Shoppers Are Asking AI Where To Go. Most Shopping Center Owners Have No Idea What It’s Telling Them.
By Yvette Elliot, Mall Maverick
Years ago, when I worked in a shopping center, I walked our leadership team around the property to make the case for Wi-Fi. My pitch wasn’t about shopper convenience. Airports had worked it out already. Own the network and you own what happens on it, who’s there, where they go, how long they stay. We owned the concrete. I wanted us to own the air.
They didn’t see it. In fairness, I was early.
I’ve thought about that walk a lot this year. The same question is back with far more money attached. Somebody will own the layer where shoppers decide where to go, and today that’s ChatGPT, Gemini and Google’s AI answers, places where centers may not appear at all. A shopper asks an AI assistant where to buy running shoes this afternoon, or who’s open until nine. It doesn’t hand back 10 links. It names one or two places, and the shopper drives there. If it didn’t name your center, nothing in your reporting will ever show it. No bounce, no abandoned cart, no line item. The visit went somewhere else, and the tenant sales went with it.
That’s the most expensive kind of loss in retail. The one you can’t see. And it isn’t a marketing question. Traffic, tenant sales and asset value make it one for ownership.
Image above and at top courtesy of Mall Maverick
The Money Is Moving Before the Shopper Does
This isn’t a future trend. The numbers are already in.
Shoppers have moved. NielsenIQ reported in May that 42% of U.S. consumers had used an AI tool to shop in the past month.
The click is disappearing. Pew Research Center found that when Google shows an AI summary, people click a traditional search result in 8% of visits, compared with 15% when it doesn’t. Your website now has fewer chances to earn the click once the answer arrives first.
AI referrals are exploding, and converting. Adobe Analytics measured a 693.4% year-over-year increase in traffic from generative AI tools to U.S. retail sites over the 2025 holiday season, and those visitors converted at a 31% higher rate than traffic from other sources, according to Adobe.
Mall Maverick’s centers tell a sharper version of the story. Across a subset of U.S. and Canadian shopping centers whose digital presence Mall Maverick manages, 96% of AI assistant-referred visits in August came from one source: ChatGPT. Mall Maverick’s analytics show that those visitors behave similarly to traditional search visitors once they land, matching them on actions per visit, and AI-driven traffic may be even higher than these figures show because Google’s AI answers report as ordinary search traffic.
That traffic also isn’t arriving evenly across centers. Year over year, AI assistant-referred traffic more than doubled at one center in the analysis while falling by about a fifth at another. This is not a tide lifting every property. It’s a set of decisions being made about specific centers, every month, whether their marketing teams are watching or not.
Agents are next. An ICSC and McKinsey report released in April, Shopping in the Age of AI, estimates agentic commerce could reach up to $1 trillion in U.S. B2C retail revenue by 2030. Separately, McKinsey estimates that the top decile of the retail sector will capture more than 85% of its economic profit.
An AI answer names one or two destinations, not a page of 10 results. Fewer names, more trust. There is no comfortable middle.
Image courtesy of Mall Maverick
Why Shopping Centers Have More To Lose Than Retailers
A retailer that loses the AI answer can still win the sale online. The agent ships the order to the house.
A shopping center can’t. Your business is the trip.
When an agent buys without a visit, the center loses the dwell time, the stop at the second and third store, the food court, the media impression and the parking revenue. Tenant sales are what renewals and percentage rent are built on, and every tenant will eventually ask the same question of their landlord: What are you doing to bring shoppers through the door when the door is now an AI answer?
A good website and a live assistant are table stakes and they earn their keep. They help shoppers at 9 pm and give AI assistants a reliable source of information about your center. Without them, you are not in the running at all. But both are built for shoppers who have already found you. Neither shows you what AI is telling the shoppers who have not.
Run the Numbers for Your Own Center
Four numbers you already have: annual visits, the share of trips that start with a question, the share of those questions where an assistant doesn’t name you and average spend per visit. Multiply.
Take a hypothetical center with 8 million annual visits. Assume 5% of trips start with a question to an assistant and the center isn’t named in half of those answers. That’s 200,000 trips a year going to a competitor, and at an assumed $60 a visit, $12 million in tenant sales exposed.
Use your own figures, then halve them to be safe. Still not a number you’d want to explain to ownership.
A Dashboard Won’t Save You
The instinct is to buy a report. Reports tell you where you’re losing, but they don’t stop it. By the time the monthly PDF lands, the wrong hours have been served to thousands of shoppers, the closed tenant is still listed, and an outdated directory is still feeding the model. Then the report gets cut at budget time, because it never changed anything.
In June I wrote that customer-facing AI can run at full speed while internal teams can’t. A coordinator checking prompts on a Friday afternoon is no match for agents that work the shopper’s side around the clock.
The answer to agentic commerce is agentic intelligence. A system that doesn’t just record what’s true about your property but acts on it: watching what the assistants say, catching what’s wrong, fixing it at the source and proving the result.
Where Mall Maverick Has Been Heading
Your CMS already holds the most accurate truth about your center. Tenant roster, hours, events, maps, promotions. For years its job was to publish that truth and stop. That’s a system of record. The AI era needs a system of action.
Mall Maverick has learned what that means the practical way. At one property where it deployed an AI assistant on the website this summer, more than a quarter of shopper questions arrived while the center was closed, and shoppers kept asking things the site had never answered. That’s a fix list written by shoppers, and it’s the same gap AI search fills with someone else’s answer.
So every release Mall Maverick has shipped this year leans one way. An assistant that answers around the clock and logs what shoppers couldn’t find. AI that keeps website content and maps current without adding headcount. Tenant data structured so AI assistants can use it. What comes next closes the loop: checking what the assistants say about each center, fixing it where the data lives and reporting it so ownership can act.
One outcome sits behind all of it. When a shopper — or a shopper’s agent — asks where to go, your center and your tenants provide the answer.
The Wi-Fi question got answered eventually, mostly by other people, and the centers that waited bought back in on someone else’s terms. This question is still open.
Your competitors are being described by AI right now. So are you. The only difference is who’s checking.
On Monday, ask your marketing team one question: What does AI say about our centers today, and who’s fixing it?
If the answer takes longer than a sentence, book a visibility baseline assessment by emailing info@mallmaverick.com or visiting mallmaverick.com.