Search Is Dying. Asking Is Next.

Google still processes billions of queries a day. But something shifted underneath it, and most restaurant operators haven't noticed yet.

37% of consumers now start their searches with AI tools instead of a search engine. Not alongside Google. Instead of Google. Gartner projected a 25% drop in traditional search volume by 2026, and the numbers are tracking. When someone wants dinner tonight, a growing share of them are opening ChatGPT or Perplexity and typing a full sentence: "Where can I get a good pasta dinner for two in the Annex on Saturday around 7?"

That's not a search. That's a question. And the difference matters more than most operators realize.

Searching was a skill. Asking is a conversation.

Think about how Google trained us to behave. We learned to compress our intent into three or four words. "Best Italian restaurant Toronto." We'd scan ten blue links, open three tabs, cross-reference reviews, check the menu on one site and the hours on another.

Now picture the same person asking ChatGPT: "My partner and I want somewhere cozy for pasta, not too loud, ideally with a good wine list, somewhere in midtown. What do you think?"

Twenty words instead of four. Context instead of keywords. The query itself tells the AI about the occasion, the vibe, the neighbourhood, the priorities. And the AI responds with three to five specific recommendations, not ten links to sort through.

This is the shift: from filtering results to receiving answers. From searching to asking. From ten options you evaluate to three options someone (or something) already evaluated for you.

The wrong restaurants are getting recommended

Here's where it gets uncomfortable. Uberall's May 2026 report found that 83% of restaurant locations are invisible in AI-generated recommendations. Not poorly ranked. Invisible. They don't exist.

And it gets worse for independents. Despite representing over 60% of restaurant locations in North America, independent restaurants appear in fewer than 3% of AI dining recommendations. The top three brands per category capture 53.4% of all AI recommendation share.

The math is brutal. AI recommends three to five options per query. If you're not in that handful, you're not in the conversation at all. There's no page two to scroll to. There's no "more results" button. You either made the cut or you didn't.

Why AI recommends what it recommends

This is the part most operators skip over, and it's the part that matters most.

AI doesn't rank the way Google ranks. It doesn't count backlinks or domain authority. It pulls from a different set of signals: your Google Business Profile, review platforms, third-party mentions in food blogs and guides, your website content (if it's text-based and crawlable), and structured data.

Two findings from Local Falcon's research stand out. First, AI-recommended restaurants average 3.6 times more Google reviews than comparable non-recommended restaurants. Not better reviews. More reviews. Volume is the signal, not score. Second, star ratings above 4.4 had almost no impact on recommendation likelihood. Getting from 4.4 to 4.8 doesn't move the needle. Getting from 200 reviews to 600 does.

Each AI platform also weights sources differently. Birdeye's analysis showed that ChatGPT leans on Yelp and directories. Gemini favours first-party websites. Perplexity splits between sources. There's no single playbook, but there is a common thread: the restaurants AI recommends are the ones it can understand clearly. Complete profiles. Consistent information across platforms. Text-based menus. Recent reviews with specific detail.

The zero-click reality is already here

Here's the context that makes this urgent. Searches that trigger Google's AI Overviews produce zero-click rates of 83%. In Google's AI Mode specifically, 93% of searches end without a single click to an external website. The person gets their answer and acts on it without ever visiting your site, your Instagram, or your booking page.

For restaurants, this means being the answer is the entire game now. If AI summarizes your competitor when someone asks "best brunch near Kensington Market," that person books there without browsing. They never see your menu, your photos, your vibe. They never had the chance to fall in love with your place.

The link between "search" and "visit" is collapsing. What's replacing it is the link between "ask" and "act." Google's own data shows searches for "when to book a table" surged 140% this year in the UK, and their AI Mode now handles restaurant bookings directly in eight countries, including Canada. No visit to a booking platform. No visit to your website. Ask, choose, book.

Why this is actually good news for independents

This is the non-consensus part, so stay with me.

The old game was stacked against independents. SEO rewarded backlinks, domain authority, and content volume. Chains had teams and budgets for that. An independent operator couldn't compete with a chain that published 200 blog posts and had links from every food directory in Canada.

The new game is different. AI doesn't care about your backlink profile. It cares about specificity, clarity, and being talked about in the right places. Your Google Business Profile is free. Your reviews are free. Your menu being text on your website instead of a PDF is free. A food blogger mentioning you in a neighbourhood guide costs nothing.

Here's the shift that matters: in the old search world, scale won. In the ask world, specificity wins. When someone asks "where can I get handmade pasta in a quiet spot near the Danforth," the AI isn't looking for the biggest brand. It's looking for the most specific match. A 30-seat pasta bar with a strong profile and 300 detailed reviews beats a chain with 2,000 generic ones for that query.

Independents are specific by nature. You ARE the neighbourhood spot. You ARE the place with the story. The problem isn't that you don't match what AI wants to recommend. The problem is that AI can't find the information to recommend you.

What changes and what doesn't

The fundamentals haven't changed. Cook great food, give a great experience, and people talk about you. What's changed is where that talk needs to live for it to count.

Word of mouth used to spread over tables and text messages. It still does. But now it also needs to spread in places AI can read: Google reviews, blog mentions, your own website with actual text on it. The dishwasher who tells his friends about your lamb shank is still valuable. But the guest who writes "the lamb shank at [your restaurant] on Ossington is the best in the city" in a Google review is doing double duty: telling their friends AND telling every AI assistant at the same time.

Three things need to be true for AI to recommend you: it needs to know you exist (complete Google Business Profile), it needs to know what you're about (text-based menus, descriptive website, consistent information), and it needs social proof that you're worth recommending (review volume and quality).

If any of those three are missing, you're part of the 83%.

The window

Google AI Mode is live in Canada. ChatGPT and Perplexity are already the go-to for a growing share of diners. But most restaurants, and nearly all independents, haven't adapted yet. The Uberall data is from last week. This is still early.

That means the independents who get this right now will be the ones AI learns to recommend first. And unlike SEO, where catching up meant months of content and link building, the basics here can be done in an afternoon: update your profile, make your menu crawlable, ask for reviews.

The shift from search to ask is real. The question isn't whether it's happening. It's whether your restaurant is part of the answer.

---

Trudy's Table is building reservation software for Canadian independents. If you want to be part of the answer when someone asks "where should I eat tonight?", join the waitlist.

---

Sources: Uberall QSR Benchmark 2026, Search Engine Land, Local Falcon, Birdeye, Gartner, Google UK Blog.

---