Are you a red wine drinker? A high spender? Or perhaps you’re a slow eater, the sort who takes up a restaurant’s table for longer than they’d like. You might not even know — but OpenTable does.
Those are just a few of the notes that the reservation platform has started serving up to some restaurant staff when you make a booking, all based on the orders you’ve made and money you’ve spent at other restaurants in the past.
Kat Menter, a host at a Michelin-starred restaurant who posts about food online as Eating Out Austin, first spotted the new “AI-assisted” tags at work a few weeks ago and shared a look at the system on TikTok. Most flag that a customer frequently orders specific drinks, like red wine or cocktails, but others note customers who spend more than average, frequently leave reviews (“Be nice to them,” Menter jokes), or have a tendency to cancel tables at the last minute. “Mine just says ‘juice,’” she admits. “I love to brunch, that is true.”
If you’re anything like me, all this might have come as a surprise. I just use OpenTable to make reservations, after all — so how does it know what I’ve ordered?
The truth is, OpenTable — like Resy and other rivals — has always done more than just help you find a table. The platform is billed to restaurants as a one-stop shop to handle reservations, waitlists, reviews, marketing, and more, but it also offers its own table management software, along with integrations into the most popular point of sale (POS) systems in the industry, such as Toast or Epos. These are the tools that run much of the day-to-day in restaurants themselves, including inventory, orders, and payments.
That’s how OpenTable knows that you usually order a couple glasses of white wine with dinner. You don’t even have to make the booking through OpenTable — so long as you have an OpenTable account, and give the restaurant your phone number or email, your booking might be paired to your profile regardless. OpenTable will then know when you arrived, what you ordered, how much you spent, what time you paid, and more besides. The data finds a way.
Still, the company might also know less about you than you think. I used its privacy rights request form to pull a copy of all the data it has on me, and it was reassuringly dull: some basic contact details, a list of the reservations I’d made through the platform, and some limited credit card information. One reservation, from 2012, had the note that I was a “first time diner,” and that’s about it.
But let’s say OpenTable knows more about you than it does me. What would a restaurant want with that information? It’s essentially a shorter, simpler version of the sort of research and notes that some restaurants — especially in fine dining — handle anyway. Certain Michelin-starred restaurants spend hours each week digging into guests’ social media profiles to predict their preferences, and San Francisco’s Lazy Bear maintains a database of 115,000 past guests in case they ever come back. Menter tells me that the Austin restaurant where she works tracks some of these details too. There are practical notes, like which customers always arrive late and who has a tendency to be a no-show, but also more personal touches — there’s the guy who always brings first dates, so there’s a note for staff to act like they’ve not seen him before, or the couple who are veterans and would both prefer to be seated with their backs to a wall and a view of the exit.
“We keep note of your kid’s name, how many visits you’ve had with us, if there are any dishes you absolutely love, things like that,” Menter says. “It’s all meant to surprise and delight each reservation.”
She’s less confident that OpenTable’s AI-assisted notes can do the same job. “We’ve been taking them with a grain of salt,” she says, adding that “many of them just seem random.” The automated notes are obviously simpler than the restaurant’s own, but worse, they’ll lump together an account holder’s data with everyone else they’ve ever dined with. Someone might pick up the “high spender” tag for handling a business dinner on a company card, or a teetotaler might be flagged as a cocktail lover if they’re often out with friends who drink. There’s a privacy problem here, but plenty of practical ones too. “I revert to not trusting the tags,” Menter notes.
OpenTable wouldn’t say how long it’s been collecting POS data, nor when it started sharing it with restaurants. Senior director of communications Mary-Kate Smitherman tells The Verge that the AI-assisted tags are a beta feature, currently only available to restaurants on its OpenTable Pro plan. She didn’t tell me what AI model the company employs, but says that it isn’t used to process individual guest data. Instead, the AI element is in analyzing restaurant item descriptions like “glass of cabernet” to categorize them as “red wine,” making it possible to classify and aggregate large, messy datasets of customer orders.
“We’ve been taking them with a grain of salt.”
The tech “both benefits the business and offers a special experience for the diner,” Smitherman says, and was introduced “following feedback and requests” from restaurants. “They might help a server suggest a dish you’ll love or recognize that you prefer a more relaxed dining pace,” for example. She confirms that OpenTable shares the information “across our network,” before defending its right to do so. “What we share with restaurants is guided by the choices you’ve made in your privacy preferences,” she says, pointing me toward the platform’s privacy policy.
The privacy policy is actually a little opaque on this. Of course it notes that data will be shared with a restaurant when you book, but only lists the details you might expect: your name, contact details, party size, and special requests, along with a vague “dining preferences” catch-all. There’s a note that it may also share “additional information about your dining activity at that restaurant or restaurant group in the past,” but no indication that information from visits to unrelated restaurants would be included.
A separate section admits that OpenTable collects POS data from participating restaurants (“such as items ordered, bill total, and time spent at the restaurant”), but only says this will be used “to provide aggregate information to the restaurant about their customers.” While the policy itself defines aggregated information as “general statistics that cannot be linked to you or any other specific user,” Smitherman tells me that it also refers to “aggregated insights about individual customers” — like the fact that I, on aggregate, drink a lot of red wine.
As Smitherman suggests, users can opt out. You can do so by logging into your account, heading to your profile, and then going to the “Preferences” page. You’ll find six options related to the privacy policy, but the one that matters most is the last one: “Allow OpenTable to use Point of Sale information.” Untick that, and your order history should be your own again.
For now, this appears to be unique to OpenTable. Its chief rival, Resy, collects “transaction data” and “metadata about your dining habits and experiences,” according to its privacy policy, which can be shared with “restaurants and their affiliates.” But Resy communications director Lauren Young tells me that “point of sale data or guestbook information” aren’t shared with “unaffiliated restaurants.” Restaurants with the same owners might be able to share information between them, but unlike with OpenTable, details drawn from your dining history with them won’t be handed to other restaurants under different ownership.
Take this as a good reminder that OpenTable was never only about reservations, and that you were always part of the product, whether you knew it or not.
But you probably don’t need to worry that a restaurant is going to treat you too differently because it knows what wine you like or how long you tend to eat for. At least not yet, while these tools are so rudimentary that staff are as likely to ignore them as take them into account.
“These tags are similar to anonymous tips, from an unreliable narrator,” Menter says. “You were probably going to get good (or bad) service anyway.”
