Why Density Beats Reach in Local Marketplaces

By Brian Tr · Updated Oct 6, 2026

A marketplace with a million users spread across a hundred cities can feel dead in every one of them. A marketplace with ten thousand users packed into one city can feel alive. That gap is the whole subject of this article. In a local marketplace, how many people are on the platform matters less than how many are close enough to each other to transact quickly.

We'll call that closeness density. It's the reason ride-hailing and food delivery companies launch city by city, the reason their filings talk about performance in "a given geographic market," and the reason a national user count can hide a weak product. This article defines density, separates it from total size, walks through how companies constrain their launch area, and covers where the idea stops applying.

We won't re-explain what a network effect is. For that, see our reference on network effects. Here we're only interested in what changes when the network is bounded by geography.

What density means, and how it differs from size

Size is a count: users, listings, drivers, restaurants, nationwide or worldwide. Density is a ratio: how many of those are reachable by a given buyer within the distance, time, or category that buyer cares about.

Take two imaginary food delivery apps, each with 2,000 restaurants and 500 couriers. App A has them spread across twenty cities. App B has them all in one. A hungry customer in App A's city sees about a hundred restaurants and about twenty-five couriers scattered around town. A customer in App B's city sees two thousand restaurants and five hundred couriers. Same total size. Completely different product. These numbers are made up to show the logic, not drawn from any company.

The difference shows up in things buyers actually notice:

• Selection. More nearby options per search, so the buyer finds something they want.
• Wait time. More nearby suppliers means a shorter distance to the closest one.
• Reliability. If one supplier cancels or is busy, another is close by.
• Price. Suppliers who get frequent orders can afford to price lower and stay available.

And it shows up on the supplier side too. A driver or seller in a dense area spends less time idle between jobs, which is the main reason they stay. We covered the supplier side in acquiring supply for a marketplace. Density is why supply acquired in the right place sticks, and why the same supply acquired in the wrong place churns.

Local network effects versus global ones

The standard picture of a network effect is global: every new user makes the product better for every other user. A messaging app works that way. A phone number in another country is just as reachable as one next door.

Local marketplaces don't work like that. A new driver in Jakarta does nothing for a rider in Manila. The value created by a new participant stays inside a small radius. Some filings say this directly. DoorDash's annual report refers to the strength of the local network effects among the merchants, consumers, and Dashers on its platform as critical to attracting and retaining all three groups. Notice the word "local" doing the work there.

Uber's 10-K for fiscal 2025 makes the same point from the risk side. It states that its success in a given geographic market significantly depends on its ability to develop network scale and liquidity in that geographic market. Not across the company. In that market. The filing also warns of continuing challenges in lower-density suburban and rural areas, where its network is smaller and less liquid.

Two consequences follow, and they run against the usual startup instinct.

1. Network effects are a collection of small networks. Each city is its own contest. Winning one doesn't win the next, though the playbook and brand may carry over.
2. Spreading thin is expensive. Every new city needs its own critical mass. Launch ten at once with the same budget and you may get ten cities below the threshold instead of one above it.

For the general cold-start framing, the chicken-and-egg problem article covers why a marketplace is hard to start. Density is the answer to a specific part of that problem: where to concentrate while you solve it.

Constrain the market

Lenny Rachitsky interviewed founders and early employees from many marketplaces for his series on kickstarting and scaling them. The first part of that series, published in November 2019, has a section called "Constrain the marketplace." The quotes there are a useful catalog of how companies narrowed their starting area. A few, attributed as the series attributes them:

• Rover (David Rosenthal): the company started in Seattle and stayed hyper-focused on Seattle for a while.
• Airbnb (Kati Schmidt): the team focused on creating critical mass in just a few markets where it could quickly unlock both supply and demand.
• Uber (Andrew Chen): the goal was to get over 30 drivers, shooting for an ETA of less than 15 minutes.
• OpenTable (Mike Xenakis): if the company could get to 50 to 100 concentrated restaurants in a city, that was enough for a consumer to land on the site and find something.
• Breather (Julien Smith): areas of density like the Flatiron District saw added units get booked immediately.

