Expanding a Marketplace to New Cities and Categories

By Brian Tr · Updated Oct 7, 2026

A marketplace that works in one place, for one kind of thing, has solved the hardest problem it will ever face. It has gotten both sides to show up and transact. The next question sounds easier and often isn't: where does it grow from here?

There are two main directions. A marketplace can go wider, into new cities or countries with the same product. Or it can go deeper and broader in what it offers, adding new categories of goods or services to the same audience. Each axis has its own logic, its own failure modes, and its own readiness tests. This article lays them out, explains why the same move that works for one marketplace sinks another, and gives a short set of questions for judging whether a new market will behave as a local one or a global one.

The two expansion axes

The axes aren't exclusive. Most large marketplaces have done both, in some order. But the order matters, and so does what's carried over. A geographic launch carries over the product and the playbook, and has to rebuild the participants. A category launch can carry over one side (the buyers, or the sellers) and has to build the other.

Geographic expansion

Local versus global: the question to ask first

Before planning any launch, decide whether the marketplace is local or global in the sense of where its network effects operate. The a16z marketplace glossary (Li Jin, February 2020) draws the line this way: in a global marketplace, additional supply anywhere in the world creates value for a user in a different country, while in a local one, value only counts within a specific geography.

NFX's marketplace expansion framework (May 2021) gives the practical consequences. Global marketplaces reward speed, because competitors can capture valuable supply if you don't scale quickly enough. Local marketplaces only gain value as more users in a specific neighborhood or city join, which makes them hard to start, but potentially defensible. Its own example: adding drivers in San Francisco only adds value to riders in San Francisco. Riders in New York get nothing from it.

The same essay adds an important caveat. It's rarely black and white. Many marketplaces have both kinds of effect, such as a ride-hailing customer who uses the service in their hometown most of the time and in other cities when they travel.

A quick way to test any marketplace:

1. Does a new supplier in another place help a buyer here? If yes, the network effect is wide. If no, it's local.
2. Does the transaction require physical presence in one place? Services delivered in person, goods delivered the same day, and anything with a travel-time limit are local by nature.
3. Can a competitor win one place without winning the others? If a rival can take a single city and hold it, you're facing a series of local fights.
4. Does demand cross borders on its own? If buyers search for supply far from home (travel, digital goods, specialist items), the market is wider than any one city.

The city-by-city model

For local marketplaces, the classic approach is a repeatable launch. Bill Gurley's 2012 essay All Markets Are Not Created Equal (November 13, 2012) notes that businesses like Yelp, Uber, and GrubHub depend on "city launches" with established processes, and that OpenTable's expansion was slowed by high-friction supplier signup that required physical installation in each location. The point isn't that these are bad businesses. It's that each launch consumes real time and money, so the process has to be cheap enough to repeat.

Uber's annual report describes the operating side of this. In its FY2025 10-K, the company says its regional on-the-ground operations teams use market-specific knowledge to launch and scale products in cities, support drivers and consumers, and build relationships with cities and regulators. The same filing says the network operates in more than 15,000 cities as of December 31, 2025. The lesson for a smaller marketplace isn't to copy the headcount. It's that a launch has local work in it (regulation, partners, supply recruiting) that software alone doesn't do.

A city-launch playbook usually bundles the same pieces:

• A supply recruiting routine that was proven in the home market
• A minimum viable level of supply before any demand marketing starts
• A defined local owner, even if that's one person part-time
• Rules for regulatory and partner contact
• A set of metrics that tell you when to stop spending on that city

The alternative: don't copy the playbook

NFX makes a point worth keeping in mind. Most marketplaces overfit their growth strategy to existing playbooks. It cites DoorDash as a case where the go-to-market differed from the Uber model. According to the essay, DoorDash didn't follow city-by-city expansion with local general managers. It started in the suburbs, which was a contrarian choice at the time, and the essay says those markets later helped fund expansion into more competitive urban ones. Treat that as NFX's account of one company's path, not a rule. What it shows is that a launch pattern should come from the market's structure, not from whichever famous company is most often cited.

Readiness signals in the home market

Expansion works best when the first market is complete. Signs that it is:

• Liquidity holds without subsidies. Matches happen at a healthy rate because the market is useful, not because you're paying for them. See marketplace liquidity.
• Density is established where you operate. Buyers find nearby supply and suppliers find nearby demand. See marketplace density.
• Repeat usage is visible. Users come back without being re-acquired each time.
• The launch process is written down. If it only lives in the heads of the people who did the first city, you're not ready to do a second one.
• The home market can run without founder attention. Otherwise the second city competes with the first for the same people.

There are no universal numeric thresholds for any of these, and none are claimed here. The right bar depends on the category. What matters is that the team has decided on its own bar before it expands, not afterward.

Why expanding before liquidity hurts

NFX warns that, for local marketplaces, expanding to new geographies too early can leave you vulnerable. The company typically has limited supply or demand-side relationships to lean on in the new place, which makes expansion expensive and can reduce liquidity in the core offering.

