SStop Building Marketplaces

The most copied Indian startup pattern of the past decade was the marketplace. Two-sided, transaction-fee, take-rate, GMV-obsessed. Cabs, food, groceries, freelancers, doctors, tutors, lawyers, real estate, used cars, even prayer services. The structure looked clean on a slide. In practice, marketplaces in India have produced one or two real winners per category and a long tail of capital incinerators. There is a reason. Marketplaces are the wrong shape for most of the value Indian builders are now trying to capture.

The right shape is community-first. The argument is not aesthetic. It is structural.

¶What a marketplace optimizes for

A marketplace optimizes for transactions. Every dollar of product, design, and growth investment goes into making the next transaction faster, cheaper, and more frequent. The incentives that follow are predictable. Suppliers are pushed toward sameness so they are easy to rank. Buyers are pushed toward price comparison because differentiation is suppressed. Reviews become noise because reputation cannot be carried between platforms. Switching costs collapse. The platform extracts a take rate that funds growth, which funds discounting, which funds the next round of supplier commodification.

This is not a flaw in execution. It is what marketplaces are built to do. A well-run marketplace makes its suppliers interchangeable. That is the product.

¶What that produces in India

The Indian context turbocharges the dynamic. Price sensitivity is real and structural at most income levels. Discovery happens on WhatsApp and through word of mouth before it happens on any platform. Suppliers, especially small ones, are happy to take leads from a marketplace and convert the relationship off-platform on the second interaction. The take rate that funds the marketplace is paid, ultimately, by the supplier who has the weakest negotiating position. Margins compress, service quality drops, the brand of the marketplace becomes synonymous with cheap-and-fast rather than trusted-and-good.

The exceptions prove the rule. The marketplaces that survived in India did so by either becoming monopolies through extraordinary capital, or by quietly turning themselves into something else, a logistics company, a private label, a credit business, a payments rail. The marketplace layer itself is rarely where the durable margin lives.

¶What a community optimizes for

A community optimizes for relationships. Investment goes into making the next interaction more meaningful, not just more frequent. The incentives invert. Suppliers are pushed toward differentiation because reputation is carried within the network. Buyers are pushed toward fit because price comparison is not the dominant frame. Reviews are not noise, they are social capital. Switching costs are real because the relationship is the product. The platform charges in a different way, membership, subscription, premium access, advisory, and the unit economics do not require commodifying anyone.

¶The category-by-category case and the builder's choice

Look at any vertical and the inversion is visible by 2026. In professional services, the marketplaces for lawyers, accountants, consultants, and designers have largely been routed by closed communities where the same professionals refer each other directly. In hiring, the matrimonial-style job marketplaces are losing senior roles to private cohorts and curated rooms. In D2C, the brand communities, paid memberships and invite-only product drops, outperform the marketplaces on margin per customer by a factor that embarrasses the dashboards. In learning, the cohort-based courses with strong alumni networks have outlasted the MOOC marketplaces that were supposed to eat them.

Even in categories where the marketplace shape seemed permanent, food delivery, ride hailing, the supplier side is building communities behind the platform's back. Driver WhatsApp groups, restaurant owner forums, kitchen partner collectives. The platform thinks it owns the network. The network is forming elsewhere.

If you are starting now and your instinct is to build a marketplace, ask one question. What happens to your moat when an LLM-powered assistant can search, rank, and book across every marketplace in your category in three seconds with no human in the loop? In most cases the answer is brutal. The marketplace was a discovery and trust layer. AI eats discovery. Trust, by contrast, is not a search problem. It is a community problem.

This is why community-first wins the next decade. Communities produce trust that is illegible to a crawler. They produce relationships that an agent cannot reproduce. They produce premium pricing because the alternative, going to the open marketplace, is no longer attractive to anyone who can afford otherwise.

¶What to build instead

Pick the same vertical you would have picked for a marketplace. Climate consultants, GCC operators, school principals, family-office investors, paediatricians, kirana modernizers. Instead of building a two-sided board with a take rate, build a closed community with a membership fee and curated programming. Layer commerce on top only after the relationships are dense enough that the commerce is incidental. Charge for access, not for transactions. Refuse the temptation to open up for growth.

The marketplace will keep getting cheaper to spin up and harder to make money on. The community will keep getting more expensive to compete with once it exists. Choose the asset that compounds. Start the room this month.