platform & marketplace strategy
Marketplaces are marketplaces first. The transaction is the product. Everything else is a feature.
If you are the PM on a two-sided marketplace or platform, you do not have one product. You have at least two — and what you build for one side often damages the other.
That is the fundamental tension that makes platform PM work different from everything else. A feature that attracts more buyers puts pressure on sellers to respond faster and handle more volume. A policy change that protects users can kill supply overnight. An algorithm tweak that increases engagement may reduce transaction quality. Every decision ripples across sides.
Most PMs who move into platform roles treat it like a normal product — ship features, track DAU, grow revenue. They fail within six months. This page is about why platforms are structurally different, and what you actually need to know to build one well.
What makes a platform a platform
A marketplace connects buyers and sellers. A platform enables interactions between two or more distinct user groups, where value is created by those interactions rather than by the platform itself.
The distinction matters. A SaaS product creates value by delivering software. A platform creates value by enabling exchanges — of goods, services, information, or attention. The platform’s job is to make those exchanges happen reliably, at scale, and at a margin.
There are three types you will encounter in India:
Transaction marketplaces. Flipkart, Urban Company, Dunzo. The platform facilitates a purchase. Revenue comes from take rate — a percentage of the transaction value.
Attention/media platforms. YouTube India, Moj, ShareChat. One side creates content, the other side consumes it. Revenue comes from advertising (charged to the consumption side’s attention).
B2B platforms. Zoho Marketplace, Razorpay Thirdwatch, IndiaMART. One side provides tools or services; the other side uses them to run their business. Revenue comes from subscriptions or API calls.
Each type has different success metrics, different cold-start problems, and different failure modes. The strategic decisions that work for Urban Company will not transfer to IndiaMART. Know which type you are building before you borrow any playbook.
Network effects: types and traps
Network effects are when the product becomes more valuable as more users join. Every PM who works on a platform uses this phrase. Very few can correctly identify which type of network effect they actually have — or whether they have one at all.
There are three that matter in practice:
Direct (same-side) network effects. Value increases as more users on the same side join. WhatsApp is the canonical example — every Indian user who joins makes the product more valuable for every other Indian user. These are the strongest moats. They are also uncommon. Most marketplaces do not have them.
Indirect (cross-side) network effects. Value increases as the other side grows. More Ola drivers means shorter wait times for riders. More riders means higher utilization for drivers. This is how most two-sided platforms work. Cross-side effects create a flywheel, but they also create asymmetry — one side typically drives the other, and you need to identify which.
Data network effects. The product improves as it collects more data, which in turn attracts more users, which generates more data. Swiggy’s delivery time estimates improve with each order — better predictions attract more users, who generate more orders, which improve the model. This is a slower flywheel but creates deep moats that are nearly impossible to replicate quickly.
The trap is claiming network effects when you do not have them. A 2016 interview cited in our knowledge base quoted Marc Andreessen on this directly — network effects are often overstated. Google Plus had access to 500 million users and feature parity with Facebook but could not replicate Facebook’s network. Why? Because network effects are not about size — they are about whether an additional user on one side creates real value for the other side. If switching platforms is cheap and the value from additional users is marginal, you do not have a moat. You have a product.
The diagnostic question: “If I added 1000 users tomorrow, would the 1001st user experience a meaningfully better product than the 10th user did?” If the answer is yes, you have network effects. If the answer is “maybe a little,” you do not.
Product review at a B2B services marketplace in Bangalore. 18 months post-launch.
CEO: “We have 5,000 service providers and 80,000 customers. Why aren't we growing faster? We should have strong network effects by now.”
PM: “I've been looking at this. Our supply is geographically fragmented — a plumber in Whitefield doesn't help a customer in Jayanagar. And our category depth is thin. The average customer transacts twice a year. There's very little compounding happening.”
CEO: “So we don't have network effects?”
PM: “We have weak local ones, but they're not the growth engine we assumed. Our real moat is review density — customers trust us because of the review volume. That's a data effect, not a classic network effect. We should be doubling down on that, not just growing supply.”
CEO: “What does that mean for the roadmap?”
PM: “Stop expanding to new cities for 6 months. Go deep in three metros — get review density per category above the threshold where a buyer feels confident. Then expand.”
Misidentifying the type of network effect had led them to a geographic expansion that spread supply too thin to create real density anywhere.
Not all network effects are equal. Growing both sides simultaneously is not always the answer — sometimes density matters more than scale.
The chicken-and-egg problem
Every two-sided platform launches with zero supply and zero demand. Neither side has a reason to show up until the other side does. This is not a marketing problem. It is a structural design problem, and how you solve it in the first 12 months shapes the platform’s trajectory for years.
