Supply-First vs Demand-First Marketplace Launch

By Brian Tr · Updated Oct 6, 2026

Every marketplace founder eventually has to answer a question that sounds simple: who do we go and get first, the sellers or the buyers? Get it wrong and you spend months acquiring people who show up, find nothing useful, and never return.

This article is a decision framework for that one choice. It sits inside the section on the marketplace chicken-and-egg problem, which frames why the problem exists. It also doesn't cover how to recruit sellers once you've decided to; that's the job of marketplace supply acquisition. Here we only decide the order, and what has to be true for each order to work.

There are four launch sequences worth naming:

1. Supply-first. Recruit sellers, providers, or listings before you push for buyers.
2. Demand-first. Gather buyers, requests, or an audience, then go and find supply to serve them.
3. Both at once in a tiny atomic network. Pick one small market and build both sides inside it at the same time.
4. Single-player, or "come for the tool." Give one side a product that's useful without the other side, and let the marketplace emerge from it.

The first two are the classic debate. The last two are what founders often reach for when neither of the first two looks workable.

Start with the evidence, not the folklore

It's tempting to say "always supply first" because the famous stories sound that way. Lenny Rachitsky's research is a better starting point. In 2019 he interviewed founders and early employees from 17 marketplaces, and in his write-up on which side to concentrate on he reports that 14 of the 17 (80 percent) focused most of their early effort on growing supply. The companies he names on that list include Eventbrite, Lyft, DoorDash, Etsy, Caviar, Instacart, and GrubHub.

The reason he gives matters more than the count. In those cases, supply either drove its own demand or word of mouth was strong enough to take care of demand. Caviar's account, as he quotes it, is that restaurants have their own brand that diners already want to order from. Eventbrite's event creators brought their own traffic.

Note the sample, though. These are 17 marketplaces that became large enough to interview, which is a survivorship filter, and the research is a set of founder recollections rather than a controlled study. Read it as "what worked for these companies", not "what always works". It's a useful prior, and the three exceptions are where the framework gets interesting.

What each sequence actually means

Supply-first

You recruit sellers before buyers exist in volume. The first buyers arrive into a catalog that already has something in it. This usually means doing things that don't scale: direct outreach, onboarding by hand, sometimes listing on the seller's behalf.

Direct sales shows up repeatedly in Rachitsky's supply-acquisition findings: he writes that direct sales was a lever for about 60 percent of the companies he talked to, and that OpenTable's early restaurant side meant hiring a sales force that knocked on doors with the software. Etsy's account, in the same piece, is that recruiting sellers in person at craft fairs worked, and that those sellers then marketed themselves. Those are tactics, which the supply acquisition article covers properly. The point here is only that "supply-first" is a commitment of founder time and cash before revenue.

Demand-first

You gather buyers, or a stream of buyer requests, and then source supply to fill them. The audience is the asset, and supply is recruited against real, visible demand. That makes the pitch to sellers easy: "there are customers waiting".

In Rachitsky's data, Rover, TaskRabbit, and Zillow were the three exceptions. His hypothesis for why is that this happens when supply is very easy to offer and flexible, with meaningful income attached, or when supply is essentially public data. Rover's account is that supply was easy and demand was hard, because customers had to change behavior and trust a stranger with their dog. TaskRabbit's account is that it had thousands of people on a waitlist to provide services and the demand side was the harder one. Zillow's supply, he notes, was publicly available data, such as county and tax assessor records, used in a new way.

The tiny atomic network

Rather than deciding which side goes first across a whole market, you shrink the market until both sides fit in your head. Andrew Chen calls the smallest self-sustaining version of a network an atomic network. For marketplaces, his guidance on the hard side of the network is a sequence: "supply, demand, supply, supply, supply". You seed the hard side, bring in the other side, and then keep returning to the hard side because it stays the bottleneck.

