The acronyms are a sales tool, not a technology
GEO. AEO. AI SEO. LLMO. Someone invents a name so they can put a new line item on your invoice. Strip the labels off all of them and they describe one thing: when a shopper asks a machine a question, does your store come up? That is the whole idea.
Here is a test that works every time. Ask the vendor what will physically change on your website. If they cannot tell you in one plain sentence, they do not know either — they are repeating something they read. Anyone who leads with the acronym instead of the outcome is selling you vocabulary.
What actually happens when a shopper asks a chatbot about your store
There is no mystery in it. The machine goes out, reads web pages, and writes an answer from what it found. That is the trick. It is a very fast reader with no opinion of its own.
So the real question is not whether you are optimized for AI. It is three plain ones: can it find your pages, can it read them, and does it trust them enough to repeat them? Everything else is noise.
If your website is a mess, none of this matters
The foundation is the same as it has always been. Pages that load, that a machine can read, that say plainly what they are about.
Where dealers lose this, over and over: every useful detail is locked inside an inventory tool a machine cannot open. The service page is one paragraph that says nothing. Hours and departments live inside a picture, and a machine cannot read a picture. The finance page says it works with all credit types and then stops.
If your service page does not say what you actually do, roughly what it costs and how to book it, you are not going to be the answer. Not because of AI. Because there is nothing there to read.
Answer the question in the first sentence
Dealers bury the answer. Three paragraphs about a commitment to the community, then maybe the thing the person asked.
Someone asks whether they can trade in a car they still owe money on. Start with: yes, and it is normal — most trades still have a loan on them. Then explain how it works, what happens when you owe more than it is worth, and what to bring with you.
Machines lift the direct answer. So do people. Nobody in the history of car buying has been annoyed at getting the answer too quickly.
The internet has to agree that you exist
Your website says you close at 7. Google says 6. The manufacturer locator has your old phone number. Some directory still has the address from before you moved.
A machine sees all four, cannot tell which is true, and recommends a store it is more sure about. You did not lose on quality. You lost on housekeeping.
The fix is boring and it works. Settle on your name, address, hours, phone and departments. Make every listing match exactly, down to whether you are Street or St. Then leave it alone until something actually changes.
Write the things only you can write
Ten tips for buying a used car — there are a million of those and the machine has read all of them. You add nothing by writing the million and first.
What nobody else has: what financing actually looks like for buyers in your market. Which trims move in your area and which sit. What your service department genuinely does better than the shop down the road. What a first-time buyer in your state actually needs to bring with them.
That is real material, and it is the stuff you already know and have never written down.
How to tell it is working, without getting conned
Write down fifteen to twenty questions a real shopper would ask. Not keywords — actual questions, the kind people say out loud. Ask them yourself once a month and write down whether you showed up.
That is a monitoring program. It costs you an hour a month and it is more honest than most of what gets sold.
If someone offers you an AI Visibility Score, ask what it measures and how it is calculated. There is no official scoreboard for this. Anyone implying otherwise has invented a number to justify a retainer.