Most dealerships carry duplicate records on 20 to 40 percent of the customers in their DMS, and bad data overall costs stores 3 to 5 percent of gross revenue annually. On a store grossing $10 million, that is $300K to $500K a year leaking through a problem most operators have never measured once.
Duplicates sound like a clerical nuisance. They are a tax on every department.
How one customer becomes four records
Nobody decides to create duplicates. The architecture creates them.
A customer buys a car and sales enters them. Two years later they hit the service lane, the advisor can't find "Bob Smith" under the number on file, and creates "Robert Smith." Marketing imports a lead list and adds a third version with a different email. A spouse co-signs and the household splits into separate identities. Four systems, four spellings, four partial histories, one human being.
Multiply by twenty years and two DMS conversions, and a 30 percent duplication rate is not negligence. It is the default outcome of running disconnected systems with no identity resolution between them.
Where the money leaks
You pay twice to acquire customers you already own. The duplicate looks like a fresh prospect, so they get conquest offers, paid retargeting, and BDC labor, all spent re-winning a relationship already on the books.
Trust erodes in public. The same household gets a conquest offer, a loyalty offer, and a "we miss you" letter in the same week, each contradicting the others. The customer concludes the store does not know them. They are correct.
Histories fragment, and gross goes with them. Half the service history sits on one record, the purchase on another. The advisor cannot see declined work from last visit. The desk cannot see the full relationship when structuring the deal. Every fragmented record makes every employee touching it slightly dumber.
Every report you trust is wrong. Customer counts inflated. Retention understated, because the "lost" customer is active under another ID. Per-customer economics calculated against a phantom denominator. You can't improve what you can't accurately measure.
Equity mining and AI misfire. The tools surface a duplicate with no trade history as a poor prospect while the real opportunity hides on the twin record. Garbage identity in, garbage intelligence out.
Pull your number this Monday
This is the first move in any serious data effort: audit your DMS for duplicate customer records. Pull the number. It will shock you.
A rough first pass takes one query: count exact matches on phone or email across customer records. That undercounts the real problem, since the worst duplicates have mismatched contact info, but even the undercount usually lands north of 15 percent and makes the point in one leadership meeting.
The real fix is identity resolution, not a one-time merge party. Records validated against third-party sources. Households resolved, not just exact-match merged. And one owned data layer where every system reads and writes against a single identity, so the duplicates stop regenerating the day after the cleanup.
That number, your duplication rate, is the baseline for the entire intelligence build. Everything else in the stack inherits it. Know it before the next vendor conversation, because every tool you are about to be pitched will perform exactly as well as the identity layer underneath it.
FAQ
How common are duplicate records in dealership DMS systems? Most stores run 20 to 40 percent duplication. Rates below 10 percent are rare without active identity resolution.
How much does bad data cost a dealership? 3 to 5 percent of gross revenue annually, through duplicate acquisition spend, fragmented histories, missed service revenue, and bad attribution.
Why do duplicate customer records happen? Disconnected systems each create their own version of the customer: sales, service, marketing imports, and DMS conversions all add variants with no identity resolution between them.
How do I find my duplication rate? Query the DMS for exact matches on phone or email across customer records. That undercount alone typically exceeds 15 percent and establishes the baseline.
Does merging duplicates fix the problem? A one-time merge helps briefly. Without an owned identity layer that all systems read and write against, duplicates regenerate immediately.
Todd Smith is the Founder and CEO of QoreAI and the author of The Intelligent Dealership: How AI and Data Transform Automotive Retail.