That's worth reading carefully. Even Chen's "both at once" version isn't symmetric. It starts with the hard side. We cover the overview in two-sided markets and marketplace business model, so we won't restate the general tactics. What matters for sequencing is that a small atomic network lets you run supply-first and demand-first as one loop in a single neighborhood, category, or customer type.

Single-player: come for the tool, stay for the network

The fourth sequence skips the standoff. You give one side a tool that's valuable even when nobody else is on the platform, and the marketplace becomes a second layer on top.

Chris Dixon described the pattern in his 2015 essay "Come for the tool, stay for the network". The tool attracts initial users, and the network delivers the lasting value. His examples are Delicious, where bookmarking was the single-player use and tagging and sharing were the network, and Instagram, where the filters brought people in and sharing made the network the reason to stay. He also notes the strategy isn't universal: Facebook and Twitter built large networks without single-player tools.

Rachitsky's research includes a marketplace version. He writes that single-player mode was less common than he expected, but when it worked, for OpenTable, Eventbrite, and Patreon, it was the most important tactic to get off the ground. His OpenTable quote is that the company couldn't lead with online bookings, because there were none yet. So it built software that restaurants could use without the demand side.

Five conditions that tilt the choice

There's no formula, but five questions do most of the work. Treat these as judgment aids drawn from the evidence above, not as laws.

1. Which side is scarcer or harder to get?

This is the single most useful question, and it matches Rachitsky's key criterion: find the constrained resource and put early effort there. If sellers are rare, specialized, or busy (restaurant owners, hosts with a spare room, skilled tradespeople), supply is the hard side and goes first. If sellers are plentiful and eager, as Rover and TaskRabbit reported, then buyers are the scarce side and demand goes first.

A quick test is the waitlist test. If you can post a "become a seller" page and get applicants without effort, supply isn't your problem. If you can't get a single applicant to a calendar slot, it is.

2. Is supply perishable or durable?

Some supply expires when it goes unused. A driver's hour, a restaurant table at 7 p.m., a hotel room tonight, and a freelancer's open week can't be stored. Other supply is durable: a listing for a used guitar sits there until someone buys it.

This is our own framing rather than a finding from the sources, but it follows from how the failure modes work. With perishable supply, an unmatched seller loses real money every hour, so a supply-first launch needs buyers to arrive fast, or you need to cover the gap. With durable supply, a listing can wait weeks without costing the seller anything, which makes supply-first much more forgiving.

3. Can supply be aggregated or seeded from public data?

Zillow is the cleanest sourced example. Rachitsky points out that its supply was publicly available records used in a new way. When you can build the catalog yourself from public or licensed data, you can launch with a full-looking supply side, and demand becomes the only thing left to win. That flips the sequence toward demand-first by default.

The same logic applies in smaller ways: a directory built from public business listings, a catalog scraped with permission, or a list of products imported from a seller's existing feed. The test is whether a buyer arriving on day one finds something real. Be careful with the claim side of this. If you list people or businesses without their consent, you may be creating a trust problem for later, so check the terms and law of your market before you do it.

4. Does demand already gather somewhere?

If buyers already congregate in one place, such as a community, a forum, an industry event, or an existing audience, you can go demand-first at low cost. The audience becomes the argument to sellers. Eventbrite's account in Rachitsky's research is the supply-side version of the same idea: event organizers arrived with their own audience, so supply carried demand.

The question to ask is which side already has a gathering place you can reach cheaply. Whichever side does is probably the side you can take first without burning cash.

5. B2B or B2C?

In a B2B marketplace, buyers are professionals with a defined need, budgets, and a longer decision process, as we cover in B2B vs B2C marketplaces. Those buyers tend to want to see vendors they recognize before they engage, which pushes toward supply-first or a managed marketplace, where the operator matches requests by hand and can start with a handful of vetted suppliers. In a B2C or C2C marketplace, individual buyers browse on impulse and bounce off thin catalogs quickly, but they're cheap to reach.

