Salesforce Forecasting: Setup, Categories and Limits

Salesforce forecasting is the native Sales Cloud feature, now called Pipeline Forecasting and long known as Collaborative Forecasts, that rolls up opportunity amounts by forecast category, time period and hierarchy into a forecast reps and managers can adjust. Each opportunity's stage sets its category: Pipeline, Best Case, Commit, Closed or Omitted. Up to four forecast types can be active.

Almost every company I work with runs Salesforce, and most of them use its forecasts page in some form. This guide covers how it works, how to set it up, the limits Salesforce documents, and where the roll-up stops being enough. Every feature claim below comes from Salesforce's own documentation, mainly the Pipeline Forecasting Implementation Guide (Winter '27, last updated July 22, 2026) and the Sales Cloud Einstein setup guide. If you want the general picture of forecasting from a CRM first, see CRM forecasting.

How does Salesforce forecasting work?

Salesforce adds up the opportunities each person owns, groups them by forecast category and period, and rolls those totals up a hierarchy so every manager sees their team's number. In Salesforce's words, a forecast "expresses expected sales based on the gross rollup of a set of opportunities."

Five pieces make it run:

Source for every row: Salesforce, Pipeline Forecasting Implementation Guide.

The forecasts page then shows a grid by rollup, period, and optionally product family or territory, with the opportunity list underneath. Managers can look at one rep or everyone below them.

What are the forecast categories in Salesforce?

The standard categories are Pipeline, Best Case, Commit, Closed and Omitted, and Omitted deals are left out of the forecast. Admins can turn on a Most Likely category in Lightning Experience, which sits between Commit and Best Case. You do not set a category on its own. Each value in the opportunity Stage picklist maps to one category, so moving a deal to "Negotiation" can move it into Commit if that is how your org maps it. You check and change the mapping in Object Manager under Opportunity, Fields, Stage.

How the categories add up depends on the rollup method. With cumulative rollups, the default, each column includes the categories that sit closer to closing:

The column names and what each includes come from Salesforce's guide. The amounts are illustrative: a quarter with $400,000 closed, $600,000 in Commit, $500,000 in Best Case and $1,500,000 in Pipeline. Note that Open Pipeline leaves out Closed, so it is not "everything." For how teams define those two middle categories, see commit vs. best case, and for the bucket that drops out, the Omitted category.

With single category rollups, each column holds only its own category, and a rep who wants the quarter total adds Best Case, Commit and Closed by hand. Most teams keep cumulative.

What forecast types can you create?

A forecast type is an object plus a measure, and Salesforce lets you run up to four at a time. Customer Support can raise the limit. The options Salesforce documents:

- Opportunity: revenue (Amount), quantity, or a custom currency or number field. Can group by a custom picklist, such as industry or business unit. - Opportunity Product: total price or quantity, grouped by product family or a custom picklist. The grid shows up to 2,000 rows, so many product families can crowd it. - Opportunity Split: revenue, overlay or custom splits, for team selling. Needs team selling and opportunity splits turned on. - Opportunity Product Split: product family only. - Line Item Schedule: revenue or quantity by schedule date, for products with revenue or quantity schedules.

You can also forecast on a custom date instead of Close Date, on Expected Revenue, or by territory if you run an active territory model. In a territory forecast each opportunity rolls up to one territory only. Salesforce also added consumption forecasting for usage-based revenue in its Spring '25 release.

One rule saves a lot of pain: design the type before you activate it. Salesforce does not let you edit filter definitions after activation, and deactivating a type deletes its forecasting data.

How do you set up Salesforce forecasting?

Turn it on in Forecasts Settings, add the tab, activate a forecast type, choose a rollup and date range, then enable users and managers. Orgs created in Winter '24 or later already have forecasting on with Salesforce's defaults, so for them this is a settings review. These steps follow Salesforce's implementation guide, in Lightning Experience:

