How to configure a sales forecast

A forecast serves three different audiences and each one wants something slightly different. The CFO wants a number tight enough to plan cash and hiring against. The CEO wants to know whether the team will hit quota and what to do about it if not. The sales manager wants a working view of which deals need help this week. If you build one forecast and expect it to serve all three, you will produce something nobody fully trusts. Decide upfront whose question the forecast is answering first. For most B2B teams the right primary audience is sales leadership, because they will set the cadence and enforce the discipline that makes the number believable to everyone else. Write the primary purpose on a single line and pin it in the forecast documentation. "This forecast answers the question: how much will we close this quarter, with what confidence, by rep, by segment." Everything else in this configuration follows from that sentence. If you cannot write it in one line, you are not ready to configure the system yet. Go back to leadership, align on scope, then return.

• Write the forecast purpose in one sentence and share it with sales leadership for approval
• Identify the primary audience: usually the CRO or VP of Sales, with CFO and CEO as secondary consumers
• Decide the forecast period: monthly, quarterly, or both, and lock the fiscal calendar
• Document who owns the forecast call, who owns the number, and who signs off on category changes

Commit, Best Case, and Pipeline are the three load-bearing categories in almost every B2B forecast. Omit is a fourth that catches noise. Each one needs a written definition that any rep or manager can apply without judgment calls that drift over time. Commit means the rep is personally promising this deal will close in the period, with evidence a manager can verify: signed order form imminent, procurement in flight with a named contact, legal review complete. Best Case means the deal has a real path to close in the period but one or more material risks remain: a stakeholder still on vacation, pricing not yet countersigned, a security review in progress. Pipeline covers everything open that is not Commit or Best Case: real deals that will not close this period but belong on the roll-up. Omit catches the deals you need in the CRM for record-keeping but do not want polluting the forecast: closed-lost, unqualified, or dormant beyond a defined threshold. Write these definitions in plain language, pin them inside the CRM where reps see them when they categorize, and resist the urge to add a fifth category. Every category you add doubles the surface area for argument.

• Draft definitions for Commit, Best Case, Pipeline, and Omit in plain language a new hire can apply
• Convert each definition into a short checklist (three to four binary criteria) that must be true to belong in the category
• Store the definitions as a picklist value and attach the criteria as a tooltip or help text inside the CRM
• Train managers to challenge any Commit deal that is missing one or more criteria during the forecast call

Your pipeline already carries win probability per stage. The forecast layer sits on top of that math as a human judgment overlay. Decide which stages are even eligible to appear in Commit and Best Case. For most teams, nothing earlier than late-stage negotiation should ever carry a Commit flag, regardless of amount. Early-stage deals might be eligible for Best Case if the deal size justifies the attention, but default them to Pipeline. Then set the system to auto-category new opportunities based on stage, with reps and managers free to adjust upward or downward as evidence shifts. The auto-category is a starting point, not a prediction. Its job is to prevent category inflation from brand-new deals that reps mark Commit the moment they create them. Build the mapping in your CRM as a dropdown rule: stage X defaults to category Y. Document any exceptions. Review the mapping every quarter after calibration data arrives.

A forecast you cannot compare to last week is just a number on a page. The whole point of the system is to track how the categorized number moves week over week, so leaders can see slippage, hear it discussed, and act on it before it becomes a quarter miss. Configure the snapshot engine to capture the full state of every opportunity on a fixed cadence: usually Monday morning before the forecast call, Thursday evening after it, or both. The snapshot should preserve category, amount, close date, stage, next step, and the forecasting rep so you can run a week-over-week delta on each dimension. Store snapshots in a dedicated table in the CRM, not in a spreadsheet that someone emails around. Dashboards and alerts fire off that table. If your CRM supports it, let managers see the delta inline on each opportunity: "Moved from Best Case to Commit on Tuesday, moved back to Best Case on Thursday." That history is where coaching conversations get real. Teams that run without snapshots almost always discover month-end surprises they could have caught two weeks earlier.

• Pick a snapshot cadence: at minimum weekly, ideally twice weekly, aligned to the forecast-call rhythm
• Capture category, amount, close date, stage, next step, and forecasting owner in every snapshot
• Store snapshots in a dedicated system table, not an exportable spreadsheet
• Expose week-over-week delta views on the opportunity record and in the manager dashboard

The submit lock is the single most important piece of discipline in the entire forecast system. On a defined day and time each week, the forecast closes. Reps can no longer change categories, amounts, or close dates on open deals. From that moment, the number is the number. Any changes after the lock go through a logged override process: a manager has to approve the change, a reason has to be written in a required field, and the entire history is preserved on the opportunity timeline. Without a submit lock, reps quietly nudge numbers through the week until the forecast matches whatever actually happened, and the whole system becomes a backward-looking narrative instead of a forward-looking commitment. Configure the lock in the CRM so that stage, category, amount, and close date fields become read-only for individual contributors after the deadline. Grant override permission to specific manager roles only. Require a reason code from a controlled picklist (slipped, pulled forward, won early, lost, re-scoped) plus a free-text note. Keep every override visible in a dedicated audit view that leadership can scan in under five minutes before the forecast call.