Read those as a pattern, not as universal thresholds. Each number is one company's early target as remembered by someone who was there. OpenTable's range worked for restaurant booking in a city. It says nothing about your category.

The same section also lists constraints by category rather than place. TaskRabbit focused on its most popular task categories, Eventbrite on tech mixers and conferences, and Etsy launched with only three categories. And there's an exception worth keeping in mind: Thumbtack's Sander Daniels says the company did all categories and all geographies from the beginning and wouldn't have survived otherwise. That's one company's account of one situation, but it's a reminder that "constrain the market" is a strong default, not a law.

Geography is the most common constraint for a reason. It's easy to define, easy to explain to suppliers, and it lines up with how buyers behave. People order dinner near where they are.

The atomic network

Andrew Chen gave this idea a name in The Cold Start Problem. His term is the "atomic network," and an excerpt published on Lenny's Newsletter defines it as the smallest network needed that can stand on its own.

The excerpt makes three points relevant here.

First, it says that growing city by city, campus by campus, or team by team is a surprisingly powerful strategy, because it leads to dense, organic connections of users that strengthen network effects.

Second, the atomic network can be very small. For Slack, the excerpt says the threshold was just the team of people around you, and under ten in a single company might be enough.

Third, even for Uber, a company we talk about as "San Francisco" or "New York," the earliest focus was on narrow, ephemeral moments, closer to a specific corner at a specific time of day than to a whole city.

The practical lesson is that "one city" may already be too big a starting unit. If a few blocks at rush hour can be an atomic network, then the question for a founder is the smallest area and the smallest set of conditions where a buyer reliably gets what they came for. Once that works, you copy it. If you can build one, and then another, you can build the rest.

This also connects to the choice of which side to recruit first. Constraining the area makes either side easier to win, because you need fewer people. See supply-first vs demand-first for how to choose.

How density shows up in filings

Public marketplace companies rarely publish a "density score." But their filings describe the mechanism in plain language, which is a good reality check against theory.

Uber uses the word liquidity throughout its 10-K. It describes needing to build network scale and liquidity in each geographic market by attracting drivers, consumers, merchants, shippers, and carriers. It also says that in certain markets it has offered, and may keep offering, significant driver incentives and consumer discounts to remain competitive and to generate network scale and liquidity. And it notes that it has experienced driver supply constraints in most markets, and that if supply is constrained it may need to raise or may not be able to reduce incentives without hurting supply liquidity in that market. In other words, liquidity is managed market by market, and money is spent to buy it.

DoorDash describes its three-sided platform of merchants, consumers, and Dashers and ties its brand and growth to local network effects. Its 10-K also says the company continues to invest in technology that improves efficiency, including better demand prediction, forecasting of food preparation times, and optimizing routing and batching algorithms, while cautioning there is no guarantee those efforts will pay off. Batching, delivering more than one order in a trip, is easier when there are more orders close together. We're drawing that inference ourselves. The filing doesn't link batching to density in so many words.

Grab shows the other side of the same coin: the cost of doing this at scale. Its 20-F says that building the synergies in its ecosystem requires it to replicate its efforts in over 900 cities in Southeast Asia, where each country has different infrastructure, regulations, systems, and user expectations. That's what a local marketplace looks like when it's grown past its first city: the work has to be done again, locally, many times.

If you want a sense of why incentives get so large, the Grab figures in acquiring supply for a marketplace are a good companion. Incentives are one of the main tools used to push a thin market past the point where it sustains itself.

When density matters less

Density is a local-market idea, so it weakens as the marketplace becomes less local. The test is whether the product is delivered in a specific place.

