Forecast the decision.
Show the assumptions.

What is a sales forecast?

A sales forecast estimates a defined commercial outcome over a defined period. For a services team, that might be new signed contract value this quarter. For another business, it might be first-year subscription bookings. The definition should tell someone which deals count, when they count and which value field is used.

A sales target is an ambition. A forecast is a current assessment. Open pipeline value is the value of opportunities in scope. These can all appear in the same meeting, but they answer different questions. Raising a forecast to match a target without new evidence makes the number less useful.

This guide describes an operational forecasting framework, with hypothetical examples. It is not an accounting policy or a claim that a particular forecasting screen is available in every CRM plan. Keep bookings, recognised revenue and cash collection distinct, and use the appropriate financial process when those are the measures you need.

Set the period, amount and inclusion rules

Choose a month or quarter and a reporting cutoff. Record the forecast date so the reader knows which evidence was available. A forecast submitted at the start of a quarter and one submitted on its final day should not be evaluated as though they had the same information.

Choose one amount basis: total contract value, first-year value or another documented measure. Exclude amounts that do not belong to that basis, and do not mix different currencies without an explicit conversion rule. Keep unknown amounts visible instead of silently turning them into zero.

Define the commitment event used for the outcome. It may be an executed agreement or an accepted order. The expected decision date is the current estimate; the actual close date records what happened. Preserve both meanings. Moving an expected date does not change historical outcomes.

Also decide how renewals, expansions and mutually exclusive proposals are represented. Two options offered for the same project cannot both be assumed to close unless that is commercially possible. Group linked alternatives and state which one each scenario includes.

Choose a method suited to the available evidence

A deal-by-deal forecast starts with the actual opportunities expected to decide within the period. Owners explain the proposed value, buyer process, remaining steps and reason for inclusion. This can be useful for a smaller set of considered purchases, but it depends on the quality and consistency of those judgements.

A stage-weighted view multiplies deal amounts by assigned probabilities. It provides a repeatable calculation, but default probabilities are not evidence of your business’s outcomes. The same stage may contain deals with very different timing, dependencies or buying processes. A weighted amount also does not establish that those deals will close in this period.

A historical model can use observed patterns from comparable opportunities. It needs reliable history, stable definitions and enough relevant examples to evaluate. A new process, new market or small sample can limit its usefulness. Keep a simple baseline so added modelling complexity has something to improve upon.

You can combine methods without pretending they are interchangeable. Use an aggregate view to expose the broad picture and inspect the individual deals that drive the result. HubSpot’s setup documentation distinguishes weighted and total amounts, forecast periods and forecast categories [1]. Check the exact behaviour and plan in whichever product you use.

Use forecast categories to express period-specific judgement

A stage describes where the opportunity stands in the process. A forecast category describes the current judgement about its inclusion or likelihood within the selected period. A late-stage deal can still fall outside this quarter if procurement cannot finish in time. Keeping those ideas separate reduces the temptation to move stages merely to change a forecast.

Write the evidence required for each category in your own operating language. A committed deal might need an agreed scope, a known approver and a credible remaining decision path. An upside deal may be possible but depend on an unresolved event. These are example definitions to adapt, not universal labels with standard probabilities.

Let the owner record uncertainty directly. If the decision date is unconfirmed, say so and identify the next step needed to clarify it. Record manager overrides with a reason rather than replacing the owner’s judgement invisibly. A forecast should preserve why the number changed.

Worked example: three scenarios, one shared dataset

Consider a fictional team forecasting new bookings for a quarter. It already has £20,000 closed won. Deal A is £15,000 and has a credible remaining decision path. Deal B is £10,000 but depends on a budget approval. Deal C is £25,000 and depends on an additional procurement decision. All values use the same amount basis and currency.

The downside scenario assumes no further deals close: £20,000. The base scenario includes the amount already won plus Deal A: £35,000. The upside scenario includes the amount already won and all three open deals: £70,000. These are conditional scenarios, not statistical confidence intervals or guarantees.

The scenarios are alternative totals. Do not add £20,000, £35,000 and £70,000 together. Within each scenario, each opportunity appears once. If Deals B and C were mutually exclusive alternatives, the upside would need to reflect that constraint rather than including both.

Now give each uncertainty an owner. Who will confirm the budget decision for B? What evidence would make C plausible within the quarter? This turns the range into a plan for reducing uncertainty. A colourful range without those dependencies is only a different way to display guesses.

Review the changes, not just the new total

At each review, preserve the previous forecast before editing the current one. Record which opportunities were added, removed, delayed, won or lost, and which amounts changed. A stable total can conceal one deal slipping out while an unrelated deal moves in. That change may matter even when the headline number does not move.

Ask owners for buyer evidence supporting the expected date. A month-end date chosen to fill a field is different from a scheduled approval meeting. Inspect the remaining steps and external dependencies rather than relying on an owner’s confidence alone.

Use a consistent review rhythm, with additional updates when significant facts change. Keep the forecast submission timestamp and author. If an opportunity moves beyond the forecast period, update its treatment and preserve the reason. Do not keep it inside the total solely because removing it creates an uncomfortable gap.

The review should end with a stated assessment and a list of unresolved conditions. Connect those conditions to owned next actions, while keeping the target visible as a separate reference. The gap between forecast and target may motivate action; it should not be hidden by changing the assumptions.

Compare like-for-like forecasts with actual outcomes

Choose a fixed forecast horizon for evaluation. For example, compare the forecast submitted four weeks before period end with the actual result for that period. Mixing early submissions with final-day estimates can make a method look more accurate simply because some forecasts were made later.

If a hypothetical forecast is £50,000 and the actual outcome is £40,000, the signed error defined as forecast minus actual is +£10,000: an overestimate. The absolute error is £10,000. If you divide that absolute error by actual, the percentage is 25%. State that denominator; other conventions answer different questions.

When actual is zero, that percentage is undefined. Report the absolute error and explain the zero outcome rather than displaying an infinite percentage. Keep several periods visible so one unusual deal does not dominate the conclusion. Examine repeated overestimation or underestimation alongside the magnitude of errors.

Use the review to improve definitions and judgement. Was the amount wrong, the timing wrong or the inclusion assumption wrong? Did a dependency remain unverified? A useful error analysis explains what the team should do differently next time, rather than merely grading the final number.

Build the first version with explicit assumptions

Start with a defined period, a clean list of eligible opportunities and a small set of scenarios. Record the amount basis and currency, known outcomes, expected decisions and dependencies. Assign an owner for the forecast and for the evidence needed to refine it.

Download the worksheet below to organise those decisions. It contains the illustrative scenario dataset and columns for replacing it with your own. It is a planning CSV, not an automatic forecasting model; review inclusions and calculate totals according to your stated rules.

Once the process is consistent, decide what to automate. Reliable field definitions and preserved snapshots are more useful than a complex model fed by inconsistent records. Improve the forecast when the evidence supports a better method, and keep its limitations visible to the people using it.

Common questions.