Salesforce deduplication you can review before anything merges

Standard matching rules compare a fixed set of fields, and they run at the moment a record is created. They do nothing about the duplicates already in the org, and they do not see the ones that came in through an import that bypassed them.

Native dedupe apps work on the object level inside Salesforce, which is convenient and also means the work is invisible outside the org. If your duplicates arrived from a spreadsheet, they will arrive that way again next quarter.

Accounts are the expensive case, because a merge moves contacts, opportunities and ownership onto one record and takes at most three records at a time. The account version of this workflow, with owner conflicts flagged before anything merges, is on Salesforce merge accounts.

The cheapest deduplication happens before data enters the CRM. Cleaning a list on the way in stops you from paying for the same cleanup twice, once in the file and once in the org. Doing that on any CRM export is CRM data cleansing, the matching it relies on is fuzzy matching rather than a fixed field rule, and the data cleaning tool runs it without the file leaving your machine. The version of this with a deadline attached, where the org is being left behind entirely, is a Salesforce to HubSpot migration. When the move runs the other way and the org is the destination, the Import Wizard and Data Loader matching rules decide what doubles, and that is covered in HubSpot to Salesforce migration. Which loader to use for each object is compared in Salesforce data migration tools.

Once the groups are approved, the merges themselves still happen in the org, three records at a time. Contacts and leads are covered on Salesforce merge contacts, and accounts on Salesforce merge accounts.

We exported an eleven row Contact report with the columns an admin would pull: Id, First Name, Last Name, Account Name, Email, Phone, Mailing State, Lead Source and Created Date. The engine proposed three duplicate groups and thirteen field fixes, and it changed nothing until each one was approved.

Rachel Nguyen came in twice, once typed in capitals with a trailing space on the email, so the pair was caught on the email address. Marcus Delgado at Copperline Freight Inc. and at Copperline Freight had two different work emails, and the pair was proposed on name and company with the legal suffix ignored. Allison and Alison Reyes at Summit Ridge Capital were proposed as a close match and flagged, because the emails differ and only a person can say whether the spelling is a typo.

This is the part that matters in Salesforce. When a group is approved, the row that is kept stays in the cleaned file and the duplicate row is removed, but its record Id is not lost. It is written into an added column called other_id on the surviving row. In our run, row 2 (0035e00000AbC01) came back with 0035e00000AbC02 beside it, so every line of the cleaned file reads as a master Id and the Id to merge into it.

Fields the two records disagree on are kept the same way. The second email address goes to other_emails, and a differing Lead Source or Created Date goes to other_lead_source and other_created_date, so nothing a sales rep typed disappears. The audit log lists each removed row with the row it merged into and every one of those values. Your admin works down that list with the native merge screen or the merge tool of their choice.

There are three ways to deal with duplicates already in an org. Each one is right for a different team, and the choice usually comes down to who does the work and whether the merges have to be explained later.

The buyer is the person who owns the list when the numbers are questioned: a RevOps or Salesforce admin cleaning up before a territory change, an ops lead about to load a trade show list, or a consultant handed an org with years of imports in it. They need the duplicate groups found this week, a list their admin can work from, and proof of what was merged.

The Analyst plan runs unlimited files of up to 25MB each, and Enterprise puts several people on one invoice with a DPA. See pricing, or run the demo above on your own report first. Nothing leaves your browser in the demo.