Salesforce forecasting: how it works, and how to make the number honest

By Giorgio Bonan. Updated October 22, 2026, published October 22, 2026.

Salesforce forecasting, called Pipeline Forecasting in Salesforce's current documentation, adds up open opportunities by forecast category and by period, rolls the totals up the role hierarchy, and lets each manager adjust the number of the people below them. It is switched on by default in every org created since Winter '24. What it does not do is ask whether a deal in Commit is still alive, and that is where most forecasts go wrong.

This article covers what the feature does out of the box, how forecast categories roll up, the settings that quietly delete data when you change them, and a worked month where the same ten deals give three different forecasts.

What Salesforce forecasting does by default

Salesforce's Pipeline Forecasting Implementation Guide (Version 1, Winter '27, last updated 22 July 2026) lists the defaults an org gets when forecasting is enabled. For orgs created in Winter '24 or later, this happens automatically; older orgs get the same defaults when an admin turns forecasting on in Forecasts Settings.

Two numbers in that table matter more than they look. First, the forecast is built on the role hierarchy: a rep who sits in no role, or in the wrong branch, does not roll up to the right manager. Second, the period is monthly. Many sales teams run on quarters, and moving from monthly to quarterly later has a cost covered below.

You can have up to four forecast types active at once, built on Opportunity, Opportunity Product, Opportunity Split or Line Item Schedule. Most teams need one: Opportunity Amount by close date. A second type on Opportunity Product helps when product families carry separate quotas.

How forecast categories roll up

Every opportunity carries a Forecast Category. The guide names five standard categories: Pipeline, Best Case, Commit, Omitted (not included in forecasts) and Closed. An admin can add a sixth, Most Likely, which appears between Commit and Best Case in Lightning Experience.

The category is not chosen freely: it comes from the stage. Each value of the Stage picklist maps to a category, and changing the stage sets the category. Reps can still override the category on the record if the page layout shows the field.

Then comes the rollup method, and the default is cumulative. With cumulative rollups, each column includes its own category plus every category closer to the win. The guide's example uses four Pipeline, three Best Case, two Commit and one Closed opportunity, each worth $50:

With single category rollups, each column shows only its own category ($200 Pipeline, $150 Best Case, $100 Commit, $50 Closed in the same example). Cumulative is easier to read in a forecast call, because "Commit" already contains what has closed.

The default stage mapping is the first thing to check

We read the stage settings of a Salesforce Developer Edition org through the API on 22 October 2026. Every open stage, from Prospecting to Negotiation/Review, mapped to the Pipeline category. Closed Won mapped to Closed, Closed Lost to Omitted.

The consequence: in an org left with this mapping, the Commit and Best Case columns stay empty unless reps set the category by hand on each deal. Your org may differ, since admins usually remap stages during setup. Check it in Object Manager, Opportunity, Fields, Stage, under Opportunity Stages Picklist Values. Our guide to Salesforce opportunity stages gives a mapping that matches stages to exit criteria.

Settings that delete data when you change them

The implementation guide carries two warnings that are easy to miss when you tune forecasting after go-live:

• Changing the period or the fiscal year. Moving from monthly to quarterly, or back, or changing the standard fiscal year, purges all adjustments, manager judgments and quotas. Enabling custom fiscal years deletes quotas and adjustments in the matching and later standard fiscal years.
• Changing the rollup method. Switching between single category and cumulative rollups deletes adjustments for all active forecast types.

Decide the period, the fiscal year and the rollup method before you load quotas, and export adjustments before any of those changes.

Why the number is still wrong

Configured well, Salesforce forecasting is a faithful adder. The difficulty is what it adds. It sums the Amount of every opportunity in a category, whatever happened to that deal since the last call. A deal in Commit with no email, call or meeting logged for six weeks weighs exactly as much as a deal whose contract is with legal today.

Confidence in the result is low across the market. In a Gartner survey published in February 2020, fewer than half of sales leaders and sellers had high confidence in their organization's forecasting accuracy. Data entry is part of the problem: in Salesforce's own research published in December 2022, reps spent just 28% of their week actually selling, with deal management and data entry taking a large share of the rest. Fields that cost time to update are the ones that go stale first: close date, next step, and the category itself.

A worked month: one pipeline, three forecasts

Here is one rep's month, ten open deals closing this month, with stage probabilities from the table above. The last column is the number of days since the last activity (task or event) logged on the deal.

Three ways to read the same month:

The gap between the rep's Commit and the clean Commit is $60,000, almost a third of the number, and all of it rests on deal B, which nobody has touched in 41 days. The weighted pipeline looks reassuring at $269,750, but $87,000 of it comes from three silent deals. None of this shows in the forecast grid, because none of these deals changed: no stage moved, no amount moved, no close date moved. Pipeline Inspection highlights what changed; a silent deal is precisely one that did not.

The 30-day threshold is a choice, not a law. Pick the length of a normal gap between two touches in your sales cycle, apply it to every deal the same way, and keep it fixed, so that the trend means something.

A five-step routine that keeps the forecast honest

1. Fix the mapping first. Map late stages to Best Case and Commit deliberately, with exit criteria for each stage, so that the category means something before anyone overrides it.
2. Freeze the forecast at a fixed point. For example, the first Monday of the month. Measure accuracy against that snapshot with the forecast accuracy formula, not against the number on the last day.
3. Challenge silent deals before the call. Every Commit or Best Case deal with no activity past your threshold gets a new date, a downgrade or a written reason.
4. Check coverage on the clean pipeline. Run the pipeline coverage calculator on what remains after silent deals are removed, not on the raw total.
5. Clean the data on a schedule. Past close dates, missing next steps and duplicate opportunities distort every column. Our CRM hygiene checklist lists what to clean and how often.

Native forecasting or an add-on

Native Pipeline Forecasting is enough for teams that keep their stages honest and review deals weekly. Revenue intelligence platforms add call recording, deal scoring and forecast predictions, at a per-seat price that suits large teams more than small ones; our comparison of sales forecasting software lays out the options and their prices.

Between the two sits a narrower job: finding, every night, the deals that inflate the forecast without moving, and scoring each rep on how clean their pipeline is. That is what TruthPipe does, at $99 per month per org, reading the opportunities and activities already in Salesforce and writing nothing back.