• Pick the exact submit day and time; most teams use Monday morning or Friday end-of-day
• Set stage, category, amount, and close date to read-only for reps after lock; manager roles retain edit
• Require a reason code and free-text explanation on every post-lock override
• Log every change to a dedicated audit view with timestamp, actor, old value, and new value

The forecast is submitted at the rep level but reviewed at the manager level. Build a dedicated view that lets a manager walk their team in one sitting: every rep listed with their Commit, Best Case, and total Pipeline numbers, plus a drill-down into every deal in Commit and every deal above the dollar threshold in Best Case. The roll-up logic should sum rep-level numbers to team totals and let leaders slice by segment, product line, or region without rebuilding the view. Store the manager-approved number separately from the rep-submitted number. The delta between those two is the manager judgment overlay and it is a critical data point: managers who consistently cut rep forecasts and still miss have a different problem than managers who consistently raise rep forecasts and still miss. Over time this delta becomes the most accurate single predictor of quarter-end performance by team. Expose it in the leadership dashboard without naming individual managers until a coaching conversation is warranted. Transparency with discretion is the right balance.

Most forecast systems drown users in alerts nobody reads. Configure yours to fire on only four events, and make them loud. First: any Commit deal that moves backward in stage or forecast category after submit. That is a leading indicator of a slip. Second: any Commit deal whose close date pushes beyond the forecast period. That is the slip itself. Third: any Commit deal whose amount drops by more than ten percent after submit. That is scope erosion masquerading as a win. Fourth: any Best Case deal that converts to Commit after submit without a logged override reason. That is category inflation and it corrodes trust fast. Route these alerts to the opportunity owner, their manager, and a shared forecast audit channel. Do not route them by email alone; most reps have filtered sales email into a graveyard folder. Use the CRM activity feed, Slack, or a dedicated notification surface so the signal lives where the work happens. Keep the list short. If you add a fifth or sixth alert, the first four stop getting read.

• Alert on Commit deals that move backward in stage or category after submit
• Alert on Commit deals whose close date slips beyond the forecast period
• Alert on Commit deals whose amount drops more than ten percent post-submit
• Alert on Best Case to Commit conversions after submit without a logged override reason

The forecast is only as useful as the delta between the submitted number and the actual closed number. Instrument that delta from day one. Capture every weekly submit, compare it to the quarter-end actuals once the period closes, and compute three accuracy metrics per rep and per manager: Commit accuracy (submitted Commit versus closed-won from the Commit pool), Best Case contribution (percentage of Best Case that converted to closed-won), and total forecast error (absolute delta between final submitted number and actual landed revenue). Publish these numbers after the second full quarter of data, not before. The first quarter is noisy because the system is new, reps are still learning the categories, and managers are still calibrating their overrides. By quarter two you have enough signal to see which reps consistently sandbag Commit and which pad it. Share the metrics with the individuals before you share them in any aggregate view. The point is coaching, not public shaming. Teams that run this feedback loop quarterly reach sub-ten-percent forecast error inside a year. Teams that skip it stay at twenty to thirty percent indefinitely.

A forecast configuration is a living artifact. Buyer behavior changes, your product changes, segments mature, and categories drift. Write down the full configuration: category definitions, stage-to-category mapping, snapshot cadence, submit day and time, override reason codes, alert triggers, and accuracy thresholds. Store the document inside the CRM admin area where sales operations, revenue operations, and sales leadership all have access. Then schedule a quarterly review on the calendar before anything else can crowd it out. In each review, pull the last ninety days of forecast data and ask three questions: did the categories predict outcomes within the accuracy threshold, did any manager consistently over-override or under-override their reps, and did any alert trigger so often that teams started ignoring it. Adjust the configuration based on answers, not opinions. Treat the forecast system the way a product team treats a shipping product: measured, iterated, and never finished. The teams that get this right do not have a forecast tool. They have a forecast practice, and the tool is just where it runs.

• Document every configuration decision inside the CRM admin area with ownership assigned
• Schedule a recurring quarterly review on the leadership calendar before anything else fills the slot
• In each review, pull ninety days of forecast data and compare submitted versus actual at rep, manager, and team levels
• Adjust category criteria, snapshot cadence, or alert thresholds based on measured drift, not anecdote