A growing share of your future customers will never see your website before they decide to trust you.
They’ll ask an AI assistant — “best gothic streetwear brands in India,” “is custom code better than Shopify for a clothing brand,” “who builds D2C storefronts without WordPress” — and they’ll act on the answer it gives. If your brand isn’t in that answer, you don’t exist in that conversation.
This is Generative Engine Optimization — GEO — and it’s where SEO was around 2010: most people aren’t doing it, the ones who are have a quiet, compounding advantage.
Why classic SEO isn’t enough anymore
Google SEO optimises to win a click. GEO optimises to be quoted — to become part of the model’s answer, with or without a click.
The mechanics are different:
- A model can’t run your JavaScript. If your content only appears after a framework hydrates, the model may never see it.
- A model rewards clarity and structure over keyword density. It’s extracting facts, not ranking pages.
- A model trusts content it can parse unambiguously — clean HTML, explicit statements, structured data.
You can rank #1 on Google and still be absent from every AI answer about your category. The two are no longer the same game.
The four things that make you quotable
1. Ship an llms.txt file — and know what it does. This is a plain-text file at the root of your domain stating, in clean prose, who you are, what you offer, and what’s true about you. Be clear-eyed about it: Google has confirmed its search systems ignore the file, and analysis across roughly 300,000 domains found no measurable link between having one and being cited. It earns its place because it costs half an hour and helps in the narrow case where an agent or a person points a model straight at your domain — not because it gets you cited. Every site I build ships one (generate a starter one free →), paired with structured data, which is the half of that pairing that actually does the work. The full evidence is here.
2. State facts as facts. Models extract sentences. “Founded in 2022, we build custom-coded storefronts and have shipped 9 live client sites” is quotable. “We’re passionate about crafting digital experiences” is not — it says nothing a model can repeat. Write the line you’d want quoted back to you.
3. Add JSON-LD structured data. Organization, Product, FAQ, Review schema. This is how you hand a machine the facts in a format it can’t misread — your prices, your offerings, your ratings, your answers to common questions.
4. Make your content server-rendered. If a crawler has to execute JavaScript to read your page, assume some won’t. Server-render or statically generate the important content so it’s there in the raw HTML.
Write the FAQ the model is looking for
AI assistants love a clear question-and-answer pair, because that’s the shape of what users ask them.
Put a real FAQ on your key pages — product pages, your about page, your service pages. Phrase the questions the way a customer would type them into ChatGPT, and answer in one or two complete, self-contained sentences. Then wrap it in FAQ schema.
Done well, your exact answer becomes the model’s answer. That’s the whole game: you’re not hoping to be summarised, you’re handing over the summary.
How to check if it’s working
You don’t need a fancy tool to start:
- Ask ChatGPT, Claude, Gemini and Perplexity a question your ideal customer would ask. Are you mentioned? Is what they say accurate?
- Check your analytics referrers for
chatgpt.com,perplexity.ai, and similar. That traffic is already arriving for brands that did this early. - Fetch your own
llms.txtand read it as if you were a stranger. Does it tell the true, complete story in plain language?
The window is open right now
Most brands in most categories have done none of this. That means the AI answer for your category is currently being written from whatever scraps the models can find — competitor pages, old directories, guesswork.
You can be the source instead. The brands that structure their facts now will be the ones quoted by default for years, the same way early SEO adopters owned Google’s first page long before everyone else noticed.
I’m Divyansh Sood. I build sites with SEO, AEO and GEO wired in from the first line of code — llms.txt, JSON-LD graph, server-rendered content, schema on every page that earns it. 9 live builds since 2022. This is the whole of SEO & GEO services, and it pairs with AI development when you want to build with models rather than just be quoted by them. If your brand should be in the answer when AI tools get asked about your category, email me — I’ll tell you within a day where the gap is.