Acquiring Buyers for a Marketplace

By Brian Tr · Updated Oct 6, 2026

Buyer acquisition is the work of bringing people who want to purchase, book, or hire onto a marketplace that already has something worth buying. The order of those words matters. Most marketplaces that work did not start by buying traffic. They started by getting supply good enough that demand had something to land on, and then they used a handful of channels to bring buyers in.

This article is a reference list of those channels, not a prescription. For each one we cover how it works, when it fits, how it scales, and what can go wrong. It's the demand-side mirror of acquiring supply for a marketplace. Whether to chase sellers or buyers first is a separate argument, and it's covered in supply-first vs demand-first. Here we assume you've made that call and need to know how buyers actually arrive.

Several of the channels below (search engine optimization, referral loops, density, and single-player mode) are big enough to deserve their own deep dives, so we keep them short. And if you want the wider theory of why two-sided businesses are hard to start, the chicken-and-egg problem and two-sided markets cover it.

Why demand usually follows supply

A buyer arrives with a question: can I find what I need here, at a fair price, soon? If the answer is no, the visit is wasted, and paying to generate that visit makes it worse. Traffic isn't the goal. A buyer who finds a good match and transacts is the goal, and the word for how often that happens is liquidity.

Liquidity is, loosely, the chance that a buyer looking for something finds it, and the chance that a seller listing something sells it. You can have a million visits and low liquidity, or ten thousand visits and high liquidity. Only the second one is a healthy marketplace.

That's why demand acquisition tends to come second. Lenny Rachitsky interviewed founders and early employees from many of the biggest marketplaces, including Airbnb, DoorDash, Etsy, Uber, and Thumbtack, and wrote up what drove their early demand. In his Part 4, published December 3, 2019, he notes that companies like OpenTable, Lyft, Uber, and DoorDash spent very little of their early resources on driving demand, while others, such as Zillow, Rover, and TaskRabbit, shifted to demand quickly.

He also gives three signals that it's time to shift attention to buyers:

• Supply is growing easily, with people signing up without much convincing.
• Supply is heavily underutilized, meaning plenty of availability and few bookings.
• You haven't yet shown you solve a problem for people.

The third signal is easy to miss. If you can't show that your offer is cheaper, more convenient, or simply better than the alternatives, more buyers won't fix it. Check that first, and then spend on demand.

Demand takes more levers than supply

One more finding from the same study is worth keeping in mind. Lenny reports that the median (and average) number of levers a company bet on for supply was roughly two, while for demand it was three. His reading is that companies see impact from more channels when driving demand. The supply-side figure is covered in our supply acquisition article.

So the advice to pick one or two channels and do them well still holds, but expect your mix to be a little broader on the buyer side, and expect it to change as you grow.

The channels at a glance

The rest of the article takes these one at a time. The percentages and examples come from Lenny's study unless we say otherwise, and we've tried to quote only what his page actually reports.

Channel 1: Supply brings demand

This is the quiet one, and it surprised Lenny. He reports that supply directly driving demand was effective for over 40% of the marketplaces he studied, which is part of why these companies focus almost exclusively on supply growth.

How it works. Sellers already have a way to reach buyers: customers, followers, a storefront, a mailing list. They tell those people to use your platform because it's now where their business happens.

The examples are concrete. A DoorDash early employee says restaurants printed placards and stickers and put them up on their own, often without DoorDash knowing, and that stickers on windows worked extremely well. GrubHub's Casey Winters describes in-store signage, Open/Closed signs, sandwich boards, printed menus, business cards, and delivery bags. For Etsy, sellers did their own grassroots marketing, and the company encouraged them to promote their shops to their communities. Patreon's model, as described by a former employee, is to acquire creators who already have followings, and those creators then bring their patrons.

When it fits. Sellers have an audience, or a place where buyers physically pass by. Creators, restaurants, event organizers, and local retailers fit this well.

Cost and scaling. Close to free in cash terms, and it scales with the number and quality of sellers. That's a real advantage. But the number of buyers it delivers depends on each seller's reach.

What goes wrong. Lenny's page lists cases where it didn't work. TaskRabbit's tasks happened behind closed doors, so there was no visible equivalent of Lyft's mustache. A Rover contact says seeing Rover t-shirts in the world wasn't a growth driver. And there's a structural risk: a seller who brings you their own customer has little reason to keep paying you a commission for that customer on the second order. That tension, often called leakage, gets worse when the thing sold is a repeat service. It's a design question for your fee model, not a channel problem, but it shows up first here.

