← All posts

Everyone Is Plugging AI Into Your Dealership Data. Ask It These Six Questions First.

September 23, 2026 · Mark Hoffman

Everyone Is Plugging AI Into Your Dealership Data. Ask It These Six Questions First.

Every dealer software company I follow is announcing an AI connection this month. Hook your reports up to Claude or ChatGPT, ask a question in plain English, get an answer in seconds.

I'm for it. We built one at Dealer Insights, and our dealers are using it now. Asking "which of my vendors is slipping?" beats digging through six logins.

But after twelve years of pulling dealer marketing data together, here is what I know: an AI will answer any question you ask it, with total confidence, from whatever data it's handed. It doesn't know when the data is wrong. It doesn't know when a number is missing. It will not say "I'm not sure."

So before you trust an AI's answer about where your money goes, ask the tool behind it these six questions.

1.1.png

1. Does it know which of your traffic is people?

A big share of dealer website traffic is bots, and some of it arrives through paid clicks. Bots can even look great on time on site. If the AI counts bot sessions as shoppers, every rate it gives you is off, and it won't tell you.

2. Can it connect your agency's campaign names to what your website sees?

The name of a campaign inside Google Ads and the tag that shows up in Google Analytics are almost never the same. Someone has to map one to the other, store by store. If nobody did, "how is my truck campaign doing?" gets answered from the wrong rows.

3. Does it know the difference between what you paid and what was spent?

Your vendor's invoice and the spend inside the ad platform are different numbers, often for perfectly good reasons. An AI that puts one vendor's invoice next to another vendor's platform spend is comparing apples to oranges, and it will call one of them cheap.

4. Does it keep Vehicle Listing Ads apart from search ads?

"Google" is not one thing. Search ads, Performance Max, and Shopping campaigns behave differently and should be judged differently. Lump them together and the average hides the one that isn't working.

5. When data is missing, does it say so, or does it show you a zero?

This is the one that worries me most. A report that shows "0 lead conversions" because the tracking broke looks exactly like a vendor that stopped producing. An AI will happily suggest cutting it. Missing should say missing.

1.2.png

6. Does it know what you actually sold?

Your DMS knows. For vendors that send you lead records, those customers can be matched to your sales, so you can see which lead conversions became cars sold. For click-based ads, like Google, there is no lead record to match, and any tool that claims to tie a Google click to a DMS sale is guessing. A good answer tells you which is which.

1.3.png

The AI isn't the problem

None of these are AI problems. They're data problems, and they were here long before AI. The AI just delivers the wrong answer faster, and more convincingly.

That's the work we've done for twelve years: the plumbing nobody sees. Cleaning out bots, mapping campaign names, keeping invoices and platform spend straight, marking what's missing, and putting what you sold next to what you spent. Our AI connection is just the newest way to get at it.

We don't sell you an AI. We make yours useful.

Want to see how your store does on these six? Request a free audit and we'll walk you through it.

Want to see what your marketing data is really telling you?

Get a Free Audit