Does ChatGPT recommend you — or your competitor?

Search used to end on a results page with ten blue links; you could rank fifth and still be seen. An assistant answer has room for two or three names. If yours is not one of them, you are not on page two — you are nowhere. That is the blunt arithmetic of AI search, and it is why "does ChatGPT recommend us?" has become a board-level question.

How assistants decide whom to name

An assistant composing a recommendation is not ranking a thousand results; it assembles an answer from the handful of sources it can retrieve, parse and trust within a few seconds. Three things decide whether you are in that handful: whether your pages can be found at all, whether they answer the question plainly enough to quote, and whether structured data and consistent facts make you safe to cite. Brand size helps less than most people expect — a small firm with a clear comparison page regularly beats a large one whose answer is buried in a PDF.

None of that is visible from the outside, which is the uncomfortable part: your classic rankings tell you nothing about it. A page can sit at position three in Google and still never appear in an assistant's answer.

Measure before you optimize

The starting point is not more content. It is knowing what the assistants answer today when someone asks the questions your customers ask: "best [your service] in [your city]", "alternative to [market leader]", "is [your brand] trustworthy". The AI visibility check in AI Rank Suite asks exactly those questions — up to 25 prompts per run, generated from your industry, your services and your location, and editable down to the letter. For every prompt it records whether your brand was named, who was named instead, and which of your pages the answer leaned on.

The report, in colors

Green rows are answers in which you appear. Red rows name the competitors that took your place — and that list is your real competitive set in AI search, which is often different from the one you assumed. The agency across town you have never heard of may own every "best in [city]" answer, while the rival you watch obsessively appears nowhere.

The three pages that decide most answers

Across industries, assistant recommendations lean on the same trio: a pricing page that states real numbers in text, a comparison page that positions you honestly against alternatives, and proof — cases, reviews, credentials — that makes recommending you safe. Most websites have none of the three, because classic SEO never forced anyone to write them. In AI search they are the difference between being the answer and being a footnote.

Then fix what is actually missing

Most sites lose not because their product is worse but because the page an assistant needs does not exist: no clear comparison page, no plain-language pricing, no page that answers the exact question being asked. Each red row in the report links to its gap, and each gap has a "draft it" action: the built-in AI writer produces a first draft — on your own API key, in your language, with your site's tone settings, as a WordPress draft that is never auto-published. You edit, you publish, you re-run the simulation, and you watch the row turn green.

What the loop costs

Because the simulation runs on your own key, a full 25-question run typically costs well under a dollar in provider fees, and the usage ledger records the exact number. Compare that with what the market charges for the same telemetry: Yoast's AI+ tier costs $358.80 per year and includes five questions per week. Monthly runs are plenty for most sites; during a campaign, run weekly. The point is not the frequency — it is that measurement precedes optimization, every time. See the pricing page for what is included in Pro.

FAQ

Which assistants can be simulated?

The check runs against the AI providers you connect — OpenAI, Anthropic or Google — and asks its questions the way a customer would. It measures the answering behavior of the models behind ChatGPT, Claude and Gemini rather than scraping any single interface.

How many questions should I test?

Twenty-five well-chosen prompts cover most businesses: your services times your locations, plus alternative questions and trust questions. Edit the generated list — you know phrasings your customers use that no generator guesses.

How fast do results change?

Faster than in classic SEO. When a missing page goes live and gets indexed, answers can shift within weeks, because assistants retrieve fresh sources instead of relying on rankings alone. Still: treat it as a loop, not an event.

Is this Generative Engine Optimization?

Yes — measuring assistant answers and closing the content gaps they reveal is the working core of Generative Engine Optimization (GEO). The classic SEO layer stays necessary; GEO adds the layer where the answer, not the ranking, is the product.