AI search is here: how to get your business cited by ChatGPT and Gemini
A purchase manager in Dubai types "reliable HDPE bottle manufacturer in India for pharma" into ChatGPT and gets three company names. Either you are one of them or you are not. This is generative engine optimisation, GEO, and it is already deciding who gets the enquiry.
What changed
For twenty years the game was ranking on a results page. Now a growing share of buyers never see that page. They ask an assistant, in plain language, and read a short answer that names two or three options. Google's own AI Overviews sit above the classic results and do the same thing.
The assistants do not invent those names. They pull them from pages they can read, understand and trust. So the question becomes: when an AI reads your website, does it find a clear, confident answer to the buyer's question, or a wall of "welcome to our website" text?
How AI assistants pick who to recommend
From our own testing across client sites, four things matter most.
1. A direct answer, early on the page
Assistants favour pages that state what the business does, for whom, and where, in the first hundred words. "Samarth MHE supplies and services material handling equipment, including battery pallet trucks, stackers and dock levellers, for plants across India" is a sentence an AI can quote. "We are committed to excellence" is not.
2. One page per question
If a buyer might ask about PET bottles for beverages, you need a page about PET bottles for beverages. Not a paragraph inside a products page. Assistants retrieve pages, and a focused page wins over a general one almost every time.
3. Entities, not adjectives
AI models reason about entities: company names, product names, places, certifications, client names, numbers. Pages dense with real entities are easier to trust than pages dense with adjectives. "ISO 9001, 84 products, clients including NTPC and L&T, serving pharma and food packaging since 2015" gives the model something to hold on to.
4. Structured data and consistency
Schema markup (Organization, Service, Product, FAQPage) tells machines what your page is about without guesswork. Consistency across your website, Google Business Profile, LinkedIn and directories tells them you are real. Mismatched names, addresses or phone numbers reduce confidence.
A practical GEO checklist
- Rewrite your homepage opener as a factual sentence: who, what, for whom, where.
- Add a "direct answer" block to every service and product page: a question as the heading and a two-sentence answer under it. We build this into every service page on our own site; scroll any of them and you will see it.
- Add FAQs that match how buyers actually ask: "Do you export to the Middle East?", "What is the minimum order quantity?", "Can you supply with our branding?". Mark them up with FAQPage schema.
- Name your clients and your proof. A clients page with logos, case studies with named companies and outcomes.
- Publish comparison and "best for" content. "HDPE vs PET for pharma bottles" is the kind of page assistants love to cite.
- Keep Google Business Profile complete and matching your website exactly. It is one of the strongest signals for local and "near me" style questions.
- Make the site fast and crawlable. Assistants use crawlers too. A site that blocks bots or loads content only through JavaScript is invisible to them.
- Get mentioned elsewhere. Industry directories, association pages, supplier lists, press. Mentions on third-party sites are how models learn you exist.
How to check if it is working
Ask the assistants yourself. Open ChatGPT, Gemini and Perplexity and type the questions your buyers would type, in the markets you serve. Note who gets named. Do it monthly. We keep a simple sheet for each client: question, assistant, named or not, date. Movement takes six to twelve weeks after the changes go live, similar to classic SEO.
Also watch Google Search Console for impressions on long, question-style queries. Those are the queries feeding AI Overviews.
What not to do
Do not stuff pages with "AI" keywords or write for the model instead of the reader. The models are trained to recognise helpful pages for humans; that is still the target. Do not fake reviews or client names; the assistants cross-check, and so do buyers. And do not abandon classic SEO. The same page structure that gets you cited also ranks.
Where to start this week
Pick your three most important services. Rewrite the opening of each page as a direct answer, add five real FAQs with schema, and name two clients. Then ask ChatGPT the question a buyer would ask and save the answer. Ask again in two months.
Key takeaways
- AI assistants cite pages that answer a buyer's question directly in the first hundred words.
- One focused page per question beats one general page for everything.
- Real entities, names, numbers, certifications and clients, build machine trust; adjectives do not.
- FAQPage, Service and Organization schema plus a consistent Google Business Profile make you easy to verify.
- Test monthly by asking ChatGPT, Gemini and Perplexity the questions your buyers ask.
Questions people ask
Is GEO different from SEO?
It builds on it. The page structure, speed and authority that rank on Google also get you cited by AI assistants. GEO adds an emphasis on direct answers, entities, FAQs and third-party mentions.
How long until an AI assistant starts naming my business?
In our experience six to twelve weeks after the content and schema changes go live, sometimes faster for niche queries with little competition.
Does this work for small businesses?
Often better than for large ones. Niche, specific questions ("GFRP rebar manufacturer in India") have few confident answers online, so a clear page can become the cited source quickly.