That's a tendency, not a rule, and we haven't found a sourced study that settles it either way. Treat it as a hypothesis to test in your own market. The practical implication for a B2B founder is that the first ten suppliers are often found through sales conversations, not ads.

The risks of each sequence

Both orders fail in recognizable ways. Knowing the failure mode tells you what to watch in the first 90 days.

Why supply churns

A seller who signs up and gets no orders learns the platform doesn't work. They don't wait politely. They stop logging in, and re-activating them later is harder than recruiting a new seller. The risk grows with perishability, since every unfilled hour costs them. This is why a supply-first launch usually needs a tight scope, such as one neighborhood or one category, so that a modest number of buyers can keep a modest number of sellers busy. Subsidy is one way to bridge the gap. Lyft, in Rachitsky's research, guaranteed drivers an income floor per hour to jump-start its marketplace from scratch, and Breather subsidized its supply side with furniture and locks to raise quality.

Why buyers bounce

A buyer who lands on an empty catalog, or searches and finds nothing, leaves and rarely comes back. Buyers are impatient because they have other options. The risk is highest when demand is bought with ads, because you pay for each visit and each failed visit is wasted money. The fix is rarely more traffic. It's usually a smaller promise: "plumbers in this one district", "this one product category", with enough supply behind it that most searches succeed.

Decision checklist

Run through these in order. The point isn't to score high; it's to see where your answers disagree, because that's where the risk is.

1. Which side is harder to get? If sellers are scarce, lean supply-first. If sellers apply unprompted, lean demand-first.
2. Is supply perishable? If yes, keep the launch market small enough that buyers can keep sellers busy, or plan to cover the gap yourself.
3. Can you build the catalog without sellers? If public or licensed data lets you, go demand-first and convert supply later.
4. Where do buyers gather today? If you can reach them cheaply, demand-first is open to you.
5. Is there a tool one side would use alone? If yes, consider single-player mode and treat the marketplace as phase two.
6. Can you shrink the market until both sides fit? If yes, run a tiny atomic network and start with the hard side.
7. What does a failed week look like? Write down the one metric that tells you supply is churning or buyers are bouncing, and check it weekly.
8. Is the first ten sellers a sales job? If so, budget founder time for it before you budget ad spend.

If the answers point in different directions, shrink the market first. A smaller market makes most of the conflicts disappear.

Choosing between named alternatives

Key Facts: Supply-First vs Demand-First

• In Lenny Rachitsky's 2019 research, 14 of 17 marketplaces (80 percent) focused most early effort on supply. The three exceptions were Rover, TaskRabbit, and Zillow.
• His hypothesis for the exceptions is that supply was very easy and flexible to offer, or, for Zillow, was publicly available data used in a new way.
• Rachitsky's supply-acquisition findings report direct sales as a lever for about 60 percent of the companies he interviewed, and single-player mode as the most important tactic where it worked (OpenTable, Eventbrite, Patreon).
• Andrew Chen's guidance on the hard side for marketplaces is the sequence "supply, demand, supply, supply, supply".
• Chris Dixon's 2015 essay uses Delicious and Instagram as examples of a single-player tool leading into a network, and notes that Facebook and Twitter grew without one.
• The sample is 17 successful marketplaces, so it carries survivorship bias and rests on founder recollections.

What to do next

Pick the sequence, then make it concrete. Write the launch market in one sentence (who, where, what category), name the side you'll take first and why, and name the single metric that will tell you the other side is failing. Then move to marketplace supply acquisition if supply is first, or to your audience plan if demand is.

And if you can't decide, that's information. It usually means the market is still too big. Shrinking it is cheaper than guessing, and it's the approach behind both the atomic network and the minimum viable product: test the smallest version that could work before you scale it.

Related Reading

• The Marketplace Chicken-and-Egg Problem
• Marketplace Supply Acquisition
• B2B vs B2C Marketplaces
• Managed Marketplace
• What Is a Marketplace Business?
• Two-Sided Markets
• Marketplace Business Model
• Minimum Viable Product