Sales Cloud · changing stages on a live org
Change the stages, keep the deals.
The order of work matters more than the new list.
You can change opportunity stages on a live Salesforce org without losing a deal if every deal is moved to a new stage before an old one is retired, and no stage that still holds records is ever deleted. What goes wrong is rarely a lost record; it is a quietly wrong one, because each stage move also rewrites probability, forecast category, the closed and won flags and the date the deal last changed stage. This is the order of work, the queries that prove nothing was left behind, and how it went for an industrial services group, a holding and its operating companies.
Object Reference and Metadata API Developer Guide, Winter ’27, read on 16 November 2026. Reviewed 16 November 2026.
What a stage move rewrites
What changes on a deal when its stage changes?
More than the label. Salesforce documents that updating the stage automatically updates the forecast category, the closed and won flags and the probability from the settings of the new stage. A migration therefore overwrites every probability a rep adjusted by hand, and moves the date the deal last changed stage, which stuck deal views are built on.
All of this is in the Opportunity entry of the Object Reference for the Salesforce Platform, Winter ’27, last updated 6 November 2026, read 16 November 2026: Closed and Won can only be set through the stage, and the last stage change date has been a standard field since API version 52.0. The same reference lists six forecast categories, from Pipeline to Omitted, each stage carrying one by default.
| On the deal | What a stage move does | Before the move |
|---|---|---|
| Probability | Reset to the default of the new stage | Export it, and decide with sales leadership whether hand adjusted values come back |
| Forecast category | Reset to the default of the new stage | Export it with the probability; the weighted pipeline moves on migration day |
| Closed and Won | Follow the flags of the new stage; they cannot be set any other way | Decide which new stages are closed and which are won before a single deal moves |
| Last stage change date | Records the last change to the Stage field, so every moved deal reads migration day | Export it; after the move, stuck deal views count from migration day |
| Automation on stage change | Runs, as for any edit by a person | Read every flow that starts on a stage change before the script runs |
Whether your stages mean the same thing to every rep is a different question, and it comes first: it is pattern 01 of the Sales Cloud audit, together with the probabilities that pattern 04 compares against four quarters of real conversion. This article starts where that one ends, with a new list agreed and a live pipeline to move into it.
The order of work
In what order do you change stages on a live org?
Agree one written stage list first. Then add the new stages beside the old ones, move the deals by a mapping decided from the data, repoint everything that names a stage, and only then deactivate the old values. Deactivate rather than delete: Salesforce keeps inactive stages for history, and reports on past quarters still need them.
The rule about inactive values is in the OpportunityStage entry of the same reference; the rule about record types is in StandardValueSet in the Metadata API Developer Guide, Winter ’27. The reason stages come before anything else: every validation, list view and report is written in their names.
- 01
One stage list, agreed in writing
Before anything is configured. In the build below the client’s documents gave several different lists, and the stages were rebuilt more than once before a written decision existed. Every rebuild meant moving the deals again and repointing everything that named a stage.
- 02
An inventory of everything that names a stage
Validation rules, flows, list view filters, report filters, dashboard filters, the path and its guidance, sales processes, field help text and the training guide. Retrieve the metadata and search it for every old stage name, then search your own documents for the same words.
- 03
A snapshot of every deal before the move
Id, stage, probability, forecast category, amount, close date and last stage change date, for open and closed deals alike. This file is the rollback, and it is the only record of what the pipeline looked like the day before.
- 04
New stages added beside the old ones
Each with its closed and won flags, default probability and forecast category set on purpose. Add every new value to every record type (and its sales process): Salesforce documents that new values loaded through the Metadata API do not appear for a record type until they are added to it.
- 05
A mapping decided from the data, not from the names
For each old stage, look at what the deals in it actually contain before choosing where they go. Two stages with similar names can hold deals at different points of the sale, and one old stage can split in two.
- 06
Automation guarded, then the move by script
If a flow stamps the close date when a deal closes, check whether it also fires when a deal moves from one closed stage to another; if it does, the migration redates history. Then move the deals in batches with a script that reads its own results back.
- 07
Everything that named a stage repointed
Validation rules, flows, list views and reports from the inventory, and the path guidance rewritten, because sentences written for the old stages are now untrue. List views that mean “open” should filter on Closed equals false, not on a list of stage names.
- 08
Old stages deactivated, then counted
Deactivate, do not delete. Then run the three queries below. The work is finished when the count of deals on a retired stage is zero and the closed and won totals match the snapshot.
Check your own org
How do you prove no deal was left on an old stage?
Count. Group every opportunity by stage and compare the result with the active stages in OpportunityStage: any stage that holds deals and is not active is a deal left behind. Then compare the closed and won counts, and the pipeline total, with the snapshot taken before the move. Zero and equal are the only passing results.
