Duplicates are the one CRM problem every admin eventually meets. Two reps call the same customer on the same day. A report says you have more accounts than you thought. A contact’s history is split across three records, none of which tells the whole story. They accumulate slowly, so nobody notices until the damage is visible.

Where Duplicates Come From

Knowing the source tells you where to focus prevention.

• Manual entry. A rep creates a contact without searching first, or searches by name and misses a variation.
• Imports. A spreadsheet loaded without a matching step creates a fresh record for every row.
• Web forms and lead capture. Each submission creates a new lead, even when the person already exists.
• Integrations. Two connected systems each create a record for the same person with no shared identifier.
• Mergers of data. Acquisitions or team handovers that bring in separate lists.
• Name variations. “Acme Ltd”, “Acme Limited” and “ACME” are all the same company to a human and three accounts to a database.

In most organizations, imports and forms create more duplicates than people do, so those are the first places to put controls.

Define What “The Same” Means

Before running a duplicate scan, write down matching rules. For contacts, the strongest key is usually the email address. Name plus company is a weaker secondary signal, and phone number weaker still since they are often shared. For accounts, the website domain tends to beat company name. Decide what combination counts as a certain match, which needs human review, and which you will ignore.

Be cautious with fuzzy matching. It catches variants such as “Jon” and “John”, but it also flags unrelated people with common names. Treat fuzzy results as suggestions to review rather than candidates for bulk merge.

Merge Without Losing History

Merging is permanent in most systems, so approach it carefully.

1. Pick the survivor. Common criteria are the record with the most recent activity, the one linked to open deals, or the oldest one with the most complete history. Apply the same rule every time.
2. Check related records. Look at what is attached to each: deals, cases, activities, notes, campaign membership. Understand how your platform treats them in a merge, and check the vendor’s current documentation.
3. Resolve conflicting fields. When two records disagree on the phone number or title, decide which is more current. Do not just let the survivor’s value win by default.
4. Preserve external IDs. If records in other systems point at the loser’s ID, make sure that link survives or is updated.
5. Back up first. Export the affected records before a bulk merge.
6. Spot check afterward. Open a few merged records and confirm the timeline and relationships look right.

Choose the Right Tool for the Scale

Situation	Suggested approach	Caution
A few duplicates now and then	Manual merge from the record	Train users to search first
Hundreds from an old import	Report of matches, review, merge in batches	Back up before batch merge
Ongoing duplicates from forms	Matching rule at creation, update instead of create	Test with edge cases
Duplicates across connected systems	Shared identifier and system of record	Decide ownership of each field
Very large legacy clean-up	Dedicated dedupe tool or consultant	Verify tool behavior in a test copy

Prevent the Next Wave

Cleaning without prevention is mopping with the tap running. Layer your defenses:

Turn on duplicate alerts or blocking at creation, if your CRM supports it. Warning a rep that a similar contact exists when they click save catches the most common mistake. Use blocking only for exact matches on strong keys; otherwise use a warning so legitimate records are not rejected.

Make imports update, not just insert. Use the CRM’s import matching option, keyed on email or an external ID, so existing records are updated instead of duplicated.

Standardize inputs. Lowercase emails, trim whitespace and use a consistent format for company names where feasible. Picklists and validation rules reduce the variants that fool matching.

Give forms smarter behavior. Check whether the platform can match a submission to an existing contact rather than creating a new lead.

Educate users in one sentence. “Search by email and company before creating anything.” It is a surprisingly effective rule.

A Hypothetical Example

An admin at a small distributor runs a report grouping contacts by email address and finds a few hundred addresses appearing twice or more. Most trace back to a trade show list imported twice by different people, with one copy owned by a rep who has since left. She exports the affected records as a backup, defines the survivor as the one with the most recent activity, merges them in batches of fifty while spot checking each batch, and then turns on duplicate warnings for new contacts. She also changes the import template to match on email. A month later she runs the same report and finds only a handful of new duplicates, all created by one integration, which she traces and fixes at the source.

Make It a Routine

Put a duplicate check on the monthly maintenance calendar. A saved report that groups contacts by email and accounts by domain, with a count, takes minutes to review. Track the number over time; a rising count points to a new source worth investigating.

Frequently Asked Questions

Is it safe to merge automatically? For exact matches on strong keys, usually yes once you have tested the rules. For fuzzy matches, review by a human first.

Can a merge be undone? In many systems it cannot, or only partially. Assume it is permanent, export a backup, and check your vendor’s current documentation.

Which record should survive? Whichever has the most useful history and links, applied with a consistent rule. Copy over any better field values from the other record before merging.

How do duplicates affect reporting? They inflate counts, split activity history and distort metrics like leads per rep or accounts per region. They can also trigger duplicate marketing sends.

Should we delete or merge? Merge when both records have history worth keeping. Delete only true junk, such as empty test records, after confirming nothing references them.

Next Step

Build one report that groups contacts by email and shows any with a count above one. Review the top twenty and agree on your survivor rule before you merge a single record.