Sales forecasting: make your CRM pipeline more reliable

Start with a sentence that leaves no room for interpretation: “As of 8 November, we estimate the value excluding tax of new contracts that will be signed between 1 and 30 November.” These dates are illustrative. This definition fixes the event, the measurement, the window and the observation date. It prevents one salesperson from discussing quotes sent while another presents expected deposits.

A signature, an invoice and a cash receipt are different events. The value of a signed contract does not, on its own, describe recognised revenue or available cash. This sales worksheet therefore replaces neither accounting records nor a cash-flow plan. If your decision concerns paying salaries or making an investment, involve the responsible people with their own schedules, rather than renaming the pipeline “cash flow”.

Set the scope as well. Are you including only new customers, renewals, contract expansions or all three? For a first version, separate movements that follow different processes. A renewal with its terms already settled and a first contract subject to a tender do not necessarily follow the same path. Adding them together can conceal a weakness in new-customer acquisition.

Finally, write down the amount convention. In the example below, each row represents the total contractual value excluding tax of a one-off project. We mix neither monthly recurring revenue with annual value nor several alternative quotes for the same project. This decision may look administrative, but it determines whether you can compare two weeks without silently changing the unit of measurement.

Before looking for a formula, check that another person can understand every important deal without calling its owner. A usable forecast rests on a few connected pieces of information, not a growing collection of mandatory fields completed in a hurry. To begin, retain the deal identifier, customer, owner, selected amount, expected signature date and forecast category.

Above all, add dated evidence and the next obstacle. “The customer is interested” explains nothing. “The manager confirmed the scope on 6 November; purchasing approval is still pending” describes a verifiable situation. You must also distinguish what the customer actually expressed from the salesperson’s interpretation. A proposal sent is not a proposal accepted, and a scheduled meeting is not an announced decision.

Document the next action, its owner and its date. That action may belong to your business or to the customer. When you do not know the next step in the purchasing process, write “to be confirmed” instead of inventing a signature deadline. Unknowns should remain visible: they are useful information for deciding what to check, not a fault to conceal.

Keep alternatives without adding them together. If a customer is choosing between a €6,000 project and a €9,000 project, create a clear convention for the amount included and record the alternative. Two proposals for the same need do not become two independent possible sales. Similarly, link a company’s contacts to the same deal when they participate in the same decision. These simple checks prevent you from building an elegant forecast on an inflated base.

To specify what permits a stage change, complement this record with verifiable CRM exit criteria. A sales stage and a forecast category remain two distinct views: one describes progress, while the other describes the expectation for a defined period.

For this method, use four mutually exclusive categories. They describe a deal’s place in the monthly forecast, not its human value or the effort the team has invested in it. At any given moment, a deal appears in only one category.

• Signed this month: the signature meeting your definition has been obtained during the period. This is the actual signed business at the observation date.
• Committed for this month: the deal is not signed, but the available evidence supports a signature within the period. The scope, decision-maker, remaining steps and timeline are explicit enough to defend this expectation. “Committed” is your internal category here, not a legal obligation on the customer.
• Additional upside: a signature this month remains plausible, but an important condition is unresolved. Name that condition and the evidence that would allow you to reclassify the deal.
• Outside this month’s forecast: the timeline is not credible or remains unknown, the deal is insufficiently qualified, or the signature belongs to another period.

Then build three cumulative totals: the baseline equals signed business; the working scenario adds signed and committed business; the high scenario adds additional upside. Do not present these totals as three baskets to add together. The high scenario already contains the other two included categories.

The baseline is not an accounting guarantee, and the high scenario is not an absolute ceiling: a new deal might emerge. These are views of the known portfolio under stated assumptions. An opportunity outside the forecast still receives sales follow-up. Excluding it from the calculation means neither abandoning it nor deciding that it has no value.

The distinction between exclusive and cumulative totals also exists in software: Salesforce’s forecasting documentation explains that, under cumulative rollups, Commit includes Closed. Check each column’s definition before adding exported totals together.

Consider a fictional services company on 8 November. It tracks only one-off projects, in euros excluding tax, with a signature targeted for November. The amounts and situations below are invented to illustrate the calculation. They do not come from a customer business or a study.

