The data workstream in a Salesforce org merge: what it covers and who owns it

An org merge has a configuration workstream and a data workstream. What the data work covers, who owns each task, and what to put in the SOW.

The short answer: A Salesforce org merge has two workstreams: configuration (objects, fields, automation, and permissions in the surviving org) and data (profiling both orgs, deduplicating each one, matching accounts and contacts across the two, deciding which values survive, keeping a crosswalk of old and new record IDs, loading in order, and checking the result after cutover). An implementation partner often leads the configuration and runs the loads. The data decisions belong to your business, and the cleaning and matching belong to whoever the statement of work names, so if it names nobody, that work is yours.

For the full sequence of an org merge, see how to merge two Salesforce orgs after an acquisition. For the matching method, see how to deduplicate accounts across two Salesforce orgs.

The two workstreams in a Salesforce org merge

Salesforce's own Help article on moving data from one org to another describes two kinds of work. First the target org needs matching metadata: record types, page layouts, validation rules, automation, and the sharing model. Then the records are imported in sequence so the relationships between them can be rebuilt. The same article says that planning, scoping, running the migration, and manipulating the data are outside what Salesforce Support covers, and that Salesforce Professional Services or a consulting partner can be brought in for help.

If you bring in an implementation partner (Salesforce calls them consulting partners), the statement of work settles who does what. The data workstream is the harder one to write down. Part of it is decisions only your company can make, such as which of two owners keeps a shared account, and the rest is cleaning and matching whose size nobody knows until someone has looked at both orgs.

What the data workstream covers

There are seven tasks. The table shows a common split between deciding and doing. Your split may differ, as long as every cell has a name in it.

Task Who usually decides Who usually does the work
Profile both orgs RevOps or business systems lead Each org's admin, or the partner's data lead
Deduplicate inside each org Each org's admin Each org's admin
Match accounts and contacts across the orgs RevOps, with account owners confirming unclear pairs Whoever the statement of work names
Decide which values survive Sales leadership and finance RevOps writes the rules down
Keep the crosswalk of old and new IDs RevOps Whoever runs the loads
Load in order, with rehearsals RevOps signs off each rehearsal Whoever runs the loads
Check the result after cutover RevOps accepts or rejects The partner fixes load errors

Because the tasks are spread across several people, the workstream needs one person above them, usually the head of RevOps or business systems, who makes sure each task has a name and a date. Profiling and in-org deduplication can start before a partner is chosen, since each org's admin can do them. The other five need more explanation than a table cell.

Cross-org matching needs a producer and reviewers

Salesforce's matching rules identify duplicates among the records in one org. Each admin can use them to clean their own org, but they cannot compare the two, so cross-org matching happens outside Salesforce and produces a list of pairs for review. Check this task first for missing names: who produces the list, and who reviews the uncertain pairs.

The review stays with your account owners and the acquired company's, because they know whether two records are one customer. A reviewer can only judge a pair once someone has pulled the names, domains, contacts, and open deals for both records into one place, and for two orgs of any size that preparation is a project of its own. If nobody on either side has those hours, Quill matches accounts and contacts across both orgs and attaches the evidence to every proposed pair, so the people who know the customers only have to approve or reject.

Survivorship decisions stay with the business

Somebody has to say which owner and which address each merged account keeps. Sales leadership decides account ownership and how the two sets of stages line up, and finance decides anything billing depends on. Your partner can apply the rules and will need them early. Survivorship rules for CRM records covers how to write them.

The crosswalk outlives the project

Salesforce's guidance says relationships are stored by record ID, and that IDs from the source org have to be replaced with the new IDs created in the target org. The table mapping each old ID to its new one is the crosswalk, and whoever runs the loads builds it. Settle who keeps it afterward, because finance and the data team will need it to repoint anything that stored an old ID.

Loads and cutover checks need someone to accept them

Salesforce's migration guidance says to migrate objects in dependency order and test in a sandbox first, then compare record counts with the source, spot-check a sample of records, and review the error logs. Your part is deciding what counts as a pass and what happens to rejected rows. If a duplicate rule is active in the surviving org, for example, Data Loader inserts can fail on rows that match it, and each one is a matching decision for the data side. What to check after a CRM migration has a fuller list.

What to write into the statement of work

Salesforce's Trailhead training for consultants on writing a statement of work treats data migration as a scope question to answer in the document: "Who moves the data from the old systems to Salesforce?" An older Salesforce checklist for assessing statements of work, last updated in 2015, says the data migration item should give details on importing data from other systems, including de-duping. A partner should be used to this conversation. In an org merge, moving the data and making it fit to move are different jobs, and the document should say who does each.

