llms.txt Adoption in India: Surprising Data From 1,100 Sites

Everyone in SEO has an opinion on llms.txt. Very few people have checked who in India actually uses it. So I did. I requested /llms.txt from the top 1,000 .in domains and from 100 of the biggest Indian brands, then opened every file that came back to see whether it was real. This is the first original study of llms.txt adoption in India that I know of, and the results are more uneven than the hype suggests.

Short version: 12.7% of India’s top 1,000 .in sites now serve an llms.txt file. That is ahead of global benchmarks. But not one of 272 government domains has one, plenty of the files are broken, and some of the biggest names in Indian ecommerce and banking have none at all. Indian teams training or fine-tuning their own models can compare the best GPU cloud providers in India.India

Below you will find the method, the full numbers by segment and industry, the brands that have one and the ones that don’t, the mistakes I saw again and again, and a simple template if you want to add your own. Gathering data from thousands of sites at scale usually needs proxies, so see our picks for the cheapest residential proxies for scraping.

llms.txt adoption in India: key findings

• 12.7% of the top 1,000 .in domains serve an llms.txt file (127 sites). Remove government domains and the figure rises to 17.4% (127 of 728).
• Government: 0 of 272. No .gov.in or .nic.in domain in the sample has one.
• 41 of 100 major Indian brands have an llms.txt file. Travel and local services lead at 60%. News and media trail at 20%.
• Quality is patchy. Among the 127 .in adopters, 103 follow the basic format. The other 24 are missing a proper H1 or links, are placeholders, or are actually robots.txt rules saved under the wrong name.
• llms-full.txt is rarer. 55 of the 127 .in adopters also serve llms-full.txt. Among the big brands, only 4 do: Razorpay, BankBazaar, upGrad and Postman.
• Big names are missing. Flipkart, Myntra, Swiggy, PhonePe, HDFC Bank, ICICI Bank, Infosys and Zoho did not serve a valid llms.txt file on the day I checked.

What is llms.txt, in plain words?

llms.txt is a plain markdown file at the root of a website, at yoursite.com/llms.txt. It was proposed in 2024 by Jeremy Howard of Answer.AI, and the format is documented at llmstxt.org. The idea is simple: give AI tools a short, clean map of your most useful pages, so they don’t have to dig through menus, scripts and cookie banners to understand what you do.

A valid file starts with an H1 containing the site or brand name, then a one-line summary in a blockquote, then sections of links with short descriptions. That’s it. There is no special syntax beyond basic markdown.

One honest caveat before the numbers. No major AI company has publicly confirmed that its search or answer engine reads llms.txt when ranking or citing sources, and Google has said it does not use it for Search. So this study measures llms.txt adoption in India, not its impact. Read it as a signal of how Indian websites are preparing for AI search, not as proof that the file gets you cited.

How I ran the llms.txt adoption in India study

Here is exactly what was checked, so you can judge the numbers or repeat the study yourself.

1. Sample 1, top 1,000 .in domains: taken from the Tranco top 1 million list, which ranks sites by combined traffic signals. I kept the first 1,000 domains ending in .in, including .co.in, .ac.in, .edu.in, .gov.in, .nic.in, .org.in and .net.in. Adult, gambling and lottery sites stay in the counts but are never named in this report.
2. Sample 2, 100 major Indian brands: a hand-picked list across seven categories (ecommerce, fintech and banking, news and media, travel and local, jobs, education and health, auto and real estate, tech and SaaS). Most are on .com, which is why they are not in the .in sample.
3. The check: an HTTPS request for /llms.txt and /llms-full.txt, following redirects. A site counted only if it returned a non-empty text file.
4. False-positive filter: many sites return a 200 status with an HTML error page or a CDN Access Denied page. Those were rejected. Every adopter was then fetched again with a normal browser user agent and re-classified.
5. Date: all checks were run on 8 October 2026. Websites change often, so treat this as a snapshot.

Sites that block automated requests completely may be undercounted, so these figures are a floor, not a ceiling.

llms.txt adoption in India: the headline numbers

Out of the top 1,000 .in domains, 127 serve an llms.txt file. That is 12.7%. For context, published global trackers have put adoption among the world’s top 1,000 sites at roughly 9%, so Indian sites are not behind on this one. If anything, they are slightly ahead.

The bigger story sits underneath the average. Government domains make up more than a quarter of the .in top 1,000, and their adoption is zero. Strip them out and llms.txt adoption in India’s commercial and educational web is 17.4%, or about one site in six.

Does traffic rank make a difference?

You might expect the biggest sites to adopt first. They don’t. Once government domains are removed, adoption is close to flat across the list, at around one in five sites in every band.

That pattern says something useful about llms.txt adoption in India. llms.txt is not a big-company project that needs a dedicated team. Mid-sized Indian sites, many on WordPress and Shopify, are adopting it just as fast, often because their SEO plugin or agency added it in an afternoon.

Indian brands with llms.txt: 41 out of 100

The brand sample tells a sharper story about llms.txt adoption in India. 41 of the 100 brands serve an llms.txt file, which is far higher than the .in average. Large consumer brands clearly see AI assistants as a channel worth preparing for.

