How ready is Indian e-commerce for AI search?
We ran the same audit against 20232 Indian online stores and scored what an AI assistant can actually read, verify and quote. Every score links to the report behind it.
Where the 19,686 measurable stores landed
Scores run from 30 to 95. This audit measures a floor - whether a machine can fetch, read and identify the page - so a high score is a baseline met, not a guarantee of being cited.
A quarter of stores score below 63, half below 72, and a quarter above 79. The index reports each store's standing as a quarter of this run rather than a pass mark, because a fixed pass mark told us almost nothing: nearly every store cleared it.
A further 546 domains returned a bot challenge, disallowed our crawler in robots.txt, or did not respond. They carry no score: a blocked response is not a bad score, and reporting it as one would be wrong.
Where the points are lost
Median score on each pillar, out of 100, weakest first. This is the honest to-do list for the sector: the overall score hides it, because a store can clear the basics and still be unquotable.
Separately, on speed: Chrome has real-user field data for 10869 of these stores, and 6057 of them load their main content within Google's 2.5-second threshold. Speed is reported apart from the score on purpose - it is a different question from whether a machine can read the page.
Two cuts of the same data
Median score by platform
Detected from the served HTML - asset hosts, generator tags and script handles. Groups of fewer than three sites are left out.
Median score by category
Derived from each store's own homepage title and description. Anything that does not match cleanly is left Unclassified rather than guessed.
What most stores are missing
Individual signals, across 19686 audited stores. Each is one check on one site, counted once however many pages it appeared on.
Percentage of audited stores failing each check outright. Warnings are not counted here.
What the best stores do that the worst don’t
We ranked all 19,686 stores, then compared the top quarter -79 and above - against the bottom quarter, 63 and below, and asked one question: what do the good ones do that the bad ones don’t? Six of the eight biggest differences are the same thing - what your product pages tell a machine.
Read each row left to right: how many low-scoring stores do this, then how many high-scoring ones do. The last column is the difference. If you fix one thing, fix your product pages - that is where the gap is.
Your platform already decided most of this
Every store here was checked for one thing: does its product page say “this is a product” in code an AI can read? Below are the platforms with at least twenty stores in the index. 96% of stores on Zoho Commerce pass it. 7% of stores on WordPress (no WooCommerce) do. Almost none of them chose that - they chose a platform, and it came with the answer.
Stores is how many sites in this index run on that platform. Pass is how many of them get this right. If yours is near the bottom, that is usually a plugin or a theme setting - not a reason to rebuild the shop.
Where product data falls apart
Describing a product to a machine is not one thing, it is five, and each one is close to useless without the one before it. A price with no offer around it, an offer on a page that never says it is a product - an assistant reads those the way you would read a price tag lying on the floor.
Out of 17,827 stores whose product pages we could read. The amber number is how many fall out at that step.
Stores that lock the door
Everything else on this page is about being readable. This is about being allowed. A store that disallows a crawler in robots.txt cannot be read, quoted or recommended by the assistant behind it, and no amount of markup changes that.
2,181 of 19,686 stores.
A blanket rule, not a position on AI. Search engines are usually exempted by name and everything else is turned away.
The only stores here that have actually decided something about AI. Almost nobody has.
807 stores publish no robots.txt at all. That allows everything by default, which happens to be the right outcome here, but it is luck rather than a decision.
If each store fixed one thing
For every store in the index we worked out which single failing check would lift its score furthest, then counted how often each one came top. This is not the most-failed list. It is the list of what to do on Monday morning.
Stores is how many have this as their single biggest win. Points is what it is worth to them on average, out of 100.
Does a .in domain read any better?
Every store here sells in India. Some say so in the domain and some do not, and it is a fair question whether that tracks anything real about how the shop is built.
The stores an assistant cannot read
Every blank in this index is a result. If we cannot tell what a shop sells after reading its homepage and thirty of its pages, a shopping assistant reading the same HTML cannot either - and it has no support email to fall back on.
Nothing on the homepage names the trade in words a machine can read.
We followed the sitemap and the homepage links and found nothing that behaves like a product page.
No platform fingerprint in the served HTML. Usually a custom build, sometimes a heavily themed one.
Blocked our crawler, disallowed it in robots.txt, or never responded.
Fast and readable are different problems
Split these stores by PageSpeed score and the readiness medians barely move. Speed work and AI-visibility work do not substitute for each other, which is why speed is reported here and kept out of the score.
On speed itself, Chrome has real-user field data for 10869 of these stores. Median largest-contentful-paint is 2,367ms, and 6057 of 10869 load their main content inside Google’s 2.5-second threshold. Lab numbers are never mixed into that figure.
PageSpeed data powered by pagespeed.cloud.
The ten most readable stores in the run
Highest overall scores. Each links to the full report showing every signal and the URL it was measured on.
| 1 | novamaxindia.com | Unknown | 95 | Report → |
| 2 | bigfanmerch.com | WooCommerce | 94 | Report → |
| 3 | crowcrowcrow.com | Headless / Next.js | 94 | Report → |
| 4 | foxtale.in | Shopify | 94 | Report → |
| 5 | greenglownursery.com | WooCommerce | 94 | Report → |
| 6 | kashisarees.com | WooCommerce | 94 | Report → |
| 7 | metabucosmetics.com | Shopify | 94 | Report → |
| 8 | riddhihealthcare.com | WooCommerce | 94 | Report → |
| 9 | antimonyindia.com | Shopify | 93 | Report → |
| 10 | aoneglassandcrockerystore.in | WooCommerce | 93 | Report → |
Most readable store in each category
Categories with at least five measured stores, so “best” means something.
The stores we could not measure
These carry no score. A site that refuses crawlers has not earned a low mark - but it is also, by that same configuration, invisible to the assistants its customers are asking.
- Returned a bot challenge
- 250
- Disallowed our crawler in robots.txt
- 67
- Did not respond
- 229
Method
- What was fetched
- A 30-URL sample per site, or the whole site where it has fewer: the homepage, then up to twenty product pages, five category pages and four others, sampled across the sitemap. Product pages are identified by what they do - an offer, a price in the markup, a working cart - not by what they declare, because a store with no Product schema is exactly the store this index needs to reach. We identify as TikboAuditBot, obey robots.txt, and wait at least two seconds between requests to a host.
- What it does not measure
- How well a store sells, ranks or converts. A low score is a statement about machine readability, not about the business. Page speed is reported separately and is deliberately not part of the score.
- Corrections
- Every score links to its full report, showing each signal and the URL it was measured on. If something is wrong, tell us and we will re-run the site and publish the correction with its date.