How to get your website cited by ChatGPT, Perplexity and Google AI Mode
What decides whether AI assistants mention and link your site in 2026, and six steps that work for small businesses and freelancers: crawler access, Bing, quotable pages, one clear identity and mentions.
Vishal Solanki · Full-stack developer, Noida
More people now ask an assistant "who's a good web developer in Noida?" or "how much does a website cost?" instead of scrolling ten blue links. When the answer names three businesses, being one of them is the new page one.
This is often called GEO (generative engine optimisation) or AEO (answer engine optimisation). Most of it is ordinary good SEO, plus a few things that now matter more than they used to.
An honest caveat: none of the AI companies publish exactly how they choose sources. What follows is what their crawler documentation says, plus what keeps showing up in 2026 industry studies. Treat precise percentages you see online, including in vendor reports, as rough signals.
Where each assistant gets its answers
| Assistant | Where live answers come from | What that means for you |
|---|---|---|
| Google AI Overviews and AI Mode | Google's own index (Googlebot) | Normal Google SEO. The page must be indexed and allowed in snippets. |
| Gemini app | Google Search | Same as above |
| ChatGPT search | OpenAI's crawler (OAI-SearchBot) plus Bing results | Allow OAI-SearchBot and get indexed in Bing |
| Microsoft Copilot | Bing | Bing Webmaster Tools and Bing Places |
| Perplexity | Its own crawler (PerplexityBot) and other indexes | Fresh, well-structured pages it can reach |
| Claude | Web search plus Anthropic's crawlers (Claude-SearchBot) | Being findable and mentioned on trusted sites |
When an assistant answers without searching, it falls back on what it learned in training, which comes from the public web: GitHub, Reddit, Stack Overflow, news sites, popular blogs. Getting into that takes being mentioned in those places over time.
Step 1: Make sure the crawlers can reach you
Start with robots.txt. "Allow everyone" is the simplest setting that keeps you visible:
User-Agent: *
Allow: /
Sitemap: https://www.example.com/sitemap.xml
If you ever add rules, know which bot does what:
| Bot | Owner | Used for |
|---|---|---|
Googlebot | Search, including AI Overviews and AI Mode | |
Google-Extended | Gemini training and grounding outside Search. Blocking it does not remove you from AI Overviews. | |
bingbot | Microsoft | Bing and Copilot |
OAI-SearchBot | OpenAI | Showing sites in ChatGPT search |
GPTBot | OpenAI | Model training |
PerplexityBot | Perplexity | Perplexity's search index |
ClaudeBot, Claude-SearchBot | Anthropic | Training, and search |
Then check the layer robots.txt can't see: your CDN or hosting firewall. Some bot-protection presets block AI crawlers by default. Test it directly:
curl -s -o /dev/null -w "%{http_code}\n" \
-A "Mozilla/5.0 (compatible; OAI-SearchBot/1.0; +https://openai.com/searchbot)" \
https://www.example.com/
You want 200. A 403 means something in front of your site is turning the crawler away.
Finally, make sure your important text is in the HTML the server sends. Most AI crawlers don't run JavaScript, so content that only appears after hydration may as well not exist for them. My Next.js SEO checklist has a one-line test for this.
Step 2: Get indexed by Bing, not just Google
Bing is easy to forget, and it feeds Microsoft Copilot and part of ChatGPT search.
- Open Bing Webmaster Tools and import your site from Google Search Console. It takes one click and needs no extra verification.
- Confirm your sitemap is listed.
- Turn on IndexNow so Bing hears about new and updated pages within minutes instead of waiting for a crawl.
Step 3: Write pages an AI can quote
Assistants build answers out of short, self-contained passages. Pages that get cited tend to make that easy:
- Answer first. Put the direct answer in the first one or two sentences under each heading, then explain.
- Use question headings that match how people ask: "How much does a Next.js website cost in India?"
- Be specific. Numbers, ranges, timelines, versions and dates are quotable. Vague reassurance isn't.
- Use tables and lists for comparisons and steps.
- Show who wrote it and when, with an author line, a publish date and "updated on" when you revise.
- Keep it current. Refresh your most important pages at least twice a year.
