Getting cited by AI assistants is not one problem. Perplexity, Gemini and Copilot assemble answers from different sources, which means the same page can be quoted constantly in one and never appear in another.
Here is what actually influences each, what is in your control, and what is not worth your time.
Where each one gets its sources
Perplexity runs a live search for most queries and cites the pages it retrieves, prominently and with links. This makes it the most directly winnable of the three: if you rank well for the underlying query and the page contains a clean, liftable answer, you have a real chance of being cited.
Gemini draws on Google's index and increasingly surfaces in AI Overviews within search itself. What influences it looks a lot like classic SEO — relevance, authority, and content structured so a passage can stand alone.
Copilot leans on Bing. This matters more than people expect, because many businesses have never looked at Bing Webmaster Tools and have no idea whether their site is indexed properly there. It is often the cheapest win available.
ChatGPT sits in between: it browses for current questions and otherwise answers from training, where your presence depends on being written about widely rather than on anything you publish today.
Being cited by AI assistants: what you control
A liftable answer. Forty to sixty words, immediately under a heading phrased as the question, complete on its own. If the paragraph depends on the one above it, it cannot be quoted. This single formatting habit does more than anything else on this list.
Specifics that can be checked. Numbers, dates, named standards, thresholds. Assistants are built to prefer sources that commit to something, and a page of balanced adjectives commits to nothing.
Being indexed where the assistant looks. Submit your sitemap to Bing Webmaster Tools as well as Search Console. It takes ten minutes and it is the one step most Indian businesses have skipped.
Entity consistency. The same business name, address and description everywhere you appear — your site, your schema, your profiles, directories. Assistants are resolving which entity you are before deciding whether to mention you, and contradictory information makes that resolution fail.
Being mentioned elsewhere. This is the uncomfortable one, because it is the least controllable and probably the most influential. Models and retrieval systems both weight what other sources say about you. Mentions without links count here in a way they never did for rankings.
What is not worth your time
llms.txt as a ranking tactic. It is a reasonable courtesy and costs nothing to publish, but no assistant has committed to using it as a source of truth. Treat it as documentation, not optimisation.
Keyword stuffing for models. The retrieval layer is a search engine. Everything that made keyword stuffing counterproductive in search makes it counterproductive here.
Trying to get listed in "best of" articles by paying for it. Assistants do lean heavily on listicles, which is why this gets sold. Paid placements in low-quality roundups tend to appear in exactly the sources assistants have learned to discount.
A realistic order of work
- Verify the site in Bing Webmaster Tools and submit the sitemap.
- Rewrite the top twenty pages so each opens with a direct answer under a question heading.
- Make the organisation schema and public profiles agree with each other.
- Publish one thing a quarter worth citing — original data, a tool, a genuinely thorough reference page.
- Track it, so the argument is about numbers rather than opinions. The setup is in tracking AI assistant traffic in GA4.
The broader version of this, including what changes for Indian businesses specifically, is in AI search optimisation in India.
Questions people ask before hiring
How do I get cited by Perplexity?
Perplexity runs a live search for most queries, so ranking well for the underlying question is the prerequisite. Beyond that, the page needs a self-contained answer of roughly forty to sixty words directly under a heading phrased as the question.
Does Copilot use different sources from Google?
Yes. Copilot leans on Bing, so being indexed and performing well there matters independently. Verifying the site in Bing Webmaster Tools and submitting the sitemap takes ten minutes and is the step most businesses have skipped.
Does llms.txt help you get cited?
There is no evidence that any assistant treats it as a ranking input. It is cheap to publish and reasonable as documentation of your own site, but it should not be mistaken for an optimisation tactic.
What matters most for being mentioned by assistants?
Being discussed elsewhere. Retrieval systems and models both weight what other sources say about you, which makes unlinked mentions, roundups and coverage more influential here than they ever were for ordinary rankings.