Google just put AI inside the map: what Ask Maps means for your venue

On August 6, Google expanded Ask Maps to more than 150 countries and territories — Indonesia among them. If you run a café, restaurant or bar, it's worth ten minutes of your attention, because it changes where your next guest's decision gets made: not in a list you can rank in, but inside an answer that names a handful of places.

What is Ask Maps?

Ask Maps is Gemini built into the Google Maps search box. Instead of typing “restaurant near me” and scrolling, people ask the way they'd ask a friend — “somewhere quiet for a work lunch that won't be packed at noon” — then follow up with “does it have parking?” and “what do people order there?” What comes back is a short conversational recommendation, not a ranked list of forty pins.

It launched in the US and India on March 12, 2026, and since August 6 it works across most of the world, in English for now. In the US it already goes past answering: it can book a table, compare hotels, and assemble food orders for pickup through Square and Toast, with Uber Eats to follow. Those point-of-sale integrations pull menus and operating details straight from the restaurant's own dashboard — another surface reading your data automatically.

One more piece of the same update deserves owner attention, currently rolling out in the US: Maps now takes conversational suggested edits to business listings — and you can upload a photo of a storefront sign, which Maps reads to detect new opening hours before asking you to confirm. Your profile is becoming something both people and AI can read and revise in plain language.

How does Ask Maps decide which places to recommend?

The honest answer: Google hasn't said. There's no published methodology, no ranking documentation, and no API to test it against. What Google has stated is what the feature draws on — a database of 300+ million places, insights from more than 500 million contributors, Street View and aerial imagery, and the guest's own Maps history. Reporting on the feature adds that it reads the venue's Business Profile, its Google reviews, and the venue's own website alongside trusted external sources such as local directories and editorial publications.

That combination should sound familiar if you've read our research. Earlier this month we published the first census-denominated study of how AI assistants recommend venues: every café, restaurant and bar in Canggu and Ubud — 4,776 of them — with 2,208 AI answers traced back to that register. Ask Maps didn't exist in our market when we collected the data, and we make no claim to have measured it. But the machinery Google describes is the same kind we measured, and three findings are the best guide currently available.

What does our research suggest about the Ask Maps era?

1. Being findable is the game, not being ranked

In our study, 85.6% of venues were never recommended once — by any system, in any answer. Even among established venues with fifty or more ratings, 72.6% were never named. The wall wasn't position; it was the door: whether the system could find enough written about you to consider you at all. A conversational answer names three places, not forty. For that guest, everyone else does not exist.

2. Reviews are about to work harder — the text, not just the stars

In our data, review volume was among the strongest factors for entering an AI answer (1.6× per step), and fresh reviews leaned the same way (1.4× — just shy of our deliberately strict statistical bar, so we report it as very likely rather than proven). Ask Maps raises the stakes on the text: it answers questions about atmosphere, dishes and details, which means a review saying “great nasi campur, quiet in the morning, easy parking” is a document that can answer a guest's exact question. A bare five-star tap answers nothing.

3. Staleness is the failure mode to watch

Ask anyone what's wrong with AI recommendations and you'll hear “hallucinations.” We went looking and found the opposite problem. Across 12,439 venue mentions, exactly one name — ten mentions, 0.08% — survived scrutiny as likely invented. Meanwhile the systems recommended permanently closed venues 93 times, because old reviews and listicles keep dead places alive. For a living venue the lesson inverts neatly: a steady stream of recent reviews and a current profile are what tell the machine you're open, busy and real this month.

What should a venue actually do this month?

Nothing exotic — the same measured factors, now with a much bigger surface reading them.

1. 01Keep reviews flowing, and aim for detailed ones. Volume opens the door, freshness very likely keeps it open, and specific text answers the guest's actual question. Ask every guest, every service.
2. 02Complete your Business Profile. Hours, price level, attributes, photos. In our study even the humble price field was associated with 1.5× the odds of appearing in answers — and now that strangers can suggest edits conversationally, an incomplete profile is an invitation for someone else to describe your business.
3. 03Have your own website that says plainly what you are. It was the strongest single entry factor in our data (1.9×), and in our citation logs a venue's own site was the most-cited source of all, ahead of TripAdvisor. Google lists venue websites among what Ask Maps reads.
4. 04Don't chase the rating decimal. Across our data the rating did nothing until a venue was already inside the answer. 4.6 with 900 reviews beat 4.9 with 40 — twice over.

The caveat we'd want said about anyone's advice, including ours

Ask Maps is new, unmeasured, and Google hasn't shown its cards. Our findings come from the four AI assistants we could measure, in two bounded markets, before this feature existed there. Everything above is inference from adjacent evidence — good evidence, honestly collected, but not a measurement of Ask Maps itself. Measuring the Maps surface directly is what we're working on next, and we'll publish it the same way: protocol first, data released.

Frequently asked questions