IndiaMART has moved content behind logins, blocked unrestricted scraping of its website and withheld its proprietary data from Google Search and AI agents. On the Q2 FY27 analyst call, co-founder and CEO Dinesh Agarwal said the company was "okay to let go of 10-20% of its current unique business enquiries" to protect its data moat, as MediaNama reported on 2 November 2026. The B2B marketplace is choosing to shrink its audience to avoid feeding systems it does not control.
What actually happened
According to MediaNama, citing the company's investor presentation, IndiaMART recorded 27 million unique business enquiries in July-September, down 14% year on year. Agarwal said enquiries were 10-15% below their peak. He attributed the fall partly to buyers moving to AI search and partly to IndiaMART's own decision to restrict the data that used to feed those searches.
His explanation was blunt: "Google AI Overviews had expanded from 10-20% to now 50% of searches or even higher." With so much LLM-based agent activity underway, he said, "everyone wants to protect their own data and unique advantage." IndiaMART says it is open to licensing redacted, non-personal data for AI training, but only "if there is an opportunity to earn revenue."
These are the CEO's statements on an earnings call. The Google AI Overviews figure is his characterisation, not an independently verified measurement. The enquiry decline is the company's own reported number.
The case for worry
The strongest argument for intervention is about the commons. Open web indexing made it possible for small suppliers, comparison sites, researchers and newcomers to find and build on public information. If large platforms respond to AI extraction by walling off their data, discovery gets worse and the web fragments into private gardens. A regulator could reasonably worry that the first movers who scraped freely have pulled up the ladder, and that dominant search and AI firms are using publicly posted business data to compete with the very sites that created it.
There is a second concern. IndiaMART's own suppliers and buyers generate this data. If a platform locks it down, those users could become less visible to the buyers they most want to reach.
Why a mandate is the wrong response
The lockout is better read as a market correction than a market failure. IndiaMART is not breaking a rule or abusing a bottleneck. It is acting on ownership and cost. Hosting, curating and verifying millions of supplier listings is expensive, and the company is declining to subsidise a downstream product that may substitute for its own.
The tools it is using are narrow and reversible. A login wall, rate limits and bot restrictions are ordinary access controls that any site owner can adjust. The robots exclusion standard, RFC 9309, has long let publishers signal crawl preferences. The same document says plainly that "these rules are not a form of access authorization." Voluntary signals have never been enough on their own, which is why platforms are turning to authentication.
A rule forcing publicly accessible commercial data to stay open to search engines or AI crawlers would mean ordering a private firm to give a rival its work for nothing. It would also punish the investment in verification and quality that makes the data valuable in the first place. Data-access mandates tend to freeze today's market structure into law and to favour whoever already operates at scale.
The cost is already being priced in
The evidence that markets are working is in IndiaMART's own numbers. The company is accepting a hit of up to a fifth of its unique enquiries. A firm does not volunteer for that unless it believes the data is worth more kept than given away. If the trade-off proves wrong, shareholders, not regulators, bear the loss, and the company can reverse course.
MediaNama also reports that IndiaMART is piloting a paid verified-buyer programme at Rs 1,000 to Rs 10,000 per buyer per year. That is a revenue model that assumes controlled access has value. A compliant licensing market is forming around that logic, and the CEO's own words describe it: redacted, non-personal data, offered when it earns revenue.
What policymakers should do instead
India's priority should be to make licensing workable rather than to dictate openness:
- Back clear contractual and technical norms. Access terms and machine-readable crawl signals should be treated as enforceable preferences, so that publishers do not need to build walls to be heard.
- Keep personal data out of the bargain. IndiaMART's offer covers redacted, non-personal data only. Regulators should insist that any AI-training licence respects that boundary, which is a privacy question, not an openness one.
- Watch for real abuse. If a dominant search or AI provider conditions market access on free data extraction, that is a competition issue to be tested case by case with evidence. A lockout by a private marketplace does not meet that bar.
- Measure before legislating. The enquiry decline mixes at least two causes: buyers shifting to AI search, and IndiaMART's own restrictions. Policy built on one number would be policy built on a guess.
The lesson for a pro-innovation regime is restraint. The open web works because most publishers choose to participate on terms they can live with. When AI agents change those terms, publishers will renegotiate, wall off or license. Letting them do so, while protecting personal data and policing genuine abuse of dominance, keeps incentives to build valuable datasets intact. Forcing access would remove the reason to build them.