps lead.
Last period's score. Accuracy and bias of the frozen forecast against closed bookings, by segment and month, and against the baseline.
Coverage against the remaining target. Open pipeline by segment against the gap still to close, read against the coverage ratios that produced the target in past quarters. Our guide to pipeline coverage ratios covers how to set them.
Stage conversion against the weights. Whether recent stage-to-close rates still match the probabilities the forecast uses, the problem behind the proof-of-value question above.
Aging and slipped close dates. Deals that have been open longer than similar won deals took, and close dates pushed more than once.
Month-level pacing. Whether the quarter depends on its final month. The Monthly Projections view in Booking Readiness breaks projected bookings into month-level contributions for this reason.
The system number against the sales call. The gap between the model's projection and the sales leaders' forecast, with the deals and segments that explain it.
Data that would move the number. Missing amounts, stale stages and unmatched records, which belong in the standing CRM hygiene routine before they reach the forecast.
The same role covers related work under other jobs: monitoring pipeline creation, progression, coverage, aging, and marketing and sales contribution by segment; a stage conversion report from funnel analysis; and anomaly alerts with a root-cause note. People still decide. Every recommendation requires a human to approve, adjust or override it before it is acted on, and the RevOps leader and analysts set strategy, own the data model and forecast call, and approve exceptions. A Digital RevOps Analyst never runs unsupervised, and it is not your team's job to supervise it: our GTM engineers manage the role day to day. Our guide to revenue operations describes how that split changes the rest of the RevOps week.
Where does our forecast come from?
Every method in the table reads the same raw material: stages, amounts, dates, engagement and the identity of the account behind each deal. We bring data from the CRM, marketing automation, ad platforms and engagement sources into the Full Funnel Data Graph, resolve the people and accounts, and apply one set of definitions for stages and handoffs. The context layer on top gives the Digital RevOps Analyst and every agent the same pipeline and the same stage rules the forecast is scored on.
Our Pipeline Projections product shows pipeline across daily, weekly, monthly, quarterly and annual views, with daily updates, funnel-stage conversion visibility and risk alerts.
At the compliance software company from the opening, booking projections had come from a manual spreadsheet scorecard and moved whenever reps changed deal stages. We deployed booking-volume and booking-value forecasts based on the actual pipeline rather than on stage hygiene. Our monthly pipeline projections went live there in July 2026 and reached 96 percent booking forecast accuracy in our measurement, and by September the growth marketing leader was checking them against the sales leaders' Friday number.
A forecast is a claim made on a date
A forecast is a statement about the future made at a specific moment, and it can only be judged against that moment. The compliance software team's comparison had a fixed day, one definition of bookings and a second number to check against, which is why "pretty close" meant something. Teams that skip those three things can still produce a forecast every week. What they cannot produce is a record of which method was right, in which segment, by how much, and that record is what decides whose number the board believes next quarter.
FAQs
What are the main sales forecasting methods?
The main B2B sales forecasting methods are rep judgment roll-ups, stage-weighted pipeline, historical trend or time-series forecasting, opportunity-age forecasting, pipeline coverage, driver-based bottom-up forecasting and predictive models. Most teams combine two or three, using the rep roll-up for the final call and a stage-weighted or predictive number to check it, with the gap explained deal by deal.
What is the difference between sales forecasting and revenue forecasting?
Sales forecasting estimates the bookings the sales team will close in a period, mostly from open pipeline. Revenue forecasting estimates the revenue the company will recognize, adding renewals, expansion, churn and the timing rules finance applies to bookings. Sales forecasts are usually monthly or quarterly and owned by sales leadership; revenue forecasts usually sit with finance and cover a year or more.
How do you calculate forecast accuracy?
Forecast accuracy is 1 minus the absolute difference between actual and forecast, divided by actual, times 100. A forecast of $1.2M against $1.0M actual has 20 percent error and 80 percent accuracy. Record the forecast at a fixed point each period, track bias (forecast minus actual) alongside it, and use MAPE to score several months or segments together.
What is a good forecast accuracy for a B2B sales team?
There is no universal benchmark, because accuracy depends on deal size, cycle length, the period measured and how early the forecast is frozen. A useful standard is internal: beat a simple baseline such as last quarter's bookings or the stage-weighted number over several periods, keep bias near zero, and score at the monthly and segment level, not only the quarter.
How often should a B2B team update its sales forecast?
Most B2B teams set the forecast weekly, on a fixed day tied to the forecast call, and review it formally each month and quarter. Models can update daily, but accuracy should be scored against the weekly or monthly snapshot that leaders actually committed to, so the measurement matches the number people used to make decisions.
What does a Digital RevOps Analyst do for forecasting?
The Digital RevOps Analyst is a RevSure Digital Worker whose forecasting job is to model, predict and continuously monitor the forecast and identify risk and upside across the quarter. It delivers a forecast model, a recommended forecast call and a scenario view. Our GTM engineers manage it day to day, and the revenue team makes the final forecast call and board-level commitment.