Indian ecommerce sites are easy for humans to browse, but the picture changes sharply when the shopper is an AI agent. FTA Global’s Agent Readiness Study 2026, released on October 6, put the average readiness score at just 43 out of 100 across the 42 Indian ecommerce websites that could be fully assessed.
The study set out to see how well 50 leading Indian ecommerce sites support AI agents that help consumers discover, compare and buy products. Agents completed 358 of 450 assigned shopping tasks, or 80%. Return-window questions were the easiest at 91%, followed by delivery-charge checks at 74% and product-and-price discovery at 73%.
Who is doing it well
The clearest leaders were boAt (95/100), Flipkart (92), Mamaearth (92), Moglix (92) and Truemeds (91). Their common advantage: they let agents access product content and kept essential details such as price and stock inside server-side HTML and structured data.
The spread was wide. Twelve sites scored 80 or above, while 18 fell below 20.
Where agents get stuck
The most interesting finding is that blocking is not the main problem. Of 92 failed tasks, 26 failed because the answer was not clearly stated on the website. Thirteen failed because content required JavaScript execution. Only 11 were tied to blocking or site errors.
Bot blocks also turned out to be less effective than many brands assume. In the study, 21 sites blocked or failed FTA’s generic cloud crawler. Real AI agents still completed 77% of tasks on those sites, compared with 82% on sites that stayed open.
Structured data remains a weak spot. Return policies appeared in schema on just three of 24 product pages, and delivery details on only two. The lack of consistency showed up in agent answers too: all three AI agents returned the same return-window answer on 29 sites, but they disagreed on 13, usually because sites had multiple or contradictory rules.
The framework behind the score
FTA Global assessed each website across four dimensions: Reach, Read, Understand and Act. The Trust pillar was left out because it needs human review.
“AI agents are changing what it means for an ecommerce website to be discoverable.”
— Senthil Kumar Hariram, Founder and Managing Director, FTA Global
The study crawled sites on September 26 and assigned Claude, GPT and Gemini the same three tasks on each of the 50 websites on September 27, producing 450 real-agent tasks. The company notes the sites are a selection of prominent Indian ecommerce businesses, not a ranking of the top 50 by revenue.
What marketers should do
FTA Global recommends making critical shopping information explicit and accessible:
- State delivery charges and return windows clearly.
- Serve product names, prices, stock and variants without requiring JavaScript.
- Keep robots.txt rules open to shopping and AI search agents.
- Review firewall and CDN configurations that could obstruct agents.
For ecommerce marketers, the message is plain: if a machine cannot answer basic questions about price, delivery and returns, the brand may disappear from AI-assisted purchase journeys, even if human customers can still see the page.
Source: MediaNews4U
