CRM deduplication for consultants and agencies

When you clean a client's CRM, you work in their data, with their tools, on their deadline. Some of you manage several clients' CRMs month after month. Others come in for one project, often before a migration or a launch. Either way, the part I hear about most is proving the work. A good cleanup leaves a smaller database that the client can't easily check. In this post, you'll see how to agree on criteria, keep a record of every merge and report results a client can trust.

Two kinds of client deduplication work

Agencies and freelancers who manage several clients' CRMs do recurring work. Each client has its own CRM, its own tools and its own idea of what a duplicate is. New duplicates keep arriving from forms, imports and integrations, so the job never fully ends.

Consultants on one-off projects work toward a date set by something else: a migration, an integration launch or the start of a contract. The project ends with a handover, and the client lives with every merge after you leave.

The two need different rhythms, and the same proof. In both cases, the client has to be able to see what you did and why.

Why proving the work is the hard part

A finished cleanup shows up as fewer records. The client sees their contact count drop by thousands and asks what went missing.

A single wrong merge is far more visible than a thousand right ones. A rep finds an account under someone else's name, or a phone number they used last week is gone. Without a record, you can't show what happened.

That's why I treat proof as part of the project from day one. Everything below works the same in HubSpot Contacts, Salesforce Accounts or Pipedrive People.

Agree on what counts as a duplicate before you merge

Every team defines duplicates a little differently. One client treats each branch office as its own company, and another wants one record per company everywhere.

Settle these questions with the client before anything merges, and put the answers in writing. They become your criteria, and later they're your answer to "why did these two merge?"

Show the before and after

Take your "before" numbers before you change anything. Record the count of records per object, the number of duplicate sets each criterion finds, and where the duplicates came from.

Export the records you plan to merge, too. If a question comes up months later, you have the data as it was on day one.

Walk the client through a few merges before you run the rest. Show them the records going in, the values that win and the record that comes out. After the merge, take the same counts again so the before and after sit side by side.

Keep a record of every merge

For each merge, keep the original records, the winning values, the final result and a link to the merged record in the CRM. That's what lets you answer "where did this phone number go?" in a minute.

In many CRMs, a merge can't be fully undone. A merge record with the original values is how you recreate a record if a merge turns out to be wrong.

Hand the full record over at the end of the project. It shows the client what changed, and it protects you when someone asks about a merge after you've left.

Report results the client can use

A good report tells the client what changed and what to do next. I'd include:

• Duplicate sets merged and records removed, per object.
• Sets left for the client to review, and why each one needs a person.
• Sets you reviewed and rejected, because they're separate records.
• The sources still creating duplicates, with an example from each.
• What you recommend next, such as fixing a form or a recurring check.

Working across each client's tools

Every client brings a different stack: web forms, enrichment and prospecting tools, spreadsheets, and sometimes a second CRM. You can't deduplicate inside every one of them. An enrichment tool, for example, usually can't be cleaned on its side, and its data goes into the CRM anyway.

Clean data before it enters where you can: spreadsheets before import, and the old system before a migration. Then clean in the CRM for everything else. When two CRMs sync both ways, a duplicate exists on both sides, so the merge has to happen in both.

In the first week, list every tool that creates records for that client. It explains why duplicates keep coming back, and it turns your report into a plan.

For ongoing clients, automate what you're sure of

Once the backlog is clean and the client trusts your criteria, set the patterns you're most certain of to merge automatically. Records where three or four strong fields match, such as name, email and company, are good candidates.

Merging new duplicates automatically keeps them from spreading between your check-ins. Edge cases still go to review, on your side or the client's.

Send a short report on a regular schedule, monthly for most clients. The same numbers each time make the work visible long after the first cleanup.

Put the agreement in writing before you merge

Whether the work is recurring or a one-off project, a client cleanup comes down to proof. Criteria agreed in writing explain why records merged, and "before" numbers next to the same counts afterward show what changed.

A record of every merge, with its original values, ties the work together. It answers questions long after the project ends, and it's what the client keeps when you hand the work over.

Frequently asked questions