1. Turn on forecasting. Setup, Quick Find "Forecasts Settings," enable it. Review the defaults: an Opportunity Amount type by close date on the user hierarchy, manager adjustments, cumulative rollups, org currency, a monthly period and quotas shown. 2. Add the Forecasts tab. Setup, App Manager, edit your sales app, Navigation Items, and move Forecasts to Selected. 3. Create and activate forecast types. In Forecasts Settings, click Create a Forecast Type, follow the flow, then choose Activate from the type's quick actions. Check field-level security first so Amount, Quantity and any custom measure are visible. 4. Choose a rollup method. In Manage Forecast Rollups, pick single or cumulative. Decide now, because switching later deletes adjustments. 5. Set the default date range. Pick monthly or quarterly, the start and the number of periods. The maximum is 15 months, 15 fiscal periods or 8 quarters. Pick this one carefully too (see the limits below). 6. Enable users. Setup, Users, edit each user and check Allow Forecasting. Users need View Roles and Role Hierarchy for role-based forecasts, which Salesforce assigns to forecast users by default. 7. Build the forecast hierarchy. Setup, Forecasts Hierarchy, Enable Users, then Edit Manager next to each role to name its forecast manager. Until you do, a VP cannot see the team's forecasts even if the role hierarchy says the team reports to them. 8. Map stages to categories. Object Manager, Opportunity, Fields, Stage, and set the Forecast Category for each stage value. Rename rollups in Forecasts Settings if your business uses other words. 9. Turn on adjustments and judgments. In Forecasts Settings enable them, then give the right profiles the Override Forecasts permission. You can show adjustments as columns and rename them, for example "My Commit" to "Call." 10. Show quotas and load them. Turn on Show Quotas, then enter quotas in Setup under Forecasts Quotas by period, type and user, role or territory. 11. Report on it. Create a custom report type on Forecasting Items (forecasts and adjustments) or Forecasting Quotas with Forecasting Items (attainment), then build reports and dashboards from them.

Salesforce also offers Lightning App Builder pages for the forecasts page if the default layout does not fit. It is optional.

What limits and gotchas does Salesforce document?

The ones that bite are about deleted data: switching settings later can wipe adjustments, judgments and quotas. Here is what Salesforce's guides say, in one place: Source: Salesforce, Pipeline Forecasting Implementation Guide.

Where do Einstein Forecasting and Pipeline Inspection fit?

Einstein Forecasting adds an AI prediction of where each team will land, and Pipeline Inspection shows how the pipeline changed. Both sit on top of the same opportunities.

Einstein Forecasting is part of Sales Cloud Einstein. Salesforce says it is available in Performance and Unlimited editions and for an extra cost in Enterprise. Its documented requirements (Sales Cloud Einstein guide):

- Salesforce forecasting enabled, and at least 12 months of opportunity history with at least one update in each of the past 12 months - A standard fiscal year - A forecast measured by opportunity revenue; predictions cover only the oldest activated opportunity revenue type - The standard Opportunity object, Close Date and Amount fields (custom dates are not supported), with Amount filled on at least 80% of open opportunities - At least one forecasting user who reports to a forecast manager - No opportunity splits; production orgs only, not sandboxes

When it is on, predictions show in a slide-out panel from the Einstein Prediction column on the forecasts page.

Pipeline Inspection needs historical trending on the Opportunity object. It marks changes to Amount, Close Date, Forecast Category and Stage, and adds a waterfall chart of pipeline changes and a flow chart of category changes (Trailhead). That matters more than it sounds. The most interesting data in a CRM is often not the field. It is the change in the field. A deal sitting in Proposal is useful. Knowing it moved into Proposal yesterday is more useful.

On price, Salesforce's pricing page lists "Sales Quoting and Forecasting" in Pro Suite, at $100 per user per month billed annually, and lists Predictive Forecasting as available for purchase. Check your contract for what you already own.

Where does the Salesforce roll-up break?

The roll-up is a sum of what reps have entered, sorted by the stages reps chose. It is accurate when that input is disciplined and inherits every error when it is not. That is not a flaw in Salesforce. It is what a roll-up is. Here is where I see it break:

- Stages are subjective. Category comes from stage. Stage weighting works if each stage has strict entry and exit criteria. When it does not, the stage is the rep's judgment, and you are applying an objective rule to a subjective input. You will miss at the end of the quarter. - One mapping for every deal. New business, expansion and renewals usually share a pipeline and a stage map, even though they close very differently. Renewals close on a known date, usually within about 30 days either side of it. New business timing follows the last stage change. - The roll-up only sees deals that exist. A quarter has three sources: carry-over deals open on day one, deals created and closed inside the quarter, and deals pulled forward from later quarters. Opportunities not yet created cannot roll up, so a quarter with a strong in-quarter motion looks thin early. - Amounts run high. Most deals close for less than the amount in Salesforce. A pipeline can carry an average deal of $80,000 while closed-won deals average $40,000. The gross rollup counts the $80,000. - Close dates drift. Across ORM customers, only about 20% of the pipeline dated to close in the quarter on day one closes in that quarter. The best slip signal is a rep changing the close date, and the earliest one is no signal at all: no stage change, no amount change, no notes.

