Can ChatGPT actually recommend your business?

A growing share of your customers no longer scroll a page of blue links. They ask ChatGPT, Google’s AI Overviews or Perplexity for a recommendation and act on the answer. When that happens, one or two businesses get named and everyone else is invisible. The question worth asking is simple: when a customer asks an AI for a trade like yours, can it actually find and recommend you?

How does an AI decide which business to name?

An AI answer engine does not rank ten links. It reads what it can understand about the businesses in an area and names the ones it can describe with confidence. That confidence comes from a handful of signals: whether it recognises you as a real, consistent entity across the web, whether your site spells out what you do and where in plain machine-readable text, and whether your key facts are backed by structured data it can quote safely.

If those signals are missing, the AI does not take a risk on you. It names a competitor it understands better, even one with a weaker reputation. This is the uncomfortable part for established trades: decades of goodwill and hundreds of reviews do not help if the machine answering your customer cannot read and verify who you are. Reputation is built for humans. AI recommendation is decided by machine readability.

Why are most local sites invisible to AI?

The most common reason is thin, unstructured content. Key facts about services, service area and proof often live only inside images, a PDF, or a few vague marketing sentences. An AI cannot lift those into an answer, so it treats them as if they do not exist. On top of that, most local sites carry no structured data, no clear entity signals tying the business name to its profiles, and short passages that give the model nothing solid to cite.

Answer length is a real factor here. In audits I consistently find that AI engines need roughly 134 to 167 words under a clear heading to quote a passage reliably. Most local FAQ answers run well under that, sometimes a single sentence. The information might technically be on the page, but it is too thin and too unstructured for a model to use, so the business never gets named.

How do I know where I stand with AI search?

You measure it, the same way you would measure a Google rank. An AI Visibility Score checks whether ChatGPT, Google AI Overviews, Perplexity and Bing Copilot can find, understand and cite your business, scored 0 to 100 and broken down per platform. In one audit of a 65-year-old hire business the spread was telling: Google AI Overviews 58, Perplexity 45, Bing Copilot 42 and ChatGPT 38. Strong reputation, weak and uneven machine readability.

Seeing the per-platform breakdown matters because the gaps differ by engine, and so do the fixes. A score alone is not the point; the value is knowing exactly which signals are missing and which platform you are losing on.

What actually fixes AI visibility?

The fixes are concrete, not vague. Expand thin answers to the length a model can cite. Add the structured data that lets an AI quote your facts safely, including FAQ and article schema with author and date signals. Tie your business name to its profiles so the AI recognises one consistent entity. Write your services, area and proof in plain text a machine can read, not buried in an image.

None of this is exotic, but it has to be implemented, not just recommended. If you want to see your own starting point, a free AI Visibility Snapshot returns your score and top three gaps in minutes, and the GEO audit goes deeper across all four platforms with the fixes prioritised.