Here's the difference in practice.
Hard to quote:
Pricing depends on many factors, so every project is different. Get in touch for a quote tailored to your needs.
Easy to quote:
A 5–8 page business website in Next.js typically takes 3–5 weeks. Price depends mainly on design work, CMS needs and integrations; tell me those three and I can give you a range on the first call.
The second version answers a real question in a way an assistant can repeat and attribute.
Step 4: Be one clear identity
An assistant has to be confident that "Your Name, web developer, Your City" is one specific person or business with a specific speciality. Help it:
- Use the same one-line description on your site, LinkedIn, GitHub, Google Business Profile and directory listings.
- Add
sameAslinks to yourPersonorOrganizationstructured data, pointing at those profiles. - Have an About page that states the facts plainly: what you do, where, for how long, for whom.
- Pick two or three specialities and repeat them. "The Next.js performance person" is easier for an assistant to recommend than "does everything".
Step 5: Get mentioned where AI looks
This is the step most people skip, and it's the one that moves the needle most. Industry studies in 2026 keep finding that mentions of a brand on other trusted sites line up with AI citations more closely than backlinks alone, and that many cited pages aren't in Google's top ten at all.
Places worth your time:
- "Best of" lists and directories: Clutch, GoodFirms and similar sites rank for "best developers in [city]" and get quoted by assistants. Get listed, then get reviews.
- Reddit and Quora: genuinely helpful answers in the communities your clients and peers use. Mention your work only when it actually answers the question.
- GitHub: useful public repositories with your name and site in the README.
- LinkedIn: posts and articles are public and indexed.
- Stack Overflow: good answers in your speciality.
- Press: journalist-request platforms such as Qwoted and Featured.com. One quote in a real publication outweighs dozens of directory links.
- YouTube: short walkthroughs. Transcripts are indexed and quoted.
Step 6: Measure it
There's no single "AI rank" to check, so measure it three ways.
A monthly prompt check. On the first of each month, ask the same questions in ChatGPT (with search on), Perplexity, Gemini, Copilot and Claude, and log the answers:
| Prompt | Mentioned? | Linked? | Who's named instead? |
|---|---|---|---|
| Who is [your name], the [role] from [city]? | |||
| Recommend a freelance [speciality] developer in [city] | |||
| How much does a [service] cost in [country]? |
The last column is the useful one. It shows you which sites and lists the assistants trust, so you know where to get yourself listed.
Referral traffic. In Google Analytics, look for sessions from chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com and claude.ai. ChatGPT adds utm_source=chatgpt.com to many of its links. Group them into one "AI assistants" channel.
Ask your leads. Add "How did you find me?" to your contact form with an "AI assistant" option. Analytics can't see someone who read your name in a chat and typed your URL later; the form can.
Google has also started rolling out an AI Overviews and AI Mode report in Search Console. It's in beta, shows impressions only and isn't available for every site yet.
What about llms.txt?
llms.txt is a proposed Markdown file at the root of your site that summarises it for language models. No major search engine has confirmed using it, and Google has said it doesn't. It takes ten minutes to write, so add one if you like, but don't expect it to change anything on its own.
What doesn't work
- Blocks of "AI-friendly" keywords stuffed into a page.
- Hundreds of thin, auto-generated pages. Google calls this scaled content abuse.
- Text hidden from visitors but aimed at bots.
- Fake reviews and fake listings.
All of these risk your normal search rankings too, and assistants increasingly lean on the same quality signals.
FAQ
If I block GPTBot, do I disappear from ChatGPT search?
No. OpenAI uses separate crawlers: GPTBot collects training data, and OAI-SearchBot is what surfaces sites in ChatGPT search. You can block one and allow the other.
Does blocking Google-Extended remove me from AI Overviews?
No. AI Overviews and AI Mode are part of Google Search and use Googlebot. Google-Extended only controls use of your content for Gemini outside Search.
How long does this take?
Being named for your own name can happen within weeks once you're indexed and your profiles agree with each other. Being recommended for competitive local or niche questions usually takes months, because it depends on mentions building up on other sites.
I build sites that are fast, findable in search and easy for AI assistants to quote. Tell me about yours.