Check these before you sign, or raise them as a change request if the project has started:

  • Each data task, with an owner. Put the partner, your team, or a named third party next to each of the seven tasks. A single line that says "data migration" does not tell you who deduplicates or who matches.
  • The assumptions and customer responsibilities sections. A line saying you will provide data that is ready to import puts the cleaning and matching on your side. Simplus, a Salesforce consultancy, wrote in 2018 about the trouble that line causes when the data is not ready. It is a reasonable arrangement if you have staffed it.
  • Volumes, and what happens if they are wrong. Profile both orgs before the estimate is final, counting the records in each object and the duplicates among them. Coastal, a Salesforce consulting partner, also recommends profiling early. If the estimate assumes 2,000 shared accounts and profiling finds 6,000, for example, the document should say how the difference is handled.
  • Review rounds and rehearsal loads. How many of each are included, and who signs off.
  • The crosswalk as a named deliverable, with its format and who receives it at the end.
  • Acceptance criteria for the data. Counts that reconcile to the source, no child records without a parent, an active owner on every record, and a sample of shared customers confirmed by their account owners.
  • Support after cutover. How long the partner fixes data issues, and who runs duplicate checks after that.

If a third party does the cleaning and matching, name it in the document too. For the kinds of outside help available, see who can help combine CRMs after a merger.

Signs that the data workstream has no owner

  • The project plan shows data as one line with load dates, and no tasks before it.
  • Nobody can say how many accounts exist in both orgs.
  • Questions about which owner or which address wins are answered one at a time in chat, by whoever happens to be asked.
  • The first rehearsal load is the first time anyone looks at the acquired org's duplicates.
  • Finance has not been asked which systems store Salesforce record IDs.

Each of these is easy to fix early. Name the owner and run the profile, then take the gaps to your partner as a scoping conversation.

Is there a tool that cleans and matches the data in a Salesforce org merge?

Yes. Quill CRM Migration (tryquill.com) takes on the data workstream while your implementation partner builds the configuration. Its AI agents connect to both Salesforce orgs, deduplicate each one, and match the same account or contact across the two on names, domains, addresses, shared contacts, and activity. The other tools you will hear about do different jobs: Salesforce's duplicate rules and merge work inside one org, and Data Loader and similar tools move the rows they are given without judging whether two of them are the same customer. Quill's agents also apply your survivorship rules, keep each old record ID so billing and ERP links can be repointed, and check counts, owners, and links after cutover. Every proposed change shows the evidence behind it, and nothing is written until a person approves it, with each change logged so it can be undone. The Quill team runs the first diagnostic and builds the agents for your two orgs, working alongside the partner you already have. The same approach covers Salesforce, HubSpot, Attio, and other CRMs such as Microsoft Dynamics 365, Pipedrive, and Zoho CRM. If you have an org merge coming up, book 15 minutes.

This is general information, not legal or accounting advice.

Frequently asked questions

What is the data workstream in a Salesforce org merge?

It is all the work on the records, as opposed to the configuration. It covers profiling both orgs, deduplicating inside each, matching accounts and contacts across the two, deciding which values survive, keeping a crosswalk of old and new record IDs, loading in dependency order, and checking the result after cutover.

Who owns data cleanup in a Salesforce org merge?

Whoever the statement of work names, and your own team if it names nobody. A common split is that each org's admin deduplicates their own org, the business decides survivorship and account ownership, and the partner or a specialist tool does the cross-org matching and the loads. One person on your side should own the workstream as a whole.

Does a Salesforce implementation partner's statement of work include data cleansing?

Only if the document says so. Look at the scope, assumptions, and customer responsibilities sections for who deduplicates, who matches records across the orgs, and what condition the data must be in when you hand it over. If cleansing is listed as your responsibility, either staff it or ask the partner to quote it.

When should the data workstream start in an org merge?

At the same time as the configuration work, and profiling can start earlier. Deduplicate each org while the target design is being settled, and begin cross-org matching once the account model is decided. Survivorship rules should be signed off before the first rehearsal load.

Does Data Loader catch duplicates when you load another org's records?

Not in the sense of recognizing the same customer under two names. Data Loader loads the rows in your file, and Salesforce's own import guidance tells you to clean the file, including removing duplicates, before you import. An active duplicate rule in the target org can reject rows that match existing records, but it does not compare the rows in your file with each other. With an external ID field holding each source record's ID it can upsert, which keeps the same source row from being loaded twice.

Sources

  1. Migrate data from one organization to another, Salesforce Help
  2. Salesforce Data Migration Best Practices: How to Plan, Execute, and Validate a Data Import or Update, Salesforce Help
  3. Error 'Use one of these records?' in Data Loader, Salesforce Help
  4. Manage Duplicate Leads and Contacts in Salesforce, Salesforce Help
  5. Resolve and Prevent Duplicate Data in Salesforce, Trailhead
  6. Merge Duplicate Accounts in Lightning Experience, Salesforce Help
  7. Things to Know About Duplicate Rules, Salesforce Help
  8. Manipulate Records with DML, Trailhead
  9. Import Data, Trailhead
  10. Build a Statement of Work, Trailhead
  11. Best Practice: Partner Statement of Work (SOW) Checklist, (PDF, 2015), Salesforce
  12. Reduce the risks in data migration, (2018), Simplus
  13. What Is a Salesforce Org Merge?, Coastal

See what we find in your data

It starts with a 15-minute call. Then we run a diagnostic, show you the issues in your data, and build agents for them.

Prefer email? Reach us at hello@tryquill.com. We respond to every message personally.

Tell us what's broken

We reply within one business day.

One last step

Everything you entered is filled in below. Pick where you'd like to send it from.