Every AI vendor pitching your dealership will demo well. The chatbot answers fast, the voice agent sounds sharp, the dashboard looks clean. None of that tells you what matters most: whether you will still control your own operation two years from now. Use this checklist to look past the demo and judge a vendor on the things that decide whether the deal is good for you or just good for them.
Data ownership: do you keep what the AI generates?
Start here, because everything else follows from it. Ask directly: who owns the conversations, the customer records, and the insights the AI produces? Not who stores them, who owns them. If the vendor owns the data or claims broad rights to use it, you are building their asset with your customers. You want a clear answer that the data is yours, in writing.
Watch for the soft version of a no. A vendor who says you own your data but then reserves the right to use it in aggregated or anonymized form is keeping the valuable part. The patterns drawn from thousands of your customer conversations are worth more than any single record, and a broad rights grant hands that value to them while you take the risk of collecting it.
Portability: can you take it with you?
Owning data means little if you cannot get it out. Ask how you export your full history, in what format, and how fast. Can you pull your customer records, call logs, and conversation history in a usable form on demand? If the only way out is a PDF report or a support ticket that takes weeks, the data is hostage even if the contract says it is yours.
A good test is to ask for a sample export during the trial. A vendor built for portability can hand you a clean file of structured data in an hour. One built to trap you will stall, quote a fee, or send back a locked format you cannot load into anything else. What they do with that request tells you more than any slide in the pitch.
Open standards: is it built to connect or to trap?
Ask what the system is built on. AI built on open standards like MCP is designed to connect with your other systems and to let you swap pieces without tearing everything out. A proprietary, closed platform is designed to make leaving painful. The more the vendor talks about their all-in-one ecosystem and the less they talk about interoperability, the more locked in you will be.
The all-in-one pitch sounds efficient, and that is the trap. When your CRM, your chat, your voice agent, and your reporting all live inside one vendor’s walls, switching any single piece means switching all of them. A store that later wants a better CRM finds it cannot move without unplugging the whole stack, so it stays and pays. Open standards keep each piece replaceable on its own.
Audit trail: can you see what the AI did?
You are responsible for what your AI says to customers, so you need to see it. Ask whether you get a full, reviewable record of interactions, decisions, and the data behind them. Can you show, after the fact, exactly what the system told a customer and why? Without that visibility you cannot manage compliance, settle a dispute, or trust the tool with anything that matters.
This turns concrete the day a customer claims your AI quoted them a price you never approved. With a real audit trail, you pull the exact conversation and see what was said. Without one, it is their word against a black box, and you are guessing about what your own system did.
Exit terms: what happens the day you leave?
Read the offboarding terms before you sign the onboarding ones. Ask what happens to your data when the contract ends, how long you have to retrieve it, whether the vendor deletes their copies, and what it costs to get out. A confident vendor makes leaving clean because they expect you to stay by choice. A vendor who buries steep exit costs and vague data-return language is telling you how the relationship really works.
How to weigh the answers
Run every vendor through these five questions and a pattern appears fast. The ones building on ownership, portability, open standards, transparency, and clean exits give you straight answers because those answers are good for them too. The ones running a walled garden get vague, redirect to the demo, or lean on trust-us assurances. You are not just buying a feature. You are deciding who controls your dealership’s data and how easily you can change your mind. Judge accordingly.
Frequently asked questions
What if a vendor gives good answers verbally but will not put them in the contract?
Treat the contract as the only answer that counts. Verbal assurances from a salesperson do not survive a renewal, an acquisition, or a dispute. If ownership, export format, and deletion on exit are real commitments, a vendor will write them down. If they will only say them out loud, you have your answer.
Is an all-in-one platform ever the right call for a smaller store?
It can be simpler to run, and simplicity has value. The risk is that convenience today becomes lock-in tomorrow. If you do choose an all-in-one, insist on the same portability and exit terms you would demand of any vendor, so that convenience does not quietly become a cage you cannot leave.
Which of the five questions matters most?
Data ownership, because the rest depends on it. If you do not truly own the data and the intelligence built on it, portability, audit trails, and clean exits have little to work with. Start there, get it in writing, and the other four become far easier to enforce.
This is the kind of capability you should own, not rent. See how VCTRS gives dealers AI built on context you own on our AI for car dealerships page.