When AI Loses a Business a Customer, Sometimes It’s a $50,000 Customer

More and more people ask ChatGPT and Gemini for recommendations before making a purchase. For a $4 pizza slice, being left out of those recommendations is a minor annoyance. For a business where one customer means $50,000 or more, being left out is a different conversation entirely.

We tracked dozens of queries across ChatGPT and Gemini for Italian kitchen cabinets in Miami. High-end showrooms selling brands like Scavolini, Poliform, Cesar, and Pedini. The kind of purchase where a customer researches carefully before walking through any door.

The patterns we found were similar to what we’ve seen in other categories, but the financial weight behind each recommendation was incomparably higher.

Two Platforms, Two Different Realities

Linea Studio appeared in nearly every Gemini response. Top positions, mentioned no matter how the question was phrased. On Gemini, Linea Studio owns the Italian kitchen cabinet category in Miami.

On ChatGPT, Linea Studio showed up once. In last position. Referred to not even by its own name, but as “Cesar (Linea Studio flagship showroom).”

A potential customer asking Gemini would almost certainly see Linea Studio. A potential customer asking ChatGPT would almost certainly not. Same city, same category, same week.

The opposite story played out for Florida Kitchen. On ChatGPT, it appeared in 62% of responses, often in the top two. On Gemini, it wasn’t mentioned a single time. Zero.

If Florida Kitchen’s owner tested only Gemini, they’d think their business is invisible online. If they tested only ChatGPT, everything looks great. Neither picture is the full story. And without checking both, there’s no way to know which customers are finding the business and which are being sent to competitors.

For anyone curious about how to run a quick manual test, we wrote a step-by-step guide that takes about five minutes.

ChatGPT Can’t Tell the Difference Between a Brand and a Showroom

This one surprised us. And it matters most for businesses that operate as authorized dealers.

Italian kitchen cabinets work on a manufacturer-dealer model. Scavolini, Cesar, Pedini, Veneta Cucine are Italian brands that make the products. Local showrooms in Miami sell and install them. The showroom is the actual business a customer visits. The brand is a name on the cabinets.

Both ChatGPT and Gemini constantly mix these up.

ChatGPT recommended “Arclinea (via Diva Miami)” and “Valcucine (Euro Cucina dealer)” and “Cesar (Linea Studio flagship showroom).” It was trying to be helpful by noting the relationship. But in practice, the credit gets split. A customer reading “Arclinea (via Diva Miami)” might search for “Arclinea Miami” and end up on the Italian manufacturer’s global website instead of contacting Diva Miami’s showroom.

Gemini did something similar. “I Wood Decor (partnered with Cucine Lube)” appeared as one entry. Elsewhere, “Cucine Lube” appeared separately, as if it were a different business in Miami. It isn’t. There’s no Cucine Lube store in Miami. I Wood Decor is the local dealer.

For any business that sells someone else’s brand (authorized dealers, franchisees, local distributors), the AI recommendation meant to send a customer to the showroom might instead send them to the brand’s corporate website, a different dealer in another city, or nowhere useful at all.

This isn’t a problem that shows up in Google Maps. On Google, the showroom has its own listing, its own reviews, its own pin on the map. The brand and the dealer are clearly separate. ChatGPT and Gemini haven’t figured that distinction out yet, and the dealers are the ones losing customers because of it.

The Numbers Behind the Blind Spot

On ChatGPT, 76% of businesses mentioned appeared exactly once. A single mention across dozens of queries isn’t visibility. It’s a coin flip that happened to land favorably one time.

The top three showrooms on each platform captured roughly a quarter to a third of all recommendations. Everyone else split the scraps.

Now put a dollar amount on that.

Say 100 people in Miami ask ChatGPT this month where to buy Italian kitchen cabinets. That number is growing. If three showrooms capture a third of those recommendations, that’s roughly 30 potential customers pointed toward three businesses. At an average project value of $50,000, that’s $1.5 million in potential revenue flowing in one direction.

Not every recommendation converts to a sale. The buying process for a $50,000 kitchen is long and involves multiple touchpoints. But the recommendation is the first touchpoint. Being in it means being part of the consideration. Not being in it means the customer never even learned the showroom exists.

For high-ticket businesses, each lost recommendation isn’t a lost dollar. It’s a lost relationship that could have been worth tens of thousands.

This Applies Way Beyond Kitchen Cabinets

The pattern repeats in every high-ticket category we’ve looked at. A handful of businesses capture most recommendations. The rest are invisible. Different platforms recommend different winners. And almost nobody is tracking it.

Lawyers. Architects. Private schools. Luxury home services. Medical specialists. Any business where one client represents thousands in revenue faces the same dynamic. The recommendations are happening right now, and most of these businesses have no idea whether they’re in them or not.

What a Showroom Owner Can Do About It

A good Google Business Profile, strong reviews, and a polished website still matter. They’re the foundation. But they don’t guarantee that ChatGPT or Gemini will recommend the business, because these platforms pull from a wider and less predictable mix of sources. Review sites beyond Google, editorial mentions, blog posts, “best of” lists, online discussions.

The first step is simply knowing.

“When a potential customer asks ChatGPT who to buy from, does my business show up?”

Most business owners can’t answer that question today. Reachd.ai gives high-ticket businesses - showrooms, dealers, luxury service providers - real-time visibility into how search platforms recommend them versus competitors, whether brand-dealer attribution is accurate, and how their share of recommendations changes over time.

At $50,000 per customer, the cost of not knowing adds up faster than most owners realize.