Being Chosen by AI and Being Seen by AI Are Two Different Things

A while ago, a friend told me a small story.

She was heading to Tennessee on a business trip and wanted a hotel near the airport. Her requirements weren't complicated: clean, relatively new, rated 8.5 or above, and no more than $130 a night with tax. In the old days, that job would have meant a dozen open browser tabs and a comparison marathon lasting past midnight.

This time she took a shortcut and just asked Gemini.

A few minutes later, the AI handed her 15 hotels. With photos, links, and a comparison table. Price, rating, distance from the airport — everything laid out, line by line, crystal clear.

Notice: the AI didn't just "find" these hotels.

It also "judged" them.

When I heard that, my heart sank a little. Think about it: if you were the marketer for one of those 15 hotels, is "did we show up in the AI's answer" really the right question anymore?

No.

The real question is: what judgment has the AI made about you?

Showing Up Is Just the Entry Ticket

Over the past couple of years, a new term has been popping up in corporate meeting rooms: GEO.

What is GEO? Generative Engine Optimization — in plain terms, making sure your brand can be understood, accurately described, and found in AI-generated answers.

Most companies are still figuring this out as they go. But consumers aren't waiting. They're getting more and more comfortable letting AI cut 90% of their options and hand them a single, tidy shortlist.

The interesting thing is, consumers aren't entirely comfortable handing over the reins either. Letting AI recommend a hotel? Fine. Letting AI log into your account, read your membership details, charge your saved credit card, and place the order? Most people still hesitate. That final act of "buying" — for now, they want to keep it in their own hands.

But the infrastructure is already being paved. OpenAI has built the Agentic Commerce Protocol; Google has built the Universal Commerce Protocol. Both are open standards, designed to let AI glide through the entire sequence — browsing products, placing the order, paying, managing the booking.

Consumers aren't fully ready, but the track has already been laid out.

So the question left for every brand is really quite plain: when the AI comes to look, to cross-check, and to recommend, what does the "you" it sees look like?

AI Doesn't Understand "Close Enough"

Let me give you a common scenario.

A hotel lists one price on its booking engine and another on its official website. The amenities list clearly says "breakfast included," but when guests arrive at the front desk, they're told breakfast costs extra. One channel says checkout is at 11, another says 12.

Can humans make sense of this? Barely. People call to confirm, people compromise.

AI won't.

AI doesn't understand "close enough," doesn't understand "it depends," doesn't understand "you know how it is." It sees three channels with three different stories and draws exactly one conclusion: this business's information is unreliable. Then it turns around and recommends the competitor whose information is spelled out clearly.

This is where a lot of marketers haven't made the mental shift. For years, our trade has run on storytelling, on emotion, on "a whole new stay experience."

AI doesn't listen to stories.

What AI does is take what you say about yourself and reconcile it against guest reviews, ratings, and real photos. Your brochure says "newly renovated," but the photos guests post still show that old carpet — that's not copy that needs polishing. That's a claim and a fact that don't match.

So the marketer's job has changed. It's no longer just optimizing content so it makes it into the AI's answer — it's making sure the facts behind that answer can withstand the AI's repeated cross-checking.

Four Questions to Replace Your Old Dashboards

In the past, to measure performance, you'd open your web analytics tool and look at traffic and bounce rates. Going forward, that won't be enough. You need to swap in four questions to ask yourself:

First, showing up. In the relevant AI searches, am I there?

Second, making the shortlist. Does the AI describe me accurately? Did I make the shortlist?

Third, being the pick. Once the user's criteria are applied, am I the one that gets recommended?

Fourth, closing the deal. Did the user actually show up, add to cart, ask for a quote, place an order?

These four steps form a chain, and not one link can break. Showing up without making the shortlist gets you nothing; making the shortlist without closing the deal gets you nothing too. And to see all four steps clearly, no single tool can do it — you have to stitch together AI visibility testing, transaction data, user research, review analysis, on-site behavior, and customer service feedback.

A hassle? Absolutely.

But the scene where purchase decisions happen has already moved onto the AI's comparison table. If you don't go look, you're running your business with your eyes closed.

What to Do Now: Three Plain Things

So concretely, what do you do?

First, get the people in the room. The folks who manage information, the folks who manage experience, and the folks who manage transactions — three groups who may barely talk to each other normally — now have to sit at one table. Ideally, pull in a senior executive to lead, so that SEO, content, PR, product, commercial, tech, legal, experience, and measurement settle into a fixed rhythm of collaboration.

Then, honestly work through three questions:

Is your information understandable? Are prices, inventory, and terms up to date, complete, and consistent across every channel?

Are your claims verifiable? Do the reviews, ratings, and photos guests take actually back up what you brag about?

Is your process executable? Can a customer who arrives via AI — or an agent running errands on a customer's behalf — get to the next step without detours, without manual workarounds, without hidden fees?

Finally, designate one person whose whole job is to keep watch on two things: whether the information stays accurate, and whether the experience lives up to the claims.

You'll notice that none of these three things is sexy. It's all dirty work, heavy lifting, and coordination.

And yet these are precisely the tasks that determine what you look like in the AI's eyes.

One Last Thing

Back to my friend.

Gemini eventually locked in a hotel for her — ten minutes from TYS airport, rated 9.2. She didn't book through the AI. She opened the hotel's own app, $125, tax included, and booked it right there.

Which step was the slowest of the whole process? Not the AI.

It was her husband's "Fine, this one."

Good grief. No amount of GEO strategy can optimize that step.

But that's exactly the card marketers can't afford to lose: AI really is changing how people discover you and evaluate you, but the one who presses the confirm button is always a human being who needs to feel reassured.

Accurate information, honest prices, delivery that keeps its promises. Do these boring things well, and both humans and AI will choose you.