What Is Bad Data in a Dealership DMS and CRM?

The short answer

Bad data in a dealership is any customer, vehicle or transaction record that cannot be trusted to do the job you are asking it to do. Most of it does not look broken. It is a record with the right name and the right vehicle that fails the moment you try to email the customer, match them across your DMS and CRM, or hand the record to an AI tool.

In the stores we work with at QoreAI, 40 to 60% of customer data is bad or duplicated. That is the share of your database that every campaign, every tool and every AI agent is working against.

That matters more now than ever. A person can spot a bad record and work around it. AI cannot. It takes what is in the field and acts on it.

Why bad data rarely looks bad

When dealers hear "bad data," they picture garbage: blank fields, misspelled names, records anyone would flag. That is the easy kind. It is also the small kind.

The record that hurts you looks fine. The name is correct. The phone number is correct. The vehicle is correct. The email field holds something a salesperson typed years ago to get past a required box.

So here is the definition I use with dealer groups. A record is not good or bad on its own. It is good or bad for a purpose. A record with a working mobile number and no email is fine for a BDC call. It is useless for an email campaign. The question is never "is this record complete?" The question is "can this record do the job in front of it?"

I have spoken with well over 200 dealership groups in the past year. Almost none of them could tell me what share of their records would pass that test.

The eight kinds of bad records in a DMS and CRM

Every franchised new car dealership has all eight. The only difference between stores is the mix.

Notice that only one of these, miscoded fields, is what most people mean by "dirty data." The other seven are records that look clean.

For a closer look at duplicates across systems, see One Customer, Four Records. For why contact data goes stale, see Why Your Dealership CRM Data Goes Bad.

What bad data costs once AI is working on it

Bad data was a tax before AI. With AI it becomes a multiplier. A good BDC agent sees none@none.com and skips it. An AI agent sends the email. The AI is not wrong about what the record says. The record is wrong, and the AI repeats the mistake at scale, with full confidence. When 40 to 60% of the records are bad or duplicated, that is wasted spend, wrong answers that sound right, and customers who blame your store, not your database. We break down the failure pattern in AI on Bad Data Is Just Fast Bad Decisions and the dollars in What Does Dirty Data Cost a Dealership?.

How to measure it this week

You do not need a vendor or a project to find your number. You need a sample and an afternoon.

1. Pull your active customers. Everyone with a deal or a repair order in the last 24 months. Ignore the rest for now.
2. Take a random sample of 200. Random matters. Do not let anyone pick the records.
3. Ask five questions of each record.
   • Is the email a real address for this person?
   • Is the mobile number real and current?
   • Does this customer appear only once across the DMS and CRM?
   • Do they still own the vehicle on file?
   • Do you know their consent status for email and text?
4. Score it against the job. A record passes only if it would work for the channel you plan to use.
5. Divide the passes by 200. That percentage is what every dollar and every tool in your store has to work with. If you land between 40 and 60% failing, you are typical.

One more thing to watch. Notice how hard it is to get that export out of your DMS and CRM. If pulling your own customer records takes a support ticket and a fee, you have learned something about who controls your data. To keep the decay from coming back, see How Bad Is Data Decay, Really?.

Frequently asked questions

Is bad data the same as missing data? No. Missing data is easy to see and easy to count. Most bad data is present and wrong: a placeholder email, a traded vehicle, a duplicate customer. Those records look complete and fail in use.

How much dealership data is bad? In the stores QoreAI works with, 40 to 60% of customer data is bad or duplicated. A 200-record random sample scored against five questions will give you your own number in an afternoon.

Can AI clean my dealership data for me? Partly. AI is good at finding duplicates, placeholders and conflicts between systems. It cannot know a customer moved or sold a vehicle unless it has an outside source to check against. Clean the foundation first, then put AI on top of it.

How often should a dealership clean its data? Continuously. Every deal and every repair order creates new records and new errors, so a one-time cleanup starts to decay the day it ends.

Do I need to clean my entire database? No. Start with the customers you market to and serve today. Cleaning records nobody will ever use is cost with no return.

Who should own data quality in a dealership? One named person, chosen by the dealer or GM. In most stores it belongs to nobody, which is how it got this way.

What is the difference between DMS data and CRM data? The DMS holds the transactions: deals, repair orders, parts and accounting. The CRM holds the relationship: leads, contacts and follow-up. The same customer lives in both, often with different details, and that gap is where most bad records come from.

Related reading: How Do I Get My Dealership Data Ready for AI?

Todd Smith is the Founder and CEO of QoreAI and the author of The Intelligent Dealership. He has 36 years of retail and technology experience.