There are four approaches that actually work. Most successful platforms use a combination.
Seed one side heavily before launching the other. YouTube seeded its supply side by solving a genuine tool problem — video hosting was genuinely hard in 2005. Early YouTubers used YouTube to host videos they shared on other sites (MySpace, forums). They were not there for the YouTube audience; they were there for free infrastructure. The audience came later. The lesson: your supply side does not need a demand side if you give them something valuable in standalone mode.
Manufacture the missing side. Reddit had almost no real users in its early months. The founders created dozens of fake accounts and posted content to simulate an active community. Quora had the same challenge with answers — early employees wrote the first thousands of answers to make the platform feel alive. PayPal seeded its supply of merchants by buying goods on eBay using PayPal and requiring the seller to set up a PayPal account to receive payment. This feels like deception, but it is not — it is solving a coordination problem until real liquidity arrives.
Target a market where both sides overlap. One of the most underrated tactics: find a user group that is both buyer and seller. Early Airbnb targeted design conference attendees — people who were renting out rooms at one conference would become guests at the next. The Quora notes in our knowledge base describe this well: “Make the two-sided network one-sided — target a specific group which has both the consumers and producers of your content.” Fiverr and Upwork benefited from freelancers who also hire for projects they cannot do themselves.
Subsidize one side temporarily. Offer the supply side favorable economics to show up first, then normalize take rates once demand is established. Urban Company subsidized service provider acquisition in its early years — free training, guaranteed work, equipment loans. That supply quality and density attracted demand, which let them raise take rates later. The economics only work if you have the runway to sustain the subsidy and a clear trigger for when to normalize it.
Pick a two-sided marketplace you know well — one you use as a customer, or one you are building.
Answer these four questions:
- Which side is harder to acquire — supply or demand? Why? (In India, supply is usually the constraint because formal supply in most categories is fragmented and skeptical.)
- Does your supply side get standalone value from your platform even without demand? If not, can you create that?
- Is there a specific geographic or category niche where you can achieve density fast, rather than spreading thin across the country?
- What is your time horizon to natural liquidity — the point where supply and demand reinforce each other without subsidy? If it is more than 18 months, your unit economics may not survive.
There is no single right answer. The exercise is to force explicit thinking about the chicken-and-egg problem before you start spending.
Which side to charge — and how much
This is the decision that most platform PMs get wrong early. The intuition is: charge both sides, maximize revenue. The reality is: charge the side that is less price-sensitive and more abundant, and subsidize the side that is scarcer.
In two-sided markets, you rarely have symmetric pricing power. One side is typically the constraint — the side that you need more of to drive value for the other. That side should be subsidized or charged less. The other side pays more because it is getting access to a scarce resource.
The examples from India are clear:
Naukri charges recruiters, not job seekers. Job seekers are abundant; recruiters with hiring budget are the scarce side. Free access for job seekers maximizes resume supply, which is what recruiters are paying for.
Razorpay charges merchants (supply of payment acceptance), not users. There is no shortage of people with UPI apps. There is friction in getting merchants to accept digital payments reliably. Razorpay subsidizes merchant acquisition and charges a transaction fee that merchants absorb because the alternative (cash handling) costs more.
Hotstar has charged consumers for premium content — but its B2B platform (API access, distribution to device manufacturers) is how it builds supply-side reach. Different pricing for different sides of a media platform.
The rule: identify the scarcer, harder-to-acquire side. That side gets subsidized. The more abundant side — the one with stronger demand for access — gets charged.
One complication in the Indian market: price sensitivity is extreme on both sides. Sellers in agriculture, logistics, and gig economy segments resist any take rate above 5-8%. Buyers in B2C segments resist fees that are visible at the point of transaction. The workaround most Indian platforms use: hide the take rate in the gross margin (you buy at X, you see it priced at X+commission, the seller never sees the split). This is what Meesho, Amazon India, and most Indian e-commerce operates on. It is not deception; it is packaging.
Platform governance: the decisions nobody prepares for
Once your platform has meaningful supply and demand, you enter a new phase of product work that has nothing to do with features. You are now setting rules that govern interactions between parties who often have conflicting interests. This is governance, and it is the work that separates mature platform PMs from feature builders.
The governance questions that hit every marketplace PM:
Who bears the cost when a transaction fails? A customer in Mumbai books a home repair service, the provider does not show up, the customer is furious. Do you refund the customer? Do you penalize the provider? Who decides? Your policy here sets the behavioral norms for your entire provider network. Too lenient on providers and demand trust collapses. Too harsh and supply dries up.
