Data Deduplication for CRM: How to Find and Merge Duplicates

Data deduplication is the process of finding records that describe the same person or company and merging them into one. In a CRM, it is the difference between one clean account history and three half-empty records that each tell a different story.

This guide covers why duplicates pile up in HubSpot and Salesforce, the matching rules that find them, and how to merge without losing data. It ends with the workflow that stops most duplicates from ever reaching the CRM.

It is written for RevOps, sales ops, and founders who own the CRM. If you are also fixing stale fields, our guide on enriching CRM data for B2B sales teams covers the other half of the job.

TL;DR

• What it is: CRM data deduplication finds records for the same person or company, merges them, and prevents new copies.
• Why duplicates happen: list imports, form fills, integrations, manual entry, and enrichment tools that create records instead of updating them.
• How to find them: match contacts on normalized email, companies on normalized domain, then use fuzzy name, phone, and LinkedIn URL matching with review.
• Native tools: HubSpot blocks email and domain duplicates on all plans; its Manage duplicates tool needs a Professional or Enterprise hub. Salesforce uses Matching Rules plus Duplicate Rules and merges up to three records at a time.
• How to merge: pick a surviving record, decide field by field which value wins, keep all activity, and back up first. HubSpot merges cannot be undone.
• How to prevent: dedupe every list before enrichment and before CRM sync, not after.

What Is Data Deduplication in a CRM?

In a CRM, data deduplication means identifying contact, lead, and company records that refer to the same real-world entity and consolidating them into a single record. It has three parts: matching, merging, and prevention.

In storage and backup, the same term means removing repeated blocks of data to save disk space, as Wikipedia's data deduplication entry describes. This guide is about the CRM meaning only.

CRM duplicates are rarely exact copies. "Acme Inc" with acme.com and "ACME Corporation" with www.acme.com/ are the same company, but no exact-match rule will say so. That is why CRM deduplication depends on normalization and fuzzy matching, not just identical values.

Why Duplicate Records Pile Up

Duplicates are a symptom of too many systems writing to the CRM with no shared identity rule. Most come from five sources.

• List imports.A CSV from an event, a data vendor, or a rep's spreadsheet is imported without matching against existing records. Different column formats make the same person look new.
• Forms and inbound. A prospect fills a form with a personal email after a rep already created a record with their work email. The CRM sees two people.
• Integrations. Sequencers, meeting tools, chat widgets, and ad platforms each create records using their own matching logic. In Salesforce, the same person often exists as both a lead and a contact.
• Manual entry.Reps type "Jon Smith" at "Acme" without searching first. Typos, nicknames, and missing domains defeat exact matching.
• Enrichment and prospecting tools. Tools that push contacts into the CRM can create a new record when an email changes or a company name is spelled differently. People also change jobs, so one person ends up with two valid work emails.

Company records are especially fragile. Subsidiaries, regional domains, rebrands, and legal-name suffixes ("Inc", "GmbH", "Ltd") all create near-duplicates that look different to a string match.

What Duplicates Actually Cost You

Duplicates are not a cosmetic problem. They break the things the CRM exists to do.

• Split history. Emails, calls, and deals are spread across records, so no rep sees the full relationship.
• Double outreach. Two reps, or two sequences, contact the same person. That looks careless and raises unsubscribe and spam-complaint risk, which our email hygiene guide covers in depth.
• Broken routing and ownership. Round-robin and territory rules assign the same account to two owners.
• Wrong reporting. Pipeline, lead counts, and attribution are inflated, and conversion rates are diluted.
• Wasted spend. You pay to enrich, verify, and store the same person twice, and many CRMs bill by contact count.
• Weaker nurture. Lead scores split across records, so neither copy crosses the threshold. That quietly hurts any program built on lead nurturing tools.

Matching Rules: How to Find Duplicates

A matching rule defines which fields must agree, and how closely, for two records to count as the same. Start with the strongest identifier, then add weaker rules that send matches to review instead of merging them automatically.

Normalize every field before you compare it. Matching raw values misses most real duplicates.

