AI is answering your customers' questions. Is it mentioning you?

Try a controlled spot-check: ask several search-enabled AI tools the same realistic local questions and record whether your business is named, how it is described, and which sources support the answer. Do not treat one response as a ranking report. Treat the exercise as a snapshot of what a customer might encounter—and repeat it consistently enough to notice patterns.

AI-assisted discovery is now material, although it should not be confused with the share of all local searches. In BrightLocal's 2026 US consumer survey, 45% of respondents said they had used generative AI tools for local business recommendations, up from 6% in the previous year's survey.

Traditional search visibility still matters. Google's AI Overviews and AI Mode are features of Google Search, and Google says the same foundational SEO practices apply to them. Other assistants can search, combine sources, and phrase answers differently, so a strong conventional ranking does not guarantee that every assistant will name the business for every prompt—but neither are the two worlds cleanly separate.

Sources: BrightLocal — Local Consumer Review Survey 2026 and Google Search Central — AI features and your website

A conventional results page presents ranked links. A search-enabled assistant may issue several related searches, combine information from multiple sources, and generate a response with a smaller set of named examples or supporting links. Google describes this as "query fan-out" in AI Mode and AI Overviews; ChatGPT Search also rewrites some prompts into targeted searches and may use multiple search providers.

That compresses attention, but "there is no page two" is too absolute. AI products often provide citations or expandable sources, and the businesses named can change with location, wording, current web results, product mode, personalization, and model updates.

Sources: Google AI Mode Help and OpenAI — ChatGPT Search

Write down five or six questions a real customer asks before choosing a business like yours. Include the service and location explicitly. Then:

• Use the same prompt wording in each product.
• Use a fresh conversation where possible so prior chat context does not shape the answer.
• Record the product, mode, date, prompt, names returned, description, and cited sources.
• Repeat each important prompt more than once. Generated answers vary, so one run is weak evidence.
• Separate "mentioned" from "cited." A business can be named without its own website being used as a source.

Repeat the same small test set monthly or quarterly. The point is not to manufacture a universal score; it is to detect recurring omissions, factual errors, and competitor patterns.

No. Manual testing is useful for answer quality, brand accuracy, and competitor context, but it is not the only available measurement.

Google introduced dedicated generative-AI performance reports in Search Console in June 2026, covering visibility in features such as AI Overviews and AI Mode. ChatGPT also appends utm_source=chatgpt.com to referral URLs, allowing visits to be identified in analytics when someone clicks through. These measures answer different questions: Search Console reports Google visibility, analytics records attributable visits, and manual testing shows what the generated answer actually says.

Sources: Google Search Central — generative-AI performance reports and OpenAI — Publishers and Developers FAQ

That is a finding, not a diagnosis. Check the cited sources and the businesses that were named. Then look for testable gaps: inaccurate or incomplete public business information, weak coverage of the specific question, limited independent evidence, a crawl or indexing problem, or simply a prompt for which your business was not the best match.

Do not assume the competitor's review count, content format, or directory presence caused the answer merely because it is visible. The assistant does not expose a complete causal ranking formula. We cover the practical foundations in how to get your business found in AI search.

A controlled manual test is a useful part of an AI-visibility audit, not a definitive rank tracker. Pair repeated prompts with Search Console and referral analytics, then investigate patterns rather than reacting to one answer.

Once you know what is recurring, the fixes become concrete work—the kind we build into Marketing & Growth. Start a conversation if you want a second pair of eyes on what you find.