Salesforce gives you adjustments for exactly this reason: a manager who knows a rep runs optimistic can change the number. That is judgment layered on a sum. It works when the manager is right. It is hard to audit when the board asks why the number moved. The odds are not great for anyone: Gartner reports a median forecast accuracy of 70% to 79%, and only 7% of sales teams reach 90% or better.

Illustrative example: the same quarter, two views

Take the quarter in the categories table, with a $1,500,000 quota at mid-quarter. Salesforce shows Commit Forecast $1,000,000 and Best Case Forecast $1,500,000. The manager adjusts Commit up to $1,100,000 based on two deals they believe in. Gap to quota on the call: $400,000.

Now suppose your own history says Commit deals close at about 80% of their amount within the quarter, Best Case at 30%, and Pipeline at 8%, and that you usually create and close about $150,000 of new deals inside a quarter after mid-quarter. All of those rates are illustrative. The expected number is:

The call says $1,100,000. Best Case says $1,500,000. A model built on your history says about $1,300,000, with a range around it. None of those is "the" answer. The value is seeing all three, and knowing that $150,000 of the expected number is not in Salesforce yet.

When do teams add a dedicated forecasting layer?

Usually after a missed quarter with no early warning, or when RevOps is buried in spreadsheets keeping the forecast alive. Those are the two triggers I hear most. A miss by itself is survivable. A miss nobody saw coming breaks trust. On the operations side, I have seen Sales Ops teams maintaining 37 spreadsheets with 12 tabs each, and then being asked to run the forecast call too. At that point they spend more time building the forecast than reading it.

Signs the native roll-up is no longer enough:

- The forecast swings late in the quarter, and nobody trusts it until the last few weeks. - Commit regularly closes well below its rollup, and adjustments are where the real forecast lives. - New, expansion and renewal revenue need different logic, and one stage map cannot hold it. - A large share of the quarter comes from deals created inside it. - The board wants a range (low, on-target and high) that rolls up the same way from region to company.

Native Salesforce forecasting is the better choice when stages have strict criteria, the team closes a handful of deals a month (too few for a model to learn from), and managers run a disciplined weekly call. A model needs volume: at a minimum a couple of hundred closed-won deals, and at least a year of history to read seasonality.

Adding a layer does not mean replacing Salesforce. When ORM goes in, Salesforce stays the same, and so does the sales process. ORM snapshots the pipeline every day and models close timing, amount and seasonality from your own history. ORM targets 95% accuracy with no manual adjustments, from day 1 to day 90 of the quarter. Some teams keep the rep roll-up beside it on purpose: one view of what the field believes and one of what the data says. For the full comparison, see ORM vs. Salesforce forecasting and CRM forecasting vs. dedicated forecasting.

How to get more out of Salesforce forecasting today

Most gains come from process, not settings. Five habits that help whichever tool sits on top:

1. Write entry and exit criteria for every stage, because the stage sets the category. If a rep cannot show the criteria, the deal does not move. 2. Split motions. Group the forecast by a custom picklist for new business, expansion and renewal, so each can be read on its own. 3. Report on adjustments. The Forecasting Items report type carries Owner Only Amount, Amount Without Adjustments, Amount Without Manager Adjustments and Forecast Amount. The gap between them shows where judgment is doing the work. 4. Turn on historical trending and Pipeline Inspection, and review close-date pushes every week. A deal that slips from one quarter to the next is less likely to close, even in Commit. 5. Talk to sales leaders before the call. The RevOps person who owns the forecast should be talking to sales leaders daily, and always before the weekly forecast call, to test the assumptions behind the number. A forecast cannot be built in a vacuum.

If you report pipeline coverage from the same data, our guide to pipeline coverage in Salesforce shows the report and the filters that keep the ratio honest.