Signed business totals €8,000. Unsigned committed business totals €18,000, from €12,000 plus €6,000. Additional upside totals €14,000. The three views are therefore an €8,000 baseline, a €26,000 working scenario and a €40,000 high scenario. The entire portfolio is worth €60,000, but presenting that amount as a November forecast would be misleading. Fjord is explicitly excluded.

Now suppose Boréal signs on 15 November without an amount change. Its row moves from committed to signed. Signed business becomes €20,000 and remaining committed business becomes €6,000. The working scenario stays at €26,000: moving the deal has created no new value. If your total increases by €12,000, your mechanism is probably counting the same deal twice. This is a useful test before distributing the worksheet to the whole team.

HubSpot’s forecasting documentation defines weighted amount as a deal amount multiplied by its probability. A weighted pipeline adds these results together. In our fictional example, assign 80% to Boréal, 80% to Cèdre, 30% to Delta and 25% to Érable for demonstration purposes. Adding Atlas, which is already signed, produces a weighted total of €26,400: €8,000 + €9,600 + €4,800 + €3,000 + €1,000. Fjord remains excluded from this monthly estimate. These rates are arbitrary and are not industry benchmarks.

The similarity to the €26,000 working scenario makes neither result more certain. You will not necessarily sign 80% of Boréal. With a small number of indivisible contracts, a handful of possible outcomes can produce results far from the calculated average. One large deal can dominate the month even when the formula appears precise to the nearest euro.

Also ask what the probability measures. “Win eventually” and “sign before the end of the month” are not the same question. A very promising deal can move outside the period because a purchasing step remains open. Applying an eventual win rate to an uncertain signature date mixes two different assumptions.

If you have comparable historical data, examine observed conversions from the same starting situation and over the same horizon. Do not build a rate by looking only at won deals. Include relevant losses and delays, document the sample and remain cautious when it is small or the offer has changed. Without sound data, an explained category may be more honest than a probability presented as scientific.

A forecast review should not become a reading of the entire CRM or an interrogation in which everyone learns to announce the number their manager expects. Prepare the data before the meeting. During the discussion, focus on deals that have changed, amounts that concentrate risk and assumptions lacking evidence. The length will depend on the portfolio; the goal is a clear decision, not a timed ritual at any cost.

For each important movement, ask four questions: what have we learned since the previous review? What source supports that information? What could still move the signature? What action will reduce this uncertainty, and who will take it? “We do not know yet” should lead to verification or reclassification, not an automatic increase in sales pressure.

Separate the forecast conversation from the sales-target conversation. The target expresses what the business wants to achieve. The forecast describes what the known evidence supports expecting. If the working scenario remains below the target, the correct diagnosis may be insufficient coverage, a timing problem or dependence on a few deals. Relabelling opportunities to close the gap does not fix any of those problems.

Finish with the accepted changes and their justification. The owner or manager can challenge an assumption, request evidence or decide to help with a specific obstacle. If they impose a different estimate, retain the deal owner’s forecast and the management adjustment separately. You can then learn from each. Overwriting a single cell prevents you from understanding whether the error arose from the deal, the interpretation or internal pressure.

To address anomalies without inspecting every record, you can use a CRM exception review. Keep a specific output here, however: an updated period forecast accompanied by the reasons behind its movements.

A CRM often displays the current state. To evaluate a forecast, you need to know what you thought before the result arrived. Keep a dated snapshot containing the scope, included rows, categories, amounts and assumptions. Retain the start-of-month version, then revisions at an appropriate cadence. Do not systematically overwrite the past with the most recent information.

At the end of November, extend our fictional example: Atlas accounts for €8,000 signed, Boréal for €12,000 and an independent first phase of Érable for €2,000. Actual signed business therefore totals €22,000. Cèdre, worth €6,000, has moved to a later period; Delta has not signed and is excluded from actuals. Érable’s remaining €2,000 stays a separate opportunity requiring a new decision. The bridge from the initial working scenario is readable: €26,000 − €6,000 + €2,000 = €22,000. Actual minus forecast gives a negative €4,000 variance. The additional win partly offset the delay: the total alone would have concealed two different movements.