1. Exact email match

Email is the strongest key for people because it is unique to one person. Lowercase it, trim whitespace, and compare the full address.

Watch for two traps. Role addresses like sales@ or info@ are shared, and one person can have both a personal and a work email, so email alone will not catch every duplicate person.

2. Company domain match

Domain is the strongest key for companies because names vary and domains do not. Strip the protocol, "www.", paths, and trailing slashes, so https://www.acme.com/about and acme.com match.

Records with only a company name need a domain first. Resolving names to domains turns "Acme Inc", "ACME Corp." and "Acme Corporation" into one matchable value, which is exactly what the Company Domain enrichment does. Check for subsidiaries and regional domains before merging, because acme.de and acme.com may be separate accounts on purpose.

3. Fuzzy name match

Fuzzy matching scores how similar two strings are instead of requiring them to be identical. It catches "Jon" and "John", typos, swapped first and last names, and "VP" versus "Vice President".

Salesforce's standard rules use algorithms such as Jaro-Winkler and Metaphone 3 for names, according to its Trailhead duplicate management module. Always pair a fuzzy name with a second field, such as company domain, and route the result to human review.

4. Phone and LinkedIn URL match

Format phone numbers to E.164 (for example +14155550123) before comparing, or country codes and dashes will hide matches. Shared office lines create false positives, so treat phone as supporting evidence.

A LinkedIn profile URL identifies one person even after a job change. Normalize it to the /in/handle form and use it to link old and new work emails.

Match keys at a glance

Native Deduplication in HubSpot and Salesforce

Both major CRMs can find and merge duplicates without extra software. Their limits matter, though, because they decide how much cleanup you can do natively.

HubSpot duplicate management

HubSpot automatically deduplicates contacts by email and companies by primary domain on every plan, per its deduplication documentation. Imports can also match on Record ID or on custom properties set to require unique values.

For duplicates that slip past those keys, the Manage duplicates tool suggests contact and company pairs. According to HubSpot's manage duplicates article, it needs a Professional or Enterprise subscription to Marketing, Sales, Service, Data, or Smart CRM.

• Default rule: contacts are compared on first name, last name, email, IP country, phone, zip code, and company name. Companies are compared on domain, name, country, phone, and industry.
• Volume: up to 10,000 pairs for individual review. Bulk management needs Data Hub Professional (30,000 pairs) or Enterprise (100,000 pairs).
• Custom rules: Data Hub Professional or Enterprise only, up to two rules per object with up to nine properties each.
• Refresh: suggestions update as records are created, with a daily scan otherwise.

Merging is available on all plans. Per HubSpot's merge records article, the primary record's values win unless they are empty, and you can override that property by property.

• The oldest create date and the furthest-down-funnel lifecycle stage are kept.
• The secondary contact's email becomes a secondary email; a secondary company domain becomes a secondary domain.
• Associations and timeline activity from both records move to the merged record.
• Records involved in 250 or more combined merges cannot be merged again.
• There is no unmerge. Rejected duplicate pairs, however, can be restored within 14 days.

Salesforce Duplicate Rules and Matching Rules

Salesforce splits the job in two. Matching Rules decide whether two records are duplicates, and Duplicate Rules decide what happens when a user creates or edits one.

• Standard matching rules cover business accounts, contacts, leads, and person accounts, using fuzzy logic on names, titles, and addresses (Trailhead).
• Custom matching rules let admins match on any field, such as a normalized phone or a custom domain field. A rule does nothing until it is activated.
• Duplicate rules either block the save or warn the user, who can then proceed. Each duplicate rule can use up to three matching rules, and each object can have up to five active matching rules.
• Existing records are not cleaned by duplicate rules. Org-wide scans use duplicate jobs, available in Performance and Unlimited editions.
• Merging combines up to three records at a time; you choose the master record and the value to keep for each field (Salesforce Help).

Native tools work well for steady-state hygiene. Large backlogs, cross-object matching (leads against contacts), and company-name variants usually need a normalization pass before the native tools can catch them.

