Two-sided marketplace acquisition: supply or demand first? The media method to build liquidity without paying twice for an empty market, with 2026 data.
You raised money to launch a marketplace. The media plan has two lines: "seller acquisition" and "buyer acquisition", same budget on both sides, launch on the same day. It is the best way to pay twice for an empty market. The first buyers show up, find nothing, leave. The first sellers list, sell nothing, deactivate. You acquired both sides. You acquired zero transactions. A marketplace does not sell a product, it sells a probability: the probability of finding what you are looking for. As long as that probability is low, every euro of media buys a disappointment.
Two-sided marketplace acquisition: supply first, almost always
Lenny Rachitsky, formerly at Airbnb, asked 17 marketplaces (Airbnb, Uber, DoorDash, Etsy, Thumbtack, Instacart, Lyft, Eventbrite and others) one question: which side did you work on first? Fourteen out of seventeen, 80%, started with supply. Sander Daniels, co-founder of Thumbtack, puts it bluntly: "all that matters is supply". At DoorDash, restaurants were the first growth lever, supply brought demand. At Eventbrite, organisers brought their own attendees. At Etsy, sellers and buyers were the same people at the start.
The three exceptions are more instructive than the rule. Rover, the pet-sitting marketplace: supply came on its own, income and flexibility, it was demand that required a change of habit. TaskRabbit never lacked supply, thousands of applicants were waiting for approval. Zillow started with public data as its initial supply to attract agents. So the rule is not "supply first". The rule is: find the hard side, and put the effort there.
Find the hard side before spending a single euro
Andrew Chen, general partner at a16z, calls it the "hard side" in The Cold Start Problem: the minority of users who create a disproportionate share of the value, and who hold disproportionate power as a result. At Uber, 20% of drivers, the "power drivers", deliver 60% of the trips. On Tinder, women swipe right on 5% of profiles, men on 45%: the scarce side is identified in one line. That side is hard to recruit, hard to keep, and it decides whether the other side stays.
For your business, the question is not "who pays?" but "who, if missing, drives the other side away?". A buyer who finds three listings when they expected thirty leaves. A seller with no demand in thirty days deactivates. Then look at usage ratios. Julia Morrongiello at Point Nine points out that an Uber driver gives on average six rides a day while a customer takes one every eight days, roughly one driver for every fifty customers. If your target ratio looks like that, you already know which side gets recruited by hand and which side gets mass media.
The KPI is liquidity, not sign-ups
The wrong metric for a marketplace at launch is the number of accounts created. The right one is liquidity, which Point Nine defines as the probability of selling what you list or finding what you are looking for. On the supply side: the share of listings that transact within thirty days. On the demand side: the share of searches that end in a transaction, the search-to-fill rate. And you measure it per area and per category, never globally. A marketplace that is liquid in Lyon and empty in Nantes is an empty marketplace in Nantes.
That changes everything in the media plan. You do not launch a country. You launch pockets of density: a city, a category, a segment. And you do not spend a single euro on demand in an area where supply is not liquid. For Deblock, we did the exact opposite of a diluted national campaign: encircling metro stations and mobile geotargeting within 100 metres, more than 30 million impressions concentrated where the target is dense. For a marketplace, the principle is the same. Concentration builds liquidity. Dilution builds churn on both sides.
The three-step launch
1. Supply: direct sales, referrals, a temporary income guarantee, in a closed area. 2. Threshold: set a target response rate or fill rate before you start, and do not open demand below it. 3. Demand: geotargeted mass media on that area only. Then start again on the next area. No "national launch". Launches, plural.
Supply media is not demand media
In Rachitsky's study, the supply levers used at launch rank as follows: direct sales for about 60% of marketplaces, referrals for a third (a third of Uber's first trips came from referrals), piggybacking on Craigslist for a third, word of mouth for a quarter (36% of supply-side awareness at Eventbrite), subsidies for a quarter (Uber guaranteed 40 dollars an hour to its first drivers). Median: two levers per company. Paid mass media barely appears. Early supply is recruited by hand, by phone, door to door, with a cheque if needed.
Demand, on the other hand, is built with media: geotargeted social, DOOH on the open area, search on the category's queries. That is where the launch budget goes, but only above the threshold. For Essity, a gamified social mechanic generated 14 million impressions at around 6 euros per lead: the cost of a contact comes from the mechanic and the creative, not from the bid. A marketplace with liquidity to show has exactly the raw material for that kind of mechanic, because the supply itself becomes the creative. A full catalogue is the best ad you can show a buyer.
And that hierarchy still holds at scale. Vinted published its 2025 results in April 2026: 1.1 billion euros in revenue (+38%), 10.8 billion euros in gross merchandise value (+47%), and net profit down 19% to 62 million, because the platform is reinvesting in the market itself: logistics, payments, new categories (sports, collectibles), three new countries. That is not acquisition media. That is investment in liquidity. A marketplace with ten billion in GMV still thinks that way.
The three mistakes I see most often
- The 50/50 budget. Symmetrical on paper, wrong in real life. The hard side takes 70 to 80% of the launch effort, often in people rather than media.
- The national launch. A million sign-ups spread across 35,000 towns is zero liquidity everywhere. Three liquid cities are worth more than an empty country.
- Blended CAC as a compass. On a marketplace, a single CAC hides the fact that you are overpaying for one side and underpaying for the other. You need a CAC per side and per area, and above all a cost per transaction, the only number that tells you whether the two sides actually met.
This is what we are working on right now for a video game client, a competitive gaming platform: recruiting players without the first arrivals finding an empty room. Same logic, same sequence, density before reach. And when an ad platform rejects the category, the plan holds, because it never rested on a single platform.
What I tell marketplace founders
Your media plan is not a media plan. It is a liquidity plan with media in it. Find your hard side. Recruit it by hand in a closed area. Set a numeric liquidity threshold, and only open demand above it. Measure cost per transaction, per area. Then replicate. It is not slower than a national launch. It is just the only one that gets anywhere.
If you are launching or relaunching a marketplace, bring your supply/demand ratios to our free audit: together we look at which side is hard, which area to open first, and what you are spending today to fill one side the other never sees.
