You are sitting in a PM interview. The interviewer asks you to design a two-sided marketplace. You sketch a buyer side and a seller side, draw an arrow between them, and call it a completion. You do not get the offer.
Platform and marketplace design is one of the most misunderstood topic areas in PM interviews. Candidates treat it like a product feature problem. It is a network problem, a pricing problem, and a trust problem at the same time. This post walks you through the mental models that experienced PMs use when they approach platform design from scratch.
What a platform is
A platform connects two or more distinct user groups so each group gets more value from the presence of the other group. Airbnb without hosts is a search engine with no results. Airbnb without guests is a photo album. The value comes from the match.
That sounds simple on paper. The hard part is that both sides need to show up before either side gets value. This is the cold-start problem, and it kills more marketplace startups than bad product decisions.
The cold-start problem
Every marketplace starts with zero supply and zero demand. Demand will not show up without supply. Supply will not show up without demand. This is a chicken-and-egg problem with no clean solution.
The playbook most successful platforms use involves a sequenced approach. Pick one side as your starting point. Subsidize that side until the other side has a reason to join the platform. Uber seeded supply by recruiting drivers before riders showed up in force. Airbnb manually photographed early listings to make supply look attractive before demand reached scale. DoorDash focused on one city block before its expansion.
In an interview, you should name the side you will seed first and explain the choice. "I would seed the supply side because buyers need to see options before they commit to a new platform" is a defensible answer. "I would grow both sides at once" is not a defensible answer.
Liquidity is the metric that matters most
Growth metrics like total registered users mislead early-stage platform teams. A marketplace with 10,000 registered buyers and 50 active sellers has a liquidity problem. Liquidity means a buyer who shows up has a reasonable chance of completing a transaction.
Andrew Chen at a16z describes liquidity as the single most important health metric for any marketplace (Chen, Andrew. "The Supply-Side of Marketplaces." a16z. Web.). Track it as transaction success rate, match rate, or time to first successful transaction depending on your vertical.
When you propose a marketplace in a PM interview, name your liquidity metric. Interviewers who have built or studied platforms will test you on that point.
Pricing and take rate
A platform earns revenue by taking a cut of transactions. That cut is the take rate. Uber takes roughly 20 to 30 percent. Etsy takes around 6.5 percent. The difference reflects market power, competition, and the cost of trust infrastructure on each platform.
Setting take rate is a product decision as much as a finance decision. A rate that is too high pushes transactions off the platform. A rate that is too low leaves revenue on the table and can signal low quality to buyers. Airbnb discovered that hosts who tried to move transactions off-platform to avoid fees created bad guest experiences because those transactions lost Airbnb's trust layer.
In a marketplace design question, you should address take rate in explicit terms. State a range, explain your logic, and acknowledge the disintermediation risk.
Trust and safety as product infrastructure
The reason buyers use a platform instead of a direct negotiation is trust. Platforms earn that trust through reviews, identity verification, payment escrow, and dispute resolution. These are the product.
Sangeet Paul Choudary frames trust infrastructure as a core layer of platform design, not an add-on (Choudary, Sangeet Paul. Platform Scale. Platform Thinking Labs, 2015. Print.). Without it, transactions leak off the platform. With it, the platform becomes the default way to transact in a category.
When you design a marketplace in an interview, give an explicit description of your trust layer. What happens after a failed transaction? Who has recourse? What is the path from fraud detection to resolution?
Curation versus openness
Every marketplace faces a spectrum between full curation and full openness. App Store curates every app. Craigslist applies almost no curation. Neither extreme fits every market.
Curation raises quality floor and builds buyer confidence. It also creates supply-side friction and slows growth. Openness drives supply growth but, if quality signals are weak, degrades buyer trust over time.
Most successful platforms start open to solve the cold-start problem, then layer in curation mechanisms as they reach scale. Etsy started as open as eBay. It moved toward curation as handmade authenticity became a brand promise. That shift created seller backlash but protected buyer trust.
Your interview answer should specify where on this spectrum you land and give your rationale. Platform design choices that seem neutral are always tradeoffs.
Multi-homing and competitive moats
Multi-homing means simultaneous activity on more than one platform. A driver who works Uber, Lyft, and DoorDash at the same time is an example of multi-homing. This is a competitive vulnerability because your supply lacks exclusivity.
Marketplaces build moats against multi-homing through switching costs, proprietary data, and network density. A driver who has 2,000 five-star Uber ratings will not start from zero on a new platform. A buyer who has a decade of Amazon purchase history and a Prime membership does not want to rebuild that on a competitor.
Simon Rothman, a partner at Greylock, argues that the strongest marketplace moats come from liquidity density in a specific geography or category rather than broad user counts (Rothman, Simon. "Marketplace Liquidity." Greylock Perspectives. Web.). Think about what part of your supply or demand genuinely resists replication.
In an interview, name your moat with specificity. "Network effects" is not an answer. "Buyers on our platform generate proprietary quality signals that new entrants cannot replicate" is an answer.
Governance and rule-setting
Beyond the technology layer, a platform is a rule-setter for a mini-economy. Who sets prices? Who can become a participant? What behavior earns a ban? Who resolves disputes? These governance questions matter because the rules determine who wins and loses on the platform.
Platforms that set rules to favor one side create long-term imbalances. Amazon's private-label strategy put it in competition with third-party sellers who depended on the same platform. That governance decision created regulatory scrutiny and seller distrust.
When you design a marketplace, describe the governance layer. What rules protect both sides? What happens when the platform's interests collide with a participant's interests?
Putting it together in an interview
A strong marketplace design answer covers six areas. First, identify both sides and what each side wants from the platform. Second, name which side you seed first and your method. Third, define your liquidity metric and your tracking approach. Fourth, state your take rate and defend the range. Fifth, describe your trust infrastructure. Sixth, name your moat.
Most candidates stop after two or three of these areas. If you cover all six with specific reasoning, you are in the top tier of responses for that interviewer.
Platform design is a systems design problem with economic constraints, not a UX problem. Treat it that way.
Works cited
Chen, Andrew. "The Supply-Side of Marketplaces." a16z. Andreessen Horowitz. Web.
Choudary, Sangeet Paul. Platform Scale. Platform Thinking Labs, 2015. Print.
Rothman, Simon. "Marketplace Liquidity." Greylock Perspectives. Greylock Partners. Web.