Good Data Hygiene: How to Find and Merge Duplicate Customer Records

Duplicate customer records are the most common kind of dirty data in a dealership CRM. Two salespeople call the same shopper. A customer who asked to stop hearing from you gets a promotion because the opt-out sat on the other record.

This guide is for the sales manager, BDC lead or CRM administrator who has to clean that up: where duplicates come from, the matching rules that find them, a merge process that keeps history and consent, and the habits that stop new ones. For the field-by-field data standard and monthly checks, see our guide to dealership customer data quality.

Why duplicates happen in a powersports store

Most duplicates are not carelessness. They come from the number of doors a customer can walk through, and from systems that each create their own record.

• Web leads submitted twice. A shopper fills in a form on Monday, gets no reply they noticed, and fills in a different form on Wednesday with a work email.
• OEM and third-party leads. The same person can arrive from your website, an OEM lead program and a listing site within a week. Some listing sites pass a relay or masked email rather than the customer's own, so the record matches nothing you already have.
• Phone-ups and walk-ins. A quick call gets logged as "Mike, RZR question" with a mobile number and no last name. When Mike walks in, a new record is created with his full name and no phone.
• Service and parts counter. Write-ups and counter sales are entered fast, by people outside the sales process, often with a home number instead of a mobile.
• Trade-ins and units owned. A unit traded in can stay on the old owner's record while the new buyer's record also shows the same VIN.
• Households. A couple shares one email, a parent buys a youth ATV in a teenager's name, or two brothers both ride and list the family landline.
• Typos and formats. "Jon" and "John", "Mc Donald" and "McDonald", (555) 010-2234 and 5550102234 all defeat a simple search.
• Imports and multiple rooftops. Event sign-up sheets, old lists and a second store's customers get loaded without matching.

Fix the door that produces the most duplicates, and the cleanup gets smaller every quarter.

Matching rules to find duplicates

Matching is a judgment about how likely two records are to be one person. Before you compare anything, normalize the data: put every phone number into one ten-digit format, lowercase emails, trim spaces, and standardize addresses to the USPS format in Publication 28 so "123 North Main Street" and "123 N Main St" match.

Then sort candidate pairs by the strength of what they share.

Two rules keep this safe. Never merge on a weak match alone. And a strong match on contact details still needs the names to agree, because phones and emails are shared far more often than people think.

Many CRMs can run these rules and show likely pairs. Whatever tool you use, keep a person in the loop for anything below "strong".

The mistake most cleanup guides miss: false matches

Most advice on CRM duplicate records cleanup, from software vendors and consultants alike, says the same three things: run a duplicate scan, merge in bulk, then block duplicates at entry. That is sound, but it assumes that one phone number or one email means one person. In a dealership database, that assumption breaks in two predictable ways.

Placeholder values group strangers together. When a form or CRM makes email or phone required, staff who do not have the detail type something to get past the field: "none@none.com", "noemail@dealer.com", 000-000-0000, the store's main number. A duplicate scan then sees dozens or hundreds of "matches" on that value. Merge them in bulk and you have built one record out of many unrelated customers, with mixed-up units, deals and permissions. Before any automated matching, search your database for placeholder values, clear them, and exclude them from the matching rules.

Shared details are not shared identities. Two riders at one address are two customers with their own units, deal history and consent. A dad and son with the same name, address and landline look like a perfect duplicate. They are not.

The fix: placeholders get cleaned, not matched, and household members get linked or noted, not merged. Use a household or relationship field if your CRM has one, or a short note on each record ("shares phone with").

A safe merge process

Run every merge the same way, in this order.

