Why this matters
A sales forecast is not a number you report. It is a set of decisions about hiring, cash, delivery capacity and next quarter's plan. If leaders cannot see why the number moved, they stop trusting it and build private versions in spreadsheets. How to build a sales forecast you can trust fixes that by tying every deal in the forecast to evidence and a written rule.
Trust does not mean perfect accuracy. A good forecast is honest about uncertainty and explains its misses. It separates what is committed, what is likely and what is early. It records what changed since the last review. Leaders can act on it even when the total is wrong.
A forecast built on optimizm fails in predictable ways. Deals sit in late stages with no next step. Close dates roll forward quietly. Reps count deals that have not spoken to a buyer in weeks. The fix is operational, not mathematical. You need clean categories, one source of truth and a review rhythm that rewards early bad news.
Before you start
Decide which decision the forecast serves. Cash planning needs a shorter horizon than hiring or board reporting. Pick one primary horizon and one secondary. Pick the unit: new bookings, recurring revenue, units or delivered work. Do not blend them into one number.
Separate new business from renewals, expansion and one-time services. Each has different evidence. A renewal has usage and a contract date. New business has a buyer, a problem, a budget path and a next step. If you blend them, a strong renewal base can hide weak new pipeline.
Write category definitions before anyone edits a deal. Commit means the buyer has agreed commercial terms, a signature path and a date. Best case means open items are known and owned. Pipeline means early. Keep the categories few. Set rules for what enters the forecast and when a deal leaves it. Name one forecast owner and one weekly review.
How to do it
Build the forecast from deal records, not from a separate spreadsheet. The CRM should store the forecast category, close date, next step and a short note on evidence. The forecast is a view over those fields.
Review changes, not every deal. The weekly call should focus on deals that moved category, moved close date, changed amount or lost their next step. That keeps the meeting short and surfaces the deals that actually change the number.
Keep a snapshot every week. A saved table of deal, category, amount, close date and owner lets you compare promises with outcomes. Without snapshots, every forecast discussion becomes memory versus memory.
- Name the decision and horizon. Write the question the forecast answers and the date range it covers. Keep that scope fixed for at least a quarter so comparisons mean something.
- Write category rules. Define commit, best case and pipeline with observable evidence. State who can change a category and what note is required.
- Add forecast fields to the deal record. Store category, close date, next step date, evidence note and change reason. Make them required before a deal appears in the forecast view.
- Set close date discipline. A close date is the date you expect a signature, not the last day of the quarter. Require a buyer event before a deal enters commit.
- Build a weekly change list. Pull deals whose category, amount or close date changed since the last snapshot. Review those first, then new deals entering best case.
- Run the call by exception. Ask what changed, what evidence supports the new position and what must happen next. Do not re-forecast every deal from scratch.
- Snapshot and compare. Save the week's forecast, then compare it with actual closed deals. Tag each miss with a reason: slipped buyer, lost to competitor, no decision, bad qualification or pricing issue.
- Publish the variance note. Send a short note with the number, the change from last week and the top reasons. Keep it factual, not defensive.
How to check it worked
Backtest. Rebuild last quarter's forecast from old snapshots and compare it with actuals. Check whether commit deals closed in the committed month. Check whether best case deals converted at a stable rate. You are looking for repeatable behavior, not one good week.
Check forecast hygiene. Can you explain the number from the CRM in minutes? Does every commit deal have a next step and a buyer event? Are close dates moving late in the month? Is the same deal sitting in best case for several weeks with no change? These are trust signals.
Check decision use. Are finance and delivery using the forecast to plan? Do they ask fewer emergency questions? Does the sales leader bring the same number to two meetings? If people still keep private spreadsheets, the forecast is not trusted yet.
Mistakes to avoid
Mixing categories. Do not let a rep call a deal commit because the buyer is friendly. Commit requires commercial agreement and a signature path. Best case requires a known open item and an owner. If definitions are loose, the forecast becomes a wish list.
Forecasting from spreadsheets. A separate file drifts from the CRM and hides changes. If you must use a spreadsheet, import it back and reconcile weekly. Also avoid changing category definitions mid-quarter, adding many categories and weighting pipeline before you have history. Weighted pipeline looks precise but can hide bad evidence.
Ignoring slipped deals. A deal that moves from this month to next month is a forecast miss, even if it is still alive. Tag it and review why. Otherwise the same deals roll forward forever and the forecast never learns.
If your team tracks deals in inboxes or a shared sheet, one option is the Cloudverg Sales CRM implementation at $9,500 fixed price. It sets up stages with exit criteria, a deal model, quote templates, follow up sequences, handoff to delivery or finance, and forecast, win rate and source reporting. Compare it with building your own process in a general CRM or hiring an internal admin. Pick the path that gets reliable deal evidence into one place fastest.
Frequently asked questions
What is the difference between commit and best case in a sales forecast?
Commit means the buyer has agreed terms and there is a clear signature path within the period. Best case means the deal is real but an open item remains, such as a security review or final budget approval. Pipeline is earlier and should not be counted as likely revenue.
How often should I update a sales forecast?
Update weekly, with a saved snapshot each time. Daily edits create noise and monthly reviews hide problems. The weekly cadence fits most sales cycles and gives you enough history to compare promises with outcomes.
Why does my sales forecast keep changing?
Most changes come from loose category rules, close dates set to quarter end and deals entering the forecast without buyer evidence. Fix the definitions, require a next step and review changes by exception. Snapshots show whether the changes are real deal movement or reporting noise.