Travel leads llms.txt adoption in India for an obvious reason. Trip planning is one of the most common things people ask ChatGPT and Gemini to help with, so online travel agencies want their routes, fares and policies to be easy for AI to read. Fintech brokers like Zerodha, Upstox and Angel One fall into the same bucket: people ask AI about trading and investing every day.

The gaps are just as telling. Most private banks, quick-commerce apps and big news publishers have stayed out. For publishers it may be deliberate, as many are cautious about AI companies using their content. For banks it is more likely slow internal approvals.

Why does no government site in India have llms.txt?

272 government domains were checked, including central ministries, state portals and nic.in services. None returned a valid llms.txt file. This is the single starkest finding on llms.txt adoption in India.

It matters more than it seems. A huge share of what Indians ask AI tools involves government services: how to apply for a passport, how to link PAN and Aadhaar, the latest scheme eligibility, exam results. When an assistant answers those questions, it currently has to piece the answer together from scattered portal pages, PDFs and third-party blogs. A clean llms.txt pointing to the official process pages would be a low-cost way to make AI answers about public services more accurate.

Educational institutions are moving faster. 10 of 93 university and college domains have one, often added through the same WordPress SEO plugin that runs the rest of the site.

How good are the files? Most pass, many don’t

Having a file is not the same as having a useful one. I graded every llms.txt file against the basic format: an H1 at the top, and at least three real links to pages on the site.

Put simply, llms.txt adoption in India is ahead on quantity but behind on quality: roughly one in five Indian llms.txt files has a problem that would confuse a machine reading it. Here are the patterns I saw most often.

The 5 most common llms.txt mistakes

1. No space after the hash. Files starting with #Pepperfry, #Cleartrip or #Bharti Airtel Limited are not valid markdown headings. Markdown needs # Name with a space. It is a one-character fix.
2. Template text left in place. One major travel brand’s file opened with # Title, the literal placeholder from a template. Another began with the words Sample llms.txt file.
3. Robots rules in the wrong file. Some sites, including a large public sector bank and a travel portal, filled llms.txt with User-agent and Disallow lines. Crawler permissions belong in robots.txt. llms.txt is a content map, not an access policy.
4. An error page served as a file. One university domain returned a 200 status with a body that just said 404. To a machine, that looks like a real file with no content.
5. Headings with no links. Several files describe the company in paragraphs but never link to a single page. Without links, the file can’t do its main job of pointing AI to your best content.

I also found 8 files that announced they were generated by an SEO plugin: Rank Math, Yoast or All in One SEO. Plugin output is a reasonable starting point, but a few of those put the generator note above the H1, which pushes the brand name off the first line. If you use one, open the file once and tidy the top. Our Rank Math review shows how we cleaned up ours with a one-line filter. I covered Rank Math’s newer AI features in the Rank Math MCP server setup guide.

Should your business add an llms.txt file?

My honest answer is yes, but with low expectations. It takes about 30 minutes, costs nothing, and does no harm. It will not, on its own, get you cited in ChatGPT, Perplexity or Google AI Overviews.

What actually drives AI citations is the same as before: clear, factual pages, consistent business details across the web, a strong entity profile, and crawler access for bots like OAI-SearchBot and PerplexityBot. I explain how those pieces fit together in GEO vs AEO vs SEO, and the local version for small businesses is in how Lucknow businesses get recommended by ChatGPT.

Think of llms.txt as a tidy index for a library that is still deciding whether to read indexes. When AI tools do start using it more, the 17% of Indian sites that already have a clean one will be ready. The rest will be catching up.

How to create an llms.txt file in 4 steps

1. List your 10 to 30 most useful pages. Services, pricing, key guides, policies and contact details. Skip tag pages, login pages and thin archives.
2. Write the file in plain markdown. Use the template below. One H1, a one-line summary, then H2 sections with links and short descriptions.
3. Upload it to your root folder so it loads at yoursite.com/llms.txt as plain text. On WordPress, Rank Math, Yoast and All in One SEO can generate one, but check the output.
4. Test it. Open the URL in a browser. You should see raw text, not an HTML page, a redirect to your homepage, or a CDN error. Then check it again after any site migration.

# Your Brand Name

> One sentence on what you do, who you serve and where.

## Services
- [Service one](https://yoursite.com/service-one/): what it is and who it is for
- [Service two](https://yoursite.com/service-two/): what it is and who it is for

## Guides
- [Most useful guide](https://yoursite.com/guide/): the question it answers

## Company
- [About](https://yoursite.com/about/): who runs the business
- [Contact](https://yoursite.com/contact/): phone, email and address

If you want a second pair of eyes, an SEO audit can cover llms.txt alongside robots.txt, schema and crawler access. For the bigger goal of being named in AI answers, see the LLM citation visibility service and the entity and knowledge graph work that supports it.

Limits of this study

• It is a single-day snapshot. Sites add and remove files all the time.
• Sites that block all automated requests may have a file that I could not see, so true adoption could be slightly higher.
• The brand list of 100 is hand-picked and leans towards consumer brands. It is useful for comparison, not as a census.
• The quality grades check structure, not usefulness. A file can follow the format and still link to the wrong pages.

I plan to re-run this llms.txt adoption in India check every six months so the trend becomes visible. If you run a large Indian site and want your result checked, or think something here is wrong, get in touch.

Frequently asked questions