Reconstruct the movements: signatures included in the forecast, unforecast signatures, delays beyond the period, losses, amount changes and data corrections. Assign each variation to one main cause to avoid counting it several times. A category change without an amount or period change does not necessarily change the total.

Compare forecasts made at the same horizon. An estimate produced the day before the period closes is naturally different from one made several weeks earlier. Track absolute error amounts and the direction of errors over several comparable periods, without drawing a general rule from two months. A tendency to overestimate may indicate optimistic timelines; repeated underestimation may reveal deals entering the tracking process too late. Look for a correction to the method before looking for someone to blame.

Some opportunities deserve particular attention because they can shift the entire result. Start by identifying concentration: if the working scenario depends mainly on one contract, say so explicitly. Show the total with and without that deal, without turning this variation into another promise. The business owner then sees a real dependency instead of an apparently diversified number.

Next, examine postponed dates. An opportunity can remain legitimate after several delays, but each move should rest on a fresh explanation. Do not automatically slide every date to the final day of the following month. Ask which external event will actually enable progress: a decision, an approval, a stakeholder’s availability or completion of the scope. Without new evidence, the category should reflect the continuing uncertainty.

Handle partial contracts, options and expansions separately. If the customer signs only a first phase, the actual amount corresponds to what meets your definition of a signature. The next phase remains a separate opportunity if it requires an independent decision. A contractual ceiling or an option is not automatically secured value. When the scope of a commitment is unclear, have the relevant qualified people confirm the interpretation.

Finally, do not confuse missing information with poor individual performance. Some teams receive more complex or more recent opportunities. Comparing their raw rates without accounting for context can encourage concealment or avoidance of difficult deals. Use the review to make the portfolio understandable and improve decisions, not to create a ranking that rewards the most conservative forecasts.

The following worksheet can live in your usual tool. It gives its reader enough information to reconstruct the number without inspecting every sales conversation. Adapt the labels to your process, but preserve the distinctions between facts, assumptions and actions.

Forecast header

• Snapshot date and person responsible.
• Forecast period and event being measured.
• Included scope, exclusions and amount convention.
• Signed business during the period, unsigned committed business and additional upside, shown separately.
• Baseline, working scenario and high scenario, with a reminder that they are cumulative.
• Main concentration of risk and difference from the sales target.

One row per opportunity

Identifier; customer; owner; selected amount; expected signature date; category; evidence and evidence date; remaining condition; next action; action owner; deadline; reason for the latest change.

After the period closes

Actual result under the original definition; variance against the selected snapshot; movements explaining that variance; one rule to improve for the next period.

Before distributing it, have someone who did not take part in the discussions read the worksheet. Can they explain why one deal is committed while another is merely upside? Can they identify what is missing? If they have to guess, improve the note instead of adding a colour. Also test the move from committed to signed: the working scenario must not increase twice. This reading and calculation check is often more useful than a sophisticated presentation. Finally, protect sensitive commercial information and share the detailed view only with the people who need it.

Start with a scope you can track properly: one team, one offer or one contract type. Review opportunities that might sign during the period, check amounts and duplicates, then classify them with their evidence. The first result may feel uncomfortable: some of the pipeline will probably leave the forecast because it lacks a credible timeline. This is not an observed loss of sales, but an improvement in visibility.

Next, choose an update cadence compatible with your sales cycle. The owner updates an opportunity when significant information arrives; the review consolidates the changes. Define who may change categories and who resolves disagreements. Keep the rules simple enough for someone who missed the original meeting to apply them.

After a complete period, do not immediately try to automate every exception. Identify the dominant problem first: missing information, systematically optimistic dates, incomparable amounts or missing snapshots. Correct one mechanism and observe what changes. An additional formula does not solve a purchasing process that nobody has clarified.

Finding new accounts is a different job. SprintLead’s CRM and pipeline describes statuses, notes, history and next-action features for organising follow-up. This does not mean it calculates the scenarios presented here: keep that calculation in a separate worksheet if necessary. A list of prospects does not automatically become a predictable portfolio. Discovered accounts still need to be qualified, enter real conversations and generate documented opportunities. Keep this boundary between sourcing and forecasting clear. Your next useful action is simple: take the coming month, write its definition, then examine the five deals with the greatest influence on its result. You will see where clarification adds more value than another chart.