Updated October 2026\n\nA dealership data audit comes down to three moves: pull your DMS duplication rate, test your data export rights with every vendor, and define "customer" in writing with leadership sign-off. All three can start Monday morning and finish inside a week. Together they produce the baseline every data, AI, and automation decision should be built on.
No consultants, no new tools, no budget request. Just three uncomfortable numbers and one settled definition.
Step 1: Pull your duplication rate
The move: Audit your DMS for duplicate customer records. Pull the number. It will shock you. Most stores find 20 to 40 percent duplication.
How to run it Monday: Have your DMS admin or IT contact run a match query on customer records: exact matches on phone number, then on email. That's the floor, not the ceiling, since the nastiest duplicates have mismatched contact info. Even the floor makes the case.
What you'll learn: Whether your customer counts, retention numbers, and per-customer economics are real. At 30 percent duplication, every report your leadership team reads is meaningfully wrong.
Write down one number: duplicate percentage. That's your baseline. Every future cleanup effort gets measured against it.
Step 2: Test your data rights
The move: Ask your DMS and CRM vendor one question: "Can I export my full data set?" Watch the reaction. The answer reveals who actually owns your customer relationships.
How to run it Monday: One email per vendor holding customer data. DMS, CRM, marketing platform, service scheduler, equity tool. The exact question, no softening. You want everything, structured and usable, and you want their terms for providing it.
How to score it: A clean yes with no fee and fast turnaround means you own your data. Export fees, 30-to-60-day timelines, flat files, partner-only access, and "proprietary" enriched fields each mean you own less than you think. Every "but" in their answer is a fence post.
Write down one answer per vendor: yes, yes-with-conditions, or no. That grid is your data rights map, and it belongs in every renewal negotiation from now on.
Step 3: Define "customer" in writing
The move: One definition of "customer," agreed by leadership, in writing. Most stores can't answer this the same way twice.
How to run it Monday: Ask your GM, sales manager, service manager, and marketing lead to each write one sentence defining an active customer. Compare the sentences. They will not match. Someone counts anyone in the CRM. Someone counts 24-month transactors. Someone counts service-only households, someone doesn't.
Then settle it in one meeting. Pick the definition, document the edge cases (households, co-buyers, service-only, fleet), and have leadership sign it.
Why this is step three and not a footnote: Every system, report, and AI tool you deploy inherits this definition. If the store can't agree on what a customer is, no platform can resolve identity correctly, because the disagreement gets built into the configuration. Settle it this week; everything else depends on it.
What you have at the end of the week
Three artifacts: a duplication rate, a data rights map, and a signed definition of customer.
That's the foundation audit. It tells you the quality of what you have, the control you have over it, and whether your own leadership agrees on what it means. Months one to three of any serious AI build start exactly here, which means a store that runs this audit now is a quarter ahead of one that starts when the first vendor contract is already signed.
The most expensive thing in your store is what you don't know. This is how you start knowing.
The 10-point AI readiness checklist
Before you buy another AI tool, score your data. Answer yes only if it is true today, not planned.
| # | Question | What "yes" looks like |
|---|---|---|
| 1 | One record per customer? | Duplicates across the DMS and CRM are merged into one customer, not just flagged |
| 2 | One record per vehicle? | Each VIN is tied to its owner, its deal and its full service history |
| 3 | Same definitions everywhere? | Sold date, gross and customer status mean the same thing in every system |
| 4 | Updated daily? | Records refresh every day, not from a monthly export |
| 5 | Departments connected? | The sales desk can see a customer's service history, and service can see their deal |
| 6 | Contact data verified? | Phone numbers and emails are verified and current |
| 7 | Consent in one place? | An opt-out in any tool reaches every other tool |
| 8 | Full history reachable? | Years of deals and repair orders, not the last 90 days in a report |
| 9 | A copy you hold? | A full copy of your data outside any single vendor, readable without that vendor's software |
| 10 | Exit terms in writing? | You know what data you keep when you cancel each vendor, and in what format |
How to score it:
- 9 to 10: Ready. AI tools will have a clean foundation. Focus on which decisions to hand them first.
- 6 to 8: Close. AI will work in one department and give conflicting answers across departments. Fix the gaps before you add tools.
- 0 to 5: Foundation first. Buying AI now means paying for confident answers built on bad data. Clean and own the data before anything else.
Start with questions 1, 9 and 10. One clean record per customer makes every other tool better, and owning the copy means you only do the work once. For a fuller score, take the AI Readiness Assessment.
See Why Dealership AI Gives Different Answers for what happens when these foundation checks are missing.
FAQ
How do I audit my dealership's data quality? Three steps: pull your DMS duplication rate, test export rights with every vendor holding customer data, and get leadership to define "customer" in writing.
What duplication rate should I expect in my DMS? Most stores find 20 to 40 percent duplicate customer records on first audit. Anything over 10 percent materially distorts reporting.
How long does a dealership data audit take? The baseline version runs inside one week: one DMS query, one email per vendor, one leadership meeting.
What is the data rights test? Asking each vendor "Can I export my full data set?" and scoring the response. Fees, delays, flat files, and excluded fields reveal who actually controls your customer data.
Why does defining "customer" matter? Every report and every AI tool inherits the definition. If leadership can't state it consistently, identity resolution fails downstream by design.
Todd Smith is the Founder and CEO of QoreAI and the author of The Intelligent Dealership: How AI and Data Transform Automotive Retail.
Related: Who Owns Your Dealership Data? walks through the DMS contract clauses that decide whether you actually control the data you audited.
Related: The Dealership AI RFP: 15 Questions Every Vendor Must Answer in Writing turns this audit into procurement leverage before you sign the next contract.
Free resource: The Dealer Data Addendum is an ungated set of eight contract clauses (data ownership, export rights, deletion, schema-change notice, audit rights) you can hand to your attorney and attach to any vendor agreement.