Channel 2: SEO and inventory pages

Search engines send buyers who are already looking. A marketplace has an unusual advantage here, because its inventory is a large set of unique pages that match specific searches.

Lenny's study found that SEO, which mattered little on the supply side, was a vital early demand driver for over 40% of marketplaces. Some of his quotes are striking. A Thumbtack founder says SEO ended up being 80 to 90 percent of its growth. GrubHub's Casey Winters says SEO was 30% of all new users and its number one lever, with 10 to 15 percent from paid search, and describes building a landing page for every restaurant. Etsy's Nickey Skarstad says millions of indexed product listings made search traffic real even early on. Caviar's Gokul Rajaram, though, calls SEO low impact for them.

Public filings add a warning. Etsy's FY2025 annual report says the company obtains a significant number of visits via search engines such as Google, and that algorithm changes have negatively affected, and may continue to affect, how its marketplaces are placed in results. Airbnb's FY2025 filing makes a related point: rising keyword costs and algorithm changes can reduce visibility and raise customer acquisition costs.

When it fits. Buyers type specific queries, such as "Thai food near me" or a type of service in a city, and you have many distinct listings. What goes wrong. Thin, near-duplicate pages, and reliance on one search engine's goodwill.

Channel 3: Paid acquisition

Search ads, social ads, and display aimed at buyers. It's the easiest channel to switch on and the easiest to misuse.

Lenny's evidence is mixed in a useful way. Performance marketing was instrumental for about a third of the marketplaces he studied. Airbnb's Gustaf Alströmer says buying keywords related to staying in homes and apartments was a way of being strategic about where to grow. Rover's David Rosenthal says most of its customer acquisition was through Google Ads aimed at people searching for dog boarding, and that a later spending fight with DogVacay partly led to their merger. GrubHub started buying keywords for a cuisine only once it had four or more restaurants for it, which is a good example of tying spend to supply.

But the page also lists paid growth under what didn't work early. Thumbtack says it lacked the cash or revenue model to support it until many years in. TaskRabbit says paid ads became more important about five years after launch. Instacart says it did almost no paid marketing early on.

Public filings show how big the spend gets later. Etsy tracks "Paid GMS," the part of marketplace sales attributable to performance marketing, and reports it was 24% of overall GMS for 2025, up from 21% for 2024. The same filing puts total marketing expense at $914.8 million for 2025. So paid isn't the whole picture even at scale, and it is a growing part of it.

When it fits. Each transaction earns enough margin to pay back the click, supply for the searched category already exists, and you can measure cost per completed order, not per visit. The arithmetic is the same as in customer acquisition cost. What goes wrong. Buying visits into an empty or thin market. The ad worked, the buyer arrived, nothing matched, and you paid for a bad first impression.

Channel 4: Referrals

Buyers invite other buyers, usually with a credit on both sides. Lenny reports an early referral program was important for three companies: Instacart, Airbnb, and Uber. Uber's Andrew Chen says about 30% of trips came through referrals early on, and Instacart's Max Mullen says about a third of demand did.

It isn't universal. TaskRabbit started with a give-$10, get-$10 model in 2010, but its contact says it wasn't significant because customers shared organically and didn't rely on the program. Referrals amplify talk that already exists. They rarely create it from nothing.

This channel has its own article, so we'll leave it there. The short version: it needs a product people already want to mention, and a reward that's paid after the first completed purchase, not at signup.

Channel 5: Partnerships and embeds

Partnerships put your inventory where buyers already are: directories, other brands' sites, or the sellers' own websites.

OpenTable is the clearest example on Lenny's page. Its early employee says the company partnered with Citysearch, AOL Digital Cities, Yahoo, and Metromix, so that anywhere one of its restaurants was listed in an online directory, an OpenTable link appeared. He adds that despite the breadth of these deals, partnerships accounted for only about 15% of bookings, and the rest came from OpenTable.com directly or restaurants' own websites. That second part is itself a form of embed: OpenTable helped restaurants build sites with a booking module, so a diner clicking to reserve landed on OpenTable.

TaskRabbit took a different route, working with established brands such as Pepsi, Gap, and Sephora, mostly through in-person activations, which it says raised brand awareness and credibility.

When it fits. Buyers' intent is already expressed somewhere else, and a link or widget can capture it. Cost and scaling. Cheap per buyer once built, slow to set up. What goes wrong. It may deliver a small share of demand, as OpenTable's 15% shows, and the partner controls the relationship.

