25 Indian D2C Sites Checked: All Have llms.txt

Why we ran this study

Everyone talks about 'AI readiness' and almost nobody publishes checkable numbers for the Indian market. So we picked 25 well-known Indian D2C websites, verified each was live, and ran the same battery of checks our software runs — llms.txt, robots.txt AI-crawler rules, JSON-LD schema types, homepage titles and meta descriptions. Every number below is from a live HTTP check on 23 October 2026, and the raw per-site data is downloadable so you can verify any cell yourself.

Finding 1 — 100% have llms.txt. Almost none of them wrote it.

This floored us, briefly. All 25 sites serve an llms.txt file — which would make India a world leader in llms.txt adoption. Then we opened the files: nearly every one is identical 'Agent Instructions / Shop skill' boilerplate with only the brand name swapped — auto-generated by Shopify for its stores. (We ruled out false positives by checking content types and testing garbage paths on the same domains.)

The real story: Indian D2C brands became 'AI-ready' on paper without a single merchant lifting a finger. Shopify's default file tells AI agents how to browse a catalogue — it does not tell them why anyone should buy from THIS brand, what makes it trustworthy, or which pages matter. The layer that influences AI recommendations is still almost universally missing, and any brand that writes a real one today is ahead of 24 of these 25.

Finding 2 — the door is wide open: only 1 in 25 blocks AI crawlers

Just one site blocks GPTBot in robots.txt; none block PerplexityBot. Contrast with Western publishing, where blocking OpenAI's crawler became common. Indian D2C has decided — mostly by default — to be fully visible to AI engines. That means AI-visibility winners here will be decided by content quality and structured data, not by access. Which brings us to the actual gap:

Finding 3 — the real gap: FAQ schema at 12%, titles failing at 64%

92% of the sites carry some JSON-LD schema and 84% have Organization or Store markup — again, largely platform defaults doing the work. But FAQPage schema — the single strongest 'quote me' signal for AI answers — appears on just 3 of 25 sites. And 64% of homepages have title tags longer than the 60 characters Google displays. The basics are platform-handled; the layer that wins citations is almost untouched.

What this means if you run a business in India

• If you are on Shopify: you have a generic llms.txt. Replace it with one that actually describes your brand — the free llms.txt generator writes it in two minutes.
• If you are NOT on Shopify: you likely have no llms.txt at all, which our full study will test on traditional SMB sites next. Five minutes fixes it.
• Everyone: add FAQPage schema to pages with real FAQs (the free schema generator builds it), and get your title under 60 characters — the single most common fixable mistake we found.
• Check where you stand on all of these in 60 seconds with the free scan.

Methodology, limitations and the raw data

Full methodology, the per-site results table and the downloadable CC-BY dataset live on the research page. Honest limitations: n=25 is a pilot and the sample is online-first D2C brands, mostly on Shopify — not representative of Indian SMBs broadly. The full two-cohort study (D2C brands vs traditional service businesses) is in progress; the gap between those two groups is the question we want answered next. Journalists and bloggers: quote any number with a link — that is what the dataset is published for. If you would rather have these exact checks run on YOUR site automatically, that is one job of the 34 tools in the GrowKitt desktop app.

Frequently asked questions