Prevent duplicate contacts in fast-growing sales teams

Fast-growing sales teams want a CRM where every contact exists once. In practice, they add tools, imports and people faster than anyone can check what comes in. From what I've seen, full prevention is rarely possible, because CRMs and integrations do things you can't change. What you can do is stop duplicates from spreading and from staying. In this post, you'll learn which prevention steps hold up in a fast team, and where each one falls short.

Why you can't stop every duplicate at the door

Every CRM, tool and integration has native behavior you can't switch off. Some of them create records on their own, with rules you don't control.

One example is HubSpot creating orphan companies: it creates a company from an email domain, with no contact attached. It's a known issue, and it hasn't changed yet.

Fast teams also run on results. Reps add tools that help them sell, marketing adds forms and events, and nobody wants to slow down to check data.

In an ideal world, you'd prevent every duplicate. In reality, bad data will come in, and the goal is to keep it from spreading and from staying.

What native deduplication catches, and where it stops

Native deduplication works, to an extent. I've seen Pipedrive, Salesforce and Nimble catch some duplicates when a record comes in and ask whether you still want to add it.

Each check has its own limits, too. In some CRMs, for example, imports without an email, domain or record ID create new records, and records created through the API can skip the check.

The bigger limit is time. In a fast team, users don't stop to read a warning, so they click through it.

In many teams, more people than the CRM admin can import and upload whenever they need to. Even with a protocol in place, they skip it, and the duplicates become the admin's problem.

Native tools also struggle once duplicates are already in. Many teams that try to clean up with them stop partway.

What realistic prevention looks like

Prevention in a fast team works as a set of layers. Each one catches part of the problem, and the last one catches what the others miss.

Agree on who imports, and how

Limit imports to a small group, or at least agree on one way to do them. Decide which fields must be filled, how names and emails are formatted, and which object each list goes into.

Write it down where reps will see it, and keep it short. A protocol that takes too long is the one people skip.

Name one person who owns the protocol and answers questions about it. When a new tool or list source shows up, that person decides how its data gets in.

Check spreadsheets for duplicates before they go in

Event lists, partner lists and rep spreadsheets are some of the easiest duplicates to stop. Check each file for duplicates inside itself, and against what's already in the CRM, before you import it.

Use the same matching you'd use in the CRM. Allow for typos, nicknames and company suffixes, so near matches don't slip in as new people.

Use your CRM's native deduplication for what it catches

Turn on the native deduplication your CRM offers, and tune its native matching to your data. Every duplicate it stops is one less to merge later.

Check them now and then, too. If reps click through the same warning every day, the rule may be too broad. It may also point to a source worth fixing.

Merge new duplicates in a continuous loop

For everything that still gets through, schedule automatic merging to run ASAP. The tool scans, matches and merges right away, then starts over as soon as it finishes. New duplicates are found and merged within minutes of arriving.

Those minutes matter. A duplicate merged early doesn't reach reports, email sequences or ownership rules, where it turns into double calls and commission disputes.

Merge the matches you're most certain of automatically, such as the same email, or the same name and company. Send edge cases to review.

Where to start

Clean the duplicates you already have before you turn on any automatic merging. A loop on top of a messy backlog spends its time on old problems.

Then add the layers one at a time, starting with who imports. Watch which duplicates still get through, and adjust the layer that let them in.

Build your prevention in layers

You won't stop every duplicate in a fast team, and you don't need to. Prevention works as layers: an agreement on who imports, a check of each spreadsheet before it goes in, and the native deduplication your CRM already has.

Each layer catches part of the problem, and none of them catches all of it. What gets through is merged within minutes, before it reaches reports, sequences and ownership, so the team can keep moving without spreading bad data.

Frequently asked questions