How AI Assistants Pick Local Businesses for You

AI assistants have quietly taken over the comparison shopping that local customers once handled on their own. Traditional local search placed the burden on the user: type a query, scan a list, open several tabs, and make a judgment call. Now, when someone asks an assistant for a business recommendation, the response arrives as a short, pre-selected list.

Google positions AI Mode as especially useful for questions requiring exploration, reasoning, or complex comparisons, the type of inquiries it says previously demanded multiple searches. In standard organic search, a business could check its ranking and understand its position. Google’s generative AI performance report now shows how often links to your site appeared in its AI features, but it doesn’t reveal the query behind each impression or what happens across other assistants.

Getting Listed Is The Entry Fee

Whether an assistant considers a business at all depends heavily on listings, reviews, and reputation. These signals were covered in Reviews, Reputation & Listings: The Local Signals AI Now Reads, and that guidance remains current.

What Google Says Is Behind The Answer

Google maintains that its AI features in Search rely on the same ranking and quality systems as regular results, with two techniques specific to answer generation. Retrieval-augmented generation (RAG), which Google also calls grounding, uses existing ranking systems to pull relevant pages from the Search index. The AI crafts its answer from those pages while linking to supporting sources. Query fan-out runs multiple related searches simultaneously, gathering more results than the original question would produce alone.

For a business’s website content to contribute to Google’s AI responses, it must live on a page Google can retrieve. A request for a quiet restaurant for a client lunch requires that detail to appear somewhere Google can read it, such as in reviews or the business description. When multiple sources conflict about the same business, Google doesn’t document how it resolves discrepancies in a given answer, so consistency across every page you control is essential.

Uberall’s quick-service restaurant benchmark typically produced three to five recommended brands per query. That’s one vendor’s reading of one industry, but it illustrates how narrow the field becomes.

Most local SEO efforts center on making a business well-qualified for recommendation. Accurate listings, an active review profile, and consistency across sources all contribute to this eligibility. Meeting Google’s requirements, however, doesn’t guarantee your content will appear in its AI-driven features.

What Search Console Shows And What It Skips

Google folds AI Overviews and AI Mode data into the main Performance report under the Web search type, mixed with everything else. Those clicks appear in the totals and can’t be separated there. The Generative AI performance report does separate them, though Google is still rolling it out, so not every property has access yet. It counts impressions, meaning the times links to a site appeared in one of Google’s generative AI features, broken down by page, country, device, and date.

There’s no query dimension, so an impression only confirms a link appeared. It doesn’t distinguish between an answer that recommended the business and one that recommended a competitor while listing the business as a supporting source. The report covers Google only. Nothing in Search Console reveals what ChatGPT, Perplexity, or Claude told someone asking which local business to use.

Whitespark’s analysis from July tested 540 queries across three U. S. cities and six industries. It found AI Overviews on 15% of direct local-intent queries, 92% of informational ones, and 97% of hybrid ones, questions like “should I hire a lawyer after an accident” that embed a purchase decision within an information request. Those are the queries where an AI answer can shape which businesses a customer considers.

Until reporting improves, teams are gauging performance two ways. Some run a fixed set of local queries against the major assistants on a schedule and record which businesses get named. That covers only the queries on the days they ran, and assistant answers shift with phrasing, location, and session, so the numbers are directional. Others separate AI referral traffic in analytics where that data exists, tracking volume and behavior over time. That shows what happened after someone clicked through, which is a smaller group than those who saw the answer.

Google’s guidance also warns against third-party tools claiming access to its internal ranking or AI systems.

Each AI visibility method measures a different part of the customer journey, leaving a different blind spot. The Google generative AI report shows link impressions by page, country, device, and date but misses queries, recommendation status, and non-Google assistants. Fixed assistant-query tracking shows which businesses appear for selected prompts but misses changes by phrasing, location, session, and timing. AI referral analytics shows visits and post-click behavior but misses people who saw an answer and never clicked.

Give The Assistant Something To Quote

Google’s optimization guide describes how to position content for inclusion in its AI features. First, a site must be included in Search generative AI features to be eligible. Google is rolling that setting out to some site owners and not others. Include is the default. A child property inherits from the closest parent with a manually configured setting unless the child overrides it. If the setting is available, confirm the site hasn’t been excluded, and businesses running each location as its own property should check those too, since a location property can inherit an exclusion from the domain above it.

Google’s guidelines also recommend keeping crawling permissions open in robots.txt and at the CDN, and making important content available as text. They emphasize structured data that aligns with the visible page content, along with current Business Profile details. Google can process JavaScript content when it isn’t blocked, though its guide says JavaScript-framework sites are generally more complex to work with. Each rendering step is another point where a business’s hours could fail to appear.

AI responses can also pull from pages a business doesn’t control. An outdated directory carrying last year’s hours, an old article, or a competitor’s comparison page describing the business in the competitor’s terms can all surface. The testing covered in July shows what happens when those descriptions aren’t accurate.

Make sure the hours, services, and service area are clearly listed in plain text on the site, matching what’s on the listings. That way, Google receives accurate information directly from the business itself. Google’s optimization guide also clears up misconceptions: Google Search ignores llms.txt and similar AI text files, and there’s no need to break content into small chunks for Google’s AI systems to understand it.

When Agents Start Doing The Booking

Google describes AI agents as systems that carry out tasks for people, like booking a reservation or comparing product options. It says browser agents may access a site directly, analyzing rendered screenshots, inspecting the DOM, and interpreting the accessibility tree that screen readers rely on.

That last point makes it important to check for accessibility gaps. A button built as an unlabeled div, a form field identified only by placeholder text, or a phone number that exists only inside an image are problems for screen readers. Google hasn’t said those defects will stop an agent, but the accessibility tree is one of the layers it says browser agents may interpret. Google points to a web.dev guide for preparing a site for current browser agents, and describes protocols like the Universal Commerce Protocol (UCP) as emerging.

For a local business, being prepared for agents likely means having the same clear information mentioned earlier, plus a smoothly functioning booking or ordering system that works without someone having to fix a broken form.

Where That Leaves Local Businesses

Customers asking an assistant for a local recommendation receive a short list rather than a page of options, and they see it before reaching anything the business controls. Google’s performance report is still rolling out, and tracking tools are nascent. Google’s public guidance doesn’t explain why one local business gets named over another, and Search Console doesn’t expose that decision either. Until that changes, the priorities are keeping business information accurate and consistent across sources, and available as text on pages Google can crawl.

SEJ Live takes this up on August 26. Session 2 covers attributing leads across channels including AI search, and Session 3 covers building local pages AI engines can read, trust, and cite, with Darren Shaw of Whitespark on the panel. Registration is free, and recordings are included.