How do you handle reviews that are gamed? Positive review gating — where sellers only ask happy customers to review — is endemic in every Indian marketplace. Flipkart, Amazon, Urban Company all deal with this. The policy decision is whether you allow seller-initiated review requests (yes, with constraints), block them (unenforceable), or create a verified-purchase review system (best, but requires infra investment).
At what point does a power seller become a platform risk? On most Indian marketplaces, 80% of supply value comes from 20% of providers. When that top tier threatens to leave, build their own platform, or negotiate special terms, what is your response? You need a concentration policy before the crisis, not during it.
How do you set algorithmic fairness rules? Your search ranking algorithm determines who eats and who starves. A vendor whose listing falls from page 1 to page 3 on Urban Company may see 60-80% revenue drop overnight. That is enormous power. Indian courts have begun looking at algorithmic visibility as a competition issue — how you rank, whether you favor your own products, and whether you explain ranking logic to providers are increasingly regulatory questions, not just product decisions.
Platform governance is not glamorous. It does not make it into roadmaps or OKRs. But governance failures destroy platforms faster than any growth misstep. CafeMedia in the US collapsed partly due to governance implosion. Quikr in India died partly because fraud eroded buyer trust to zero.
The practical approach: document your governance principles before you need them. “Provider interests are weighed equally with buyer interests in policy decisions” or “algorithmic ranking is based solely on quality signals, not on commercial relationships.” Write it down. Make it public. Then actually enforce it — because providers talk to each other, and the moment they conclude your platform is rigged, supply begins to leak.
The platform extension trap
Once a marketplace hits scale, the temptation is to expand vertically — offer services that compete with your own supply side. Amazon sells its own private-label products on its marketplace. Ola launched Ola Electric while still being the main distribution channel for Maruti and Hyundai drivers. Zomato acquired Blinkit and now competes with the Blinkit category on its own app.
The economics are compelling. You have distribution, data, and customer relationships. Adding first-party supply lets you capture full margin instead of just take rate. The strategic risk is severe. Every supplier on your platform who sees you competing with them recalculates their dependence. They invest less in your platform. They list more aggressively on competing platforms. They seed alternatives.
Amazon India’s private label strategy has driven many major Indian electronics sellers to prioritize Flipkart. Myntra’s private label brands — Roadster, HRX — create real friction with the branded fashion companies whose exclusives they also want.
The principle: vertical integration makes sense when supply quality is the constraint and first-party supply will set a higher standard that lifts the platform. It is destructive when it signals to existing supply that your platform is a customer acquisition channel, not a genuine marketplace. The test is whether your supply side believes you will compete fair — and whether they have enough alternatives that they can walk.
Metrics that actually matter on a platform
Most product metrics are supply-side metrics in disguise. GMV, orders shipped, providers active. What you actually need to track in both directions:
Liquidity rate. The percentage of demand-side searches or requests that result in a successful match. If 40% of Dunzo order requests result in no delivery partner being available, that is a liquidity problem — not a demand problem. Track liquidity at the category and geography level.
Take-up rate. Of supply-side listings or profiles, what percentage get at least one transaction per week? Low take-up rate means supply is listing but not transacting — which means you have ghost supply inflating your “active providers” metric while actual buyer experience is poor.
Cross-side satisfaction differential. If your buyer NPS is 65 and your seller NPS is 15, that gap will eventually become a supply crisis. Platforms that obsess over buyer satisfaction while ignoring seller satisfaction run out of quality supply. Track both. Investigate gaps.
Concentration risk. What percentage of your GMV comes from your top 20% of sellers? If it is above 75%, you have a structural governance risk. One seller strike or exodus and your metrics crater.
Test yourself
You are the PM for a home services marketplace in India, operating in 8 cities. Three months ago, you raised take rates from 15% to 22%. Active providers have dropped 28% in 45 days. Liquidity rate is down from 78% to 61%. Buyer NPS is down 9 points. Board meeting in 10 days.
Your CEO asks you to present a plan. What do you do first?
your path
Urban Company is considering opening its platform to independent beauty professionals who are not part of Urban Company's employed/trained workforce, as a way to rapidly expand supply in Tier 2 cities. The quality team opposes this because Urban Company's brand promise is trained and verified professionals.
The call: Do you open the platform to independent professionals, and under what conditions if any?
Urban Company is considering opening its platform to independent beauty professionals who are not part of Urban Company's employed/trained workforce, as a way to rapidly expand supply in Tier 2 cities. The quality team opposes this because Urban Company's brand promise is trained and verified professionals.
The call: Do you open the platform to independent professionals, and under what conditions if any?
Where to go next
- Understand how to price your platform: Pricing Strategy
- Build growth models for platforms: Growth Analytics
- Go to market in a two-sided context: GTM Strategy
- Track the right platform metrics: Metrics & KPIs