Merge Strategies: Deciding What Survives

A merge strategy, often called survivorship, decides which record survives and which value wins for each field. Write it down before you merge anything, so every merge follows the same logic.

1. Pick the surviving record. Choose the one with the most activity, an open deal, or the oldest create date. Keeping the record that owns the history protects reporting and attribution.
2. Set field-level rules. Verified email beats unverified. The most recently updated phone and title win. Owner follows the record with the open opportunity.
3. Never drop history. Activities, associations, and deals should combine, not overwrite. Both HubSpot and Salesforce do this by default.
4. Keep old identifiers. Store the losing email or domain as a secondary value so future imports still match.
5. Separate auto-merge from review. Auto-merge only exact email and exact domain matches. Send fuzzy matches to a person.

How to Deduplicate a CRM, Step by Step

This process works for a first cleanup and for a quarterly hygiene pass. Run it on contacts and companies separately, companies first.

1. Export and back up. Pull contacts, leads, and companies with record IDs, owners, create dates, and activity counts.
2. Normalize. Lowercase emails, strip domains to the root, format phones to E.164, trim whitespace, and remove legal suffixes from company names.
3. Fill missing match keys. Resolve company names to domains and add LinkedIn URLs where you can. A record without a key cannot be matched.
4. Match companies on domain. Group accounts by normalized domain, check for parent and subsidiary relationships, then merge.
5. Match contacts on email, then LinkedIn URL. Merge exact matches in bulk.
6. Review fuzzy matches. Name plus domain and phone matches go to a reviewer with both records side by side.
7. Merge using your survivorship rules. Use the native merge tools so activity and associations move correctly.
8. Turn on prevention. Enable Salesforce duplicate rules or HubSpot unique-value properties, and fix the upstream source that created the most duplicates.

Once records are unique, refresh the fields that went stale. Our list of CRM data enrichment tools compares the options.

How to Stop New Duplicates: Dedupe Before You Enrich

The cheapest duplicate is the one that never reaches the CRM. Prevention means checking every incoming list against itself and against existing records before anything is enriched or synced.

Order matters. If you enrich first, you pay to find an email for the same person twice, and two enrichment runs can write conflicting titles or phones into two records.

1. Dedupe the list against itself on LinkedIn URL, email, and domain.
2. Dedupe against the CRM so existing contacts are updated, not recreated.
3. Filter to your ICP so you only enrich records you will actually work.
4. Enrich the unique records, ideally with waterfall enrichment so each contact is looked up once across several providers.
5. Sync with an update-first rule, matching on email or domain before creating anything new.

The same rule applies to account research. If you pull accounts from several sources, as in our company research tools roundup, or segment them by stack with technographics, resolve every account to a domain before it touches the CRM.

How SyncGTM Keeps Duplicates Out

SyncGTM is built around the dedupe-before-enrich order above. It cleans lists before they cost credits or reach the CRM, and it flags existing duplicates for your ops team rather than merging them inside the CRM for you.

• Name-to-domain resolution. The Company Domain enrichment resolves company names to their domains. Companies with the same domain but different name spellings can then be identified and merged.
• Dedupe on entry. The GTM automation platform backfills missing fields, dedupes records on entry, and re-enriches on a schedule so HubSpot and Salesforce data stays current.
• Dedupe before credits are spent. GTM agents such as Warm Leads From Posts deduplicate by LinkedIn URL and apply your ICP filter before spending enrichment credits.
• Duplicate checks on CRM data. The HubSpot and Salesforce lead sources pull records into SyncGTM, run deduplication and data-quality checks, and flag duplicates for ops to resolve in the CRM.
• 2-way sync. Clean, enriched records sync back to HubSpot and Salesforce, the two CRMs with 2-way sync.

You can start free and see per-credit costs on the pricing page.

Final Thoughts

CRM data deduplication is one cleanup project and one ongoing habit. The project is matching on email and domain, reviewing fuzzy matches, and merging under written survivorship rules.

The habit is what keeps the CRM clean: dedupe every list before you enrich it, and update existing records before creating new ones. Do that, and the native HubSpot and Salesforce tools only have to catch the stragglers.