How do I get my website recommended by AI?

More and more buyers skip Google entirely and ask an AI chatbot: "What's the best [product] for my situation?" ChatGPT, Perplexity, Gemini, and Google AI answer with a short list of named companies. If you're not on that list, you don't get the click, the lead, or the sale, and you never even knew the conversation happened. This is how you get onto that list.

First, understand how AI picks who to recommend

AI assistants don't use Google's link-based ranking. They generate an answer from content they can read, understand, and trust, pulled from their training data and, increasingly, from live web search. The pages they cite tend to be machine-readable, clearly structured, factual, and fresh. That means the game isn't "rank higher". It's "be the clearest, most trustworthy source on your topic."

What we found when we tested this

We tested this across 50 buying questions put to ChatGPT, Claude, Gemini and Perplexity, ten each in five industries, and recorded every company and source named. That is 200 answers and 951 distinct websites cited between them. The methodology, including the exact prompts, the run dates and a run-to-run variance control, is published with the results rather than summarised here.

The four assistants agreed on a full set of names in 3 of those 50 questions. In the other 47, asking a different assistant produced a different shortlist. There is no single ranking to climb here, which is why the work below is about being legible to each of them rather than about beating one scoreboard.

The second finding matters more for where you spend effort. A large share of what the assistants read was not a business's own website at all:

Ten questions per industry, four assistants, run July 2026. Treat the directory figures loosely: they move depending on where you draw the line between a directory and a publisher. The pattern is the point. Read the full study, including every question, every named business and the run-to-run variance control.

In practice that means step 6 below is not filler. If a fifth to a third of what an assistant reads about your category is a directory, being absent from the ones it reads costs you more than another page on your own site would earn.

The 7 things that get you recommended

1. Add structured data (JSON-LD schema)

Schema markup tells AI what your business is, what you sell, and how you're rated, in a machine-readable format it can trust. At minimum, add Organization, Product or Service, and FAQ schema. It is the change that returns the most for the least effort on most sites, and it is usually a copy-paste job.

JSON-LD is the format Google's own structured data documentation recommends, and the same markup is what an assistant reads to decide what your page is about. If you want the exact blocks to paste, we keep a worked example and the mistakes that break each type in our schema reference.

2. Write content that directly answers questions

AI rewards content formatted the way people ask questions. Use real question phrasing in your headings ("How much does X cost?", "Is X better than Y?") and answer in the first sentence, before you elaborate. This "answer-first" structure is far more likely to be lifted into an AI response than a long, meandering paragraph.

3. Keep your content fresh

AI systems favor current information. Add and maintain datePublished and dateModified metadata, and genuinely update key pages. A page that clearly signals it was updated this month beats an undated page from three years ago.

4. Use clean, semantic HTML

AI (and the agents that browse on users' behalf) parse your page structure. Use proper nav, main, article, and heading tags, descriptive alt text, and labeled form fields. If a machine can't tell your headline from your footer, it won't confidently cite you.

5. Make sure your content is actually crawlable

If your key content only appears after JavaScript runs, many AI crawlers won't see it. Serve real HTML content, keep a clean sitemap, don't block AI crawlers in robots.txt, and keep load times fast. You can't be recommended for content the AI never read.

The robots.txt line is the one people get wrong, because the crawler that matters is not the one they have heard of. OpenAI runs separate agents, and its bot documentation is explicit that a site disallowing OAI-SearchBot "will not be shown in ChatGPT search answers", while blocking GPTBot alone only opts you out of training. On speed, Google's guidance is that sites should aim for a Largest Contentful Paint of 2.5 seconds or less. Treat that as the floor you clear, not a score you optimise.

You can check which AI crawlers your own robots.txt currently allows with our free crawler checker. It reads the file and tells you which of sixteen agents can reach you.

6. Build authority AI can verify

AI cross-checks claims. Being mentioned on review sites, industry directories, Reddit threads, and reputable third-party pages builds the "this company is real and well-regarded" signal that makes AI comfortable naming you. Encourage reviews and get listed where your category is discussed.

7. Cover the questions buyers actually ask

Create clear, dedicated content for the exact comparisons and questions in your niche ("best X for small business", "X vs Y", "how to choose an X"). These are the prompts people type into AI: and pages that answer them cleanly are the ones that get pulled into the answer.

A quick checklist

• Organization + Product/Service + FAQ JSON-LD schema in place
• Answer-first content with real question headings
• Current datePublished / dateModified on key pages
• Semantic HTML, descriptive headings, alt text, labeled forms
• Content served as HTML, crawlable, fast (< 2.5s), not blocked in robots.txt
• Reviews and third-party mentions that verify your authority
• Dedicated pages for the buyer questions in your category

Questions people ask us

How do I get my website recommended by AI assistants?

Give assistants something they can quote and verify: JSON-LD structured data, pages that answer one buyer question directly in the first paragraph, visible publish and updated dates, and mentions on third-party sources they trust. Then measure it, ask ChatGPT, Perplexity, and Gemini the buying questions in your category and see whether you're named. If you aren't, the sections above tell you what to change first.

Why doesn't AI recommend my website even though it ranks on Google?

Google ranks pages by links and relevance. Most AI assistants compose answers from pages they can parse, quote, and verify, a #1-ranked page can still fail that test if it lacks schema markup, direct-answer formatting, and freshness signals. The two systems reward different things, which is why the gap between them is worth measuring.

How long does it take to get recommended after optimizing?

Faster than classic SEO. Assistants with live web search (ChatGPT with browsing, Perplexity, Gemini) reflect on-page changes as soon as they re-crawl, often days, not months. Building enough third-party presence to be named consistently takes longer, but structural fixes like schema and answer-first formatting show up quickly.

Don't guess, measure your AI visibility

The hardest part is knowing where you actually stand. You can't improve what you can't see. That is what Greater Than Services does: it puts the buyer questions in your category to ChatGPT, Claude, Gemini, Perplexity and Google AI Overviews every week, shows whether you were named and what was cited instead, and generates the code to close the gap.

The figures in this article come from our own testing. The study sets out the questions, the assistants, the dates and the variance control, and the raw dataset is published under CC BY so you can check it. If something here does not match what you see, tell us and we will correct it.