Channel 6: Content and community

Some marketplaces draw buyers by publishing something useful. Thumbtack's founder describes surveying tens of thousands of businesses on how business-friendly their local government was and ranking every state and city. It took one person about seven months, and it began as an SEO play and turned into a branding event. Zillow made public housing data easier to find and share.

Community is the same idea with people instead of articles. Airbnb's Brian Chesky, quoted on Lenny's page, says events and PR were the main way the network was bootstrapped in the early days. Lyft's launch parties are another form: Lyft's Benjamin Lauzier says they reached about a thousand people at most, but created a core group of passionate users who became ambassadors.

When it fits. Buyers research before they buy, and you hold data nobody else has. What goes wrong. It's slow, and attribution is murky. A report that wins links but attracts readers who never buy hasn't added liquidity.

Channel 7: PR and launch

PR was impactful for about a third of Lenny's companies, and the primary lever for TaskRabbit and Zillow. TaskRabbit's Jamie Viggiano says that in 2009 to 2011 the sharing economy was new, customers hesitated to invite strangers into their homes, and the best way to build trust was word of mouth and press. Zillow's Nate Moch says the company's controversial product and its housing data got coverage, helped by the timing of the Great Recession, when home-value data was newsworthy.

Uber's version is a stunt format. Andrew Chen describes the idea of a "rider zero," a local celebrity as the first rider in a new market.

When it fits. The product is new or contested, trust is the barrier, or a news cycle makes your data relevant. What goes wrong. OpenTable shied away from PR because fine dining is planned and infrequent, and the company didn't believe press could drive meaningful demand. Attention doesn't help if the audience isn't about to buy. It's also spiky and hard to repeat.

Channel 8: Offline and local

When a marketplace works city by city, some demand is cheaper to win in person. Lyft's Lauzier says the team went door to door to startups with free cupcakes, donuts, and ride coupons. Uber used street teams who handed out referral codes at places like the Caltrain station, which later became an Ambassadors program. Rover had people hand out flyers and coupon codes at dog parks.

When it fits. The market is local and density matters more than reach. Cost and scaling. High cost per buyer, and it scales one city at a time. What goes wrong. Success in one city doesn't carry over automatically, and the work depends on local people. How concentrated buyers and sellers must be in each area is the subject of its own article on density.

Matching channels to your marketplace

A few rules fall out of the above.

1. Start from the buyer's moment of need. If buyers search, SEO and paid matter. If trust is the barrier, PR and partnerships matter. If the market is physical, offline matters.
2. Check that supply exists for the demand you're buying. Spend follows inventory, as GrubHub's keyword rule shows.
3. Measure completed transactions, not visits. Visits are the number that flatters every channel.
4. Plan for three or so levers, not one. And expect the mix to change as you scale.
5. Watch for leakage. Buyers who arrive through a seller's own channel may take the next order off-platform unless you give them a reason not to.

Buyer type changes the picture too. A business buyer is reached differently than a consumer, and B2B vs B2C marketplaces covers that contrast. In a managed marketplace, the operator often finds buyers directly and matches by hand, so the channel list shifts toward direct sales. And once buyers arrive, supply has to be ready to serve them, which is the job of seller onboarding.

Key Facts: Marketplace Buyer Acquisition

• In Lenny Rachitsky's marketplace study, supply driving demand and SEO were each effective for over 40% of companies, and performance marketing and PR for about a third (Part 4, published December 3, 2019).
• The same study found the median company relied on about three levers for demand versus roughly two for supply.
• Early referrals accounted for about 30% of Uber trips and about a third of Instacart's demand, according to people quoted on that page.
• OpenTable's directory partnerships accounted for only about 15% of its bookings, per an early employee quoted in the study.
• Etsy's Paid GMS was 24% of overall GMS in 2025, up from 21% in 2024, with total marketing expense of $914.8 million (FY2025 10-K).
• Airbnb's FY2025 10-K warns that rising keyword costs and search algorithm changes can reduce visibility and increase customer acquisition costs.

Related Reading

• Acquiring Supply for a Marketplace
• SEO for Marketplaces
• Referral Loops in Marketplaces
• Why Density Beats Reach in Local Marketplaces
• What Is Single-Player Mode?
• Supply-First vs Demand-First
• The Chicken-and-Egg Problem in Marketplaces
• B2B vs B2C Marketplaces
• Managed Marketplace
• Marketplace Seller Onboarding
• Two-Sided Markets
• Customer Acquisition Cost