Marketplaces where a buyer and seller never need to be near each other work differently. Upwork's 10-K says it has customers in over 180 countries and describes its remote model as a way around the limitations of a local-only model. For a client hiring a designer remotely, the designer's location is largely irrelevant, so what matters is the depth of talent in the relevant skill, not the number of designers per square mile. The same logic plausibly holds for any marketplace whose goods ship, though we haven't sourced a filing statement on it for Etsy, so treat that as reasoning.

For these marketplaces the equivalent of density is category depth. Does a buyer searching for a specific skill or product find enough good options? That's the same idea with a different axis. The lesson of density, which is to concentrate before you spread, still applies, but the unit you concentrate on is a niche or a category rather than a neighborhood. Our article on horizontal vs vertical marketplaces covers this trade-off from the category side.

There's also a middle zone. Marketplaces with a local delivery step, such as home services, mix both: the service is delivered locally, but the supplier pool might be recruited and vetted centrally. And a managed marketplace can sometimes substitute operator effort for density, by manually matching buyers and suppliers where the natural market is thin. That works, but it costs more per transaction.

Measuring density without inventing benchmarks

Density is a concept, so teams measure it through proxies. Here are the common ones, defined in general terms. None of these come with a universal "good" number. The right target depends on the category and has to come from your own data.

A few cautions apply to all of them.

• Measure by small area, not by company. A blended national fill rate can be healthy while three cities fail. Break the numbers down to the smallest area where the product is delivered.
• Watch both sides. High fill rate with very low supplier utilization means you may have more supply than the market can use, which is a retention risk. Low fill rate with high utilization means supply is the bottleneck.
• Look at the trend within a cohort. A new market's first months are noisy. What matters is whether the same cohort of buyers gets faster, more reliable service over time.

Uber's filing treats liquidity in much the same way: as a thing that rises and falls with driver supply and has to be maintained. If you want to compare these metrics with how you're acquiring buyers, the customer acquisition cost reference explains the cost side. In a dense market, CAC per active buyer usually looks different from a thin one, because a satisfied buyer brings repeat orders and word of mouth. That's a reason to compute CAC per market, not per company.

Practical rules of thumb

These follow from the above, and are reasoning, not data.

1. Pick the smallest area where a buyer can get what they came for. If you can't describe it in a sentence, it's probably too big.
2. Don't open a second area until the first is self-sustaining. Define "self-sustaining" in advance, with a fill rate and time-to-match you'd accept without subsidies.
3. Spend incentives where they buy density. An incentive in a market that already works is a transfer. An incentive in a market that's one step short of working can be an investment.
4. Expect to redo the work per market. Grab's description of replicating efforts across 900 cities is a warning that expansion is not free.
5. Know your unit. Neighborhood, city, campus, category, skill. The right unit depends on how the product is delivered.

Key Facts: Marketplace Density

• Uber's FY2025 10-K states that success in a given geographic market depends on developing network scale and liquidity in that geographic market, and flags lower-density suburban and rural areas as a continuing challenge.
• DoorDash's FY2025 10-K calls local network effects among merchants, consumers, and Dashers critical to attracting and retaining all three groups.
• Grab's FY2025 20-F says its ecosystem synergies require it to replicate its efforts in over 900 cities in Southeast Asia.
• In Lenny Rachitsky's 2019 series, Rover, Airbnb, and OpenTable all describe concentrating on a small number of markets first. Thumbtack is cited as an exception that launched with all categories and geographies.
• Andrew Chen defines the atomic network as the smallest network that can stand on its own, and says growing city by city, campus by campus, or team by team leads to dense, organic connections.
• Upwork's FY2025 10-K reports customers in over 180 countries, an example of a marketplace where physical density matters little.

Related Reading

• The Chicken-and-Egg Problem in Marketplaces
• Supply-First vs Demand-First
• Acquiring Supply for a Marketplace
• Horizontal vs Vertical Marketplaces
• Managed Marketplace
• Acquiring Buyers for a Marketplace
• SEO for Marketplaces