Run these in the Developer Console query editor. The first lists the stages as Salesforce holds them, with their flags; the second shows where the deals actually are; the third, with your retired stage names in it, must return 0.
If a stage in the second result has Closed set to true and you did not expect it, a list view or report that names stages instead of filtering on Closed is showing those deals as open.
One build, from the inside
Eight stages for an industrial services group.
An industrial services group, a holding and its operating companies, restarting Salesforce after a first rollout that stalled, in production with no sandbox. Its sales standard defined a tender process: New, Engagement, Proposal Preparation, Proposal Evaluation, then four ways a tender closes, won, lost, withdrawn by the group and withdrawn by the customer. The whole build is in the second Salesforce implementation case study; this is only the stage change.
- 01
Mapped from the data, moved by script
The mapping from the earlier stages was decided from what the deals held, and the move ran as a script, with a zero count of deals on retired stages as its acceptance test.
- 02
Stages that move themselves
The first task or meeting logged on a deal in New moves it to Engagement, through two small flows on Task and Event, so a rep never has to remember to do it.
- 03
A close date that sets itself
Closing a deal stamps the close date with that day in a before save flow, which Salesforce’s own decision guide calls the most performant declarative way to update the triggering record before the initial DML commit.
- 04
Guidance read again after every change
Every stage change broke sentences in the path guidance, written in the client’s own words. The rule that came out of it: after changing a value, search for the old word in all your own text, not only in the configuration.
The one step that build skipped at the start is step 01, and skipping it meant rebuilding the stages more than once before the written list existed.
What makes it worse
Four things that survive a stage change unnoticed.
None of them raises an error. Each one keeps working and starts reporting something untrue, which is the same family of fault as the dashboards in why the finance team stops trusting the numbers.
- 01
A list view that names the closed stages
A filter of “stage is not Closed Won or Closed Lost” was right when those were the only closed stages. Add a withdrawn stage and every withdrawn deal reappears as open pipeline. Filter on the Closed flag, which follows every closed stage automatically.
- 02
A report filtered on probability above zero
It looks like a way to hide dead deals. It also hides every stage whose default probability is zero, including new closed stages that the report description says it counts.
- 03
A validation rule that fires on every stage change
A rule such as “close date cannot be in the past” written on stage change freezes every old deal the moment anyone moves it, including the script. In the build below it fires only when a deal is created or its close date is edited, after a count showed how many open deals already had a past date.
- 04
Path guidance nobody reread
A path follows the order of the picklist values it is based on, and each step carries guidance only if someone wrote it (Salesforce Help, Paths). A new stage therefore starts without guidance, and the old guidance keeps describing stages that no longer exist until someone reads it word by word.
Asked often enough to answer here
Questions
- Q
Should I delete old opportunity stages or deactivate them?
Deactivate them. Salesforce documents that inactive stage values are not available in the picklist and are retained for historical purposes only, so deals closed in earlier quarters keep the stage they closed in. Delete a value only after the count query shows that no deal holds it.
- Q
Will changing stages change our forecast?
Yes, on the day of the move, even if no deal changed. Every moved deal takes the probability and forecast category of its new stage. Export both values first, agree with sales leadership whether adjusted probabilities are restored, and compare the forecast totals before and after.
- Q
Can we do this without a sandbox?
Yes, with more discipline. The group in this article had production only. The move ran as a script with a saved snapshot, the changes went in as named files, and the counts were read back from the org. A sandbox would add one thing: a rehearsal of the script on a copy of real deals.
- Q
Do the reps need retraining after a stage change?
They need the new definitions in their own words, on the screen where they work. That is what path guidance is for: one sentence per stage on what must be true before a deal moves on, rewritten whenever the stages change, and read once more by somebody who sells.
The base package includes the revenue process encoded: stages with exit criteria that mean something, conversion, ownership and sharing, for up to 25 seats and one selling motion. A rebuild of an existing org is scoped on the same page.
Serhii Skrypnyk · Senior Salesforce Administrator · 7 Salesforce certifications · on the platform since 2018. Reviewed 16 November 2026.
Read next
The question that usually comes after this one.
- 01
Case: Group of companies, second rollout
A holding and its operating companies came back to Salesforce after a first rollout stalled: tenders as leads, one customer record in either alphabet, access proved as the users.
- 02
Sales Cloud Audit: Eight Checks, €650 Day
A Sales Cloud audit in eight checks your admin can run, or a €650 data quality check in one day against your exports, €4,875 to fix all eight.
- 03
Sales Cloud Implementation Cost in 2026
Sales Cloud implementation from €11,310 for 174 hours: the seven decisions that move those hours, and why billed hours and elapsed weeks are different numbers.