1. Export both records first. In many CRMs a merge is one-way. HubSpot's help center, for example, states that "once you've merged two records, it is not possible to unmerge them." A saved export of both records is your only way back if you get it wrong.
2. Confirm it is one person. You need at least one strong match from the table above and agreeing names. If you are unsure, do not merge. Flag it and ask the customer at the next contact: "Is this still the best number for you?"
3. Choose the surviving record. Usually it is the record with the deal history, service history or the customer number your other systems point to. The newer record is not automatically the better one.
4. Decide field by field. Use the survivorship rules in the next table rather than accepting whatever the merge screen defaults to.
5. Keep all history. Notes, call recordings, texts, emails, appraisals, lost reasons and service visits from both records belong on the survivor.
6. Carry over the strictest preference. Any opt-out or do-not-contact flag on either record stays, with its original date and source.
7. Clean up open work. Close duplicate tasks and appointments, keep one owner, and check that the customer is not enrolled twice in the same follow-up or campaign.
8. Log the merge. Record who merged which records, when and why, so a wrong merge can be traced and fixed.

Which value wins

Limit who can merge. Merges are hard to reverse and touch consent, so the right to merge belongs to a small number of trained people, usually the CRM administrator and a sales or BDC manager. Our guide to the CRM administrator role at a dealership covers users and permissions.

Handling opt-outs during merges

A merge is when an opt-out is most likely to get lost. If the screen keeps the newer record's values by default, a fresh web lead can quietly switch a customer's messages back on.

Federal rules show why that matters. Under the FCC's TCPA rules at 47 CFR 64.1200, a customer may revoke consent to calls or texts "by using any reasonable method," and the rule says such requests "must be honored within a reasonable time not to exceed ten business days from receipt of such request." For company-specific do-not-call lists, "A do-not-call request must be honored for 5 years from the time the request is made." For commercial email, the FTC's CAN-SPAM guide says "You must honor a recipient's opt-out request within 10 business days."

So the merge rule is simple: keep the strictest preference, per channel, from either record. An email opt-out on one record and a text opt-out on the other means the merged customer has opted out of both. Never let a merge, import or integration reset an opt-out, and keep the date and source of each so you can show when it happened.

State laws can add more. This is general information, not legal advice. Check your own process with counsel.

How to stop new duplicates

Cleanup without prevention is a treadmill. These habits shrink the next one.

• Search before you create at every entry point, by mobile number, then email, then last name and ZIP code. The customer data quality guide sets out the full search rule and required fields.
• Make required fields honest. Give staff a "not provided" option. A blank field is better than a placeholder that matches hundreds of strangers.
• Match integrations before they create. Website, OEM and third-party leads should be checked by phone and email before a new record is made. Ask each vendor how its feed matches.
• Clean imports before loading. Remove duplicates inside the file, map every row to a lead source, match against the database, and never let an import overwrite permissions.
• Teach the household rule. A shared phone or email is not a reason to reuse someone else's record.
• Track where duplicates start. Note which entry point created each extra record you merge.

Quarterly duplicate cleanup checklist

Review flagged pairs weekly, keep to your monthly data checks, and once a quarter run a deeper sweep of the whole database. Use this checklist for the quarterly sweep.

Compare the counts each quarter. A rising number from one entry point tells you where to retrain or which integration to fix.

Where PSX Ascend fits

Clean data is a habit first, and software should make the right habit easy. Quality data comes first: more good data in the CRM gives AI more to work with, and AI on a broken process only speeds up the broken process.

PSX Ascend's AI-powered powersports CRM is built to help teams keep oversight of communication preferences across phone, text and email, with less repetitive manual entry. For dealer groups, the multi-location BDC software includes advanced record matching that brings customer activity and purchase history together before duplicate records create conflicting follow-up. The PSX process also requires quality notes and accurate dispositions when a sale does not happen, so the record holds the context that ALMA, the Automated Lead Management Assistant, uses to bring stalled prospects back to a live conversation with a person.

Clean records also make audience segmentation for powersports dealers and planning around common powersports customer types work, because each rests on one record per real customer.

You can also see the same AI-powered powersports CRM on our sister site, PSXDigital's AI-powered powersports CRM page.

Sources

• 47 CFR 64.1200, TCPA rules on revoking consent and do-not-call requests (Cornell LII)
• Federal Trade Commission: CAN-SPAM Act: A Compliance Guide for Business
• HubSpot Knowledge Base: Merge records
• USPS Publication 28: Postal Addressing Standards