I Asked ChatGPT to Recommend a Business Like Yours. Heres What It Said.

Here's a test worth running right now.

Open ChatGPT or Perplexity. Type: "What's the best [your service] for [your target customer]?"

Then watch.

Chances are, you won't see your business. You might see your biggest competitor. You might see a brand you've never heard of. You might see a listicle from 2022 that no longer reflects reality. What you almost certainly won't see is the business that has spent years building a great product and trying to rank on Google.

This is the search problem nobody warned you about.

---

The Customer Who Never Visited Google

For most of the last 20 years, the path was predictable: customer has a problem → customer types into Google → customer clicks a result → customer lands on your site.

That path hasn't disappeared. But a second path now runs parallel to it.

Millions of people now open ChatGPT, Perplexity, Claude, or Gemini and ask a full conversational question. They don't search for keywords. They describe their situation and ask for a recommendation. Something like:

• "I'm launching a new SaaS product, what tools do I need to get found on Google?"
• "What's a good SEO tool that also helps with AI search?"
• "Can you recommend an affordable alternative to Semrush for a small business?"

The AI responds with a confident, detailed answer. It names specific products, explains tradeoffs, and often includes links. The customer trusts it and acts on it.

If your business doesn't appear in that response, you don't get a second chance. The customer never visits your website. They never see your offer. They don't know you exist.

---

Why This Is Different From Traditional SEO

It's tempting to think that if you rank well on Google, you'll naturally show up in AI responses too. Sometimes that's true — but often it isn't.

AI tools don't just pull the top Google result. They synthesize from multiple sources, weight content based on structure and clarity, and rely heavily on how well your content answers direct questions. A page that ranks #1 for a keyword might completely miss the format that AI models prefer.

There's also a category shift happening. Traditional SEO optimizes for keywords. The new discipline — often called GEO, for Generative Engine Optimization — optimizes for citations. It's about being the source an AI confidently names when a user asks a question in your space.

If you're not sure what the difference looks like in practice, the GEO vs SEO comparison breaks down exactly where the two approaches diverge: different signals, different content formats, different technical requirements.

---

The Vocabulary Has Changed Too

One reason businesses fall behind on this is that the terminology shifts faster than the tactics. Six months ago, "AI SEO" wasn't a widely used phrase. Now there's a whole ecosystem of terms — GEO, AEO, LLM SEO, AI citations, RAG, llms.txt, PerplexityBot — each describing something real and distinct.

Understanding what each term actually means matters, because the tactics attached to them are different. Optimizing for AI Overviews on Google requires different actions than optimizing to appear in Perplexity's source citations, which requires different actions than getting ChatGPT to name-check your brand.

The GEO and AI SEO glossary has clean definitions for all of these — worth bookmarking as a reference as this space keeps evolving.

---

What Actually Gets You Cited

AI tools favor content that is specific, structured, and directly answers questions. The signals that matter most:

Clear entity identification. AI models need to understand exactly what your business is, what it does, who it serves, and what problem it solves. Vague positioning — "we help companies grow" — doesn't give an AI enough to work with.

Question-and-answer formatting. Content structured around questions that your customers actually ask performs significantly better than long-form content that buries the answer in paragraph six.

Structured data. FAQ schema, HowTo schema, and other JSON-LD markup give AI crawlers a pre-formatted data layer to extract from. Pages without structured data are harder for AI models to parse accurately.

llms.txt. A plain-text file at the root of your website that describes your business directly to AI models. Think of it as a business card written specifically for ChatGPT, Perplexity, and Claude to read. Without it, AI models have to infer who you are from your content — and they sometimes get it wrong.

Allowing AI crawlers. GPTBot, PerplexityBot, ClaudeBot, and Anthropic-AI all need explicit permission in your robots.txt to access your content. Many sites still block them accidentally.

None of this is especially complex. But it requires intentional action — it doesn't happen by default.

---

How to Actually Fix It

The gap between "technically SEO'd" and "visible in AI search" is bridgeable, but it requires a systematic audit of both layers simultaneously. Most traditional SEO audits weren't built to surface AI-specific gaps.

The how it works page explains the approach we use: a 100-action roadmap that covers both Google optimization and AI search optimization in a single audit. The two aren't separate workstreams — they share a foundation of technical health, content structure, and authority signals. The distinction is in the final 20–30% of tactics that are specific to each channel.

For most sites, fixing the AI visibility layer takes less time than the initial Google SEO work. The fundamentals are already there. You're adding a translation layer — repackaging what you know into formats AI models can easily read and cite.

---

The Window Is Still Open

One unusual feature of this moment is that AI search is still early. Most businesses haven't touched it. The competitive landscape in AI citations isn't nearly as settled as it is in Google rankings, where established domains have years of backlinks and authority baked in.

The businesses that move now will be the ones ChatGPT recommends six months from now, when your competitors finally realize they should have started sooner.

Run the test again: ask an AI tool about your category, your service, your customer's problem. See who it recommends. If it's not you, that's fixable — and the window to fix it cheaply is still open.