Here's the mechanism. A new city draws on the same finite pool of operators, engineers, and marketing budget as the home market. If the home market isn't yet self-sustaining, every dollar and hour sent elsewhere weakens it. Buyers who get fewer matches leave, and suppliers who see fewer orders leave, and the loop that should be pulling the marketplace upward starts running backward. A second weak market adds a second place where the same thing can happen.

The same essay also describes the opposite error. Early on, a local marketplace still needs to expand to enough similar markets to reach sufficient coverage. And a marketplace with wide network effects that doesn't expand quickly can lose to larger-scale operators. So the guidance isn't "go slow." It's "be ready, then be fast," and the definition of ready depends on whether your network effects are local or global.

Sequencing the cities

When there's a choice of where to go next, these criteria help rank candidates:

1. Similarity to the home market. Same regulations, same supply types, same buyer behavior. Similar markets let you reuse the playbook.
2. Existing pull. Buyers or suppliers already asking, or already using the product while traveling.
3. Competitive intensity. An uncontested market is cheaper than one with a strong local incumbent.
4. Supply availability. For supply-constrained marketplaces, how fast can you recruit?
5. Strategic value. Some places matter because they unlock other things, such as travel hubs for a travel marketplace.

NFX describes one version of this for local marketplaces: win the constrained side of the market first, aiming for the number one or two position, then replicate that quickly elsewhere. For more on structuring a market entry decision outside marketplaces, see the market expansion model and our piece on geographic expansion for SaaS.

Category expansion

Why add categories

Category expansion makes sense when the marketplace has a base of participants that wants more than one thing from it. Typically, one side is already acquired and the new category gives them more reasons to transact, more often. NFX frames this sequence for demand-constrained marketplaces: capture more demand with the same type of supply across more geographies first, then move to adjacent categories once you dominate.

Two shapes are common:

• Shared demand. The same buyers purchase from a new set of suppliers. The buyer relationship is the asset being reused.
• Shared supply. The same suppliers serve a new set of buyers or sell a new type of service. NFX notes that once you've embedded with suppliers, it's easier to add an adjacent offering that connects them to demand-side users.

Picking which side to reuse is a real decision. It's the same question as supply-first versus demand-first, asked again at a larger scale.

The horizontal and vertical tension

Category expansion pushes a vertical marketplace toward being horizontal. The a16z glossary defines a vertical marketplace as one hyper-targeted to the needs of a particular industry, product category, or group of customers, and a horizontal one as broadly targeted, such as Craigslist. Every added category makes the product more general, and a more general product usually offers a less tailored experience.

So the trade is concrete. A marketplace might gain reach and cross-selling and lose depth, trust, and the specialized features that made its first category work. Our article on horizontal versus vertical marketplaces covers that tension in full. For expansion planning, the practical rule is to ask whether the new category needs different supply quality, different search, or different trust mechanics. If it does, it's closer to a new marketplace than to a new shelf.

Same-side and cross-side effects

The a16z glossary defines two-sided network effects as the network becoming more valuable as the number of users on the other side increases. That's the cross-side effect, and it's what most marketplace thinking assumes. A same-side effect is the other kind: value rises because more users of the same type join. Buyers might benefit from other buyers through reviews or price comparison, for instance. A new category usually brings cross-side effects only if it shares one side with the core. If it shares neither side, none of the core's network effects carry over, and the category starts from zero.

That gives a simple screen for category ideas:

If the answer to all four is no, the category is a separate business with a shared logo.

Putting the axes in order

Neither axis is automatically first. A reasonable order of thinking:

1. Confirm the core is liquid. Neither axis should start until the home market works on its own.
2. Identify the network-effect type. Local marketplaces usually deepen geography before breadth. Global ones usually widen supply fast.
3. Pick the axis that reuses the most. Choose the move that carries over the most supply, demand, and trust.
4. Run one expansion at a time. Two at once split attention and hide what's working.
5. Define the stop rule up front. Decide in advance what a launch has to show by a given time, or the team will argue about it afterward.

For how the expansion loop connects to the wider logic of platforms, see network effects.

Key Facts: Marketplace Expansion

• Two axes: geographic (new cities or countries) and category (new goods or services). They carry over different things: geography reuses the product, categories reuse one side of the market.
• The a16z marketplace glossary (Li Jin, February 2020) defines global marketplaces as those where supply anywhere adds value for users elsewhere, and local ones as relevant only within a geography.
• Gurley's 2012 essay names Yelp, Uber, and GrubHub as businesses that depend on city launches.
• NFX's expansion framework (May 2021) says expanding to new geographies too early can make a local marketplace's expansion expensive and reduce liquidity in the core offering.
• The same essay says global-network marketplaces that don't expand quickly risk losing to larger-scale operators.
• Uber's FY2025 10-K says it operates in more than 15,000 cities as of December 31, 2025, using regional on-the-ground teams to launch and scale products.
• None of these sources gives a numeric liquidity or density threshold for "ready to expand." The bar has to come from your own category data.

Related Reading

• Why Density Beats Reach in Local Marketplaces
• What Is Marketplace Liquidity?
• Horizontal vs Vertical Marketplaces
• Marketplace Metrics: GMV, Liquidity and Take Rate
• Market Expansion Model