Marketplaces fail on the side that is harder to recruit, and founders usually assume it is the other one. The test is cheap: twenty conversations and five transactions completed by hand, before any software exists.
This guide covers what those conversations should ask, what the manual transactions teach about trust and disputes, and why deciding which side to subsidise is the decision everything else depends on.
Key takeaways
- One side is harder than you think: Founders usually assume it is the other one, and twenty conversations settle it before any build.
- Match five transactions by hand: Trust, pricing and disputes are all learned in two weeks at the cost of your own time.
- Decide which side you subsidise: Growing both evenly is how a marketplace spends a year too thin to be useful to either.
- Repeat transactions, not visits: That figure is what separates a marketplace from a directory people use once.
Look for the business signal behind the technology
The signal that a marketplace model fits is a transaction that already happens awkwardly. Buyers and sellers who are finding each other through directories, message groups or personal contacts are telling you there is demand and that the current route is poor.
A marketplace is not always the right shape for that demand. If one side is very small, or if the business can hold the inventory itself, a straightforward shop is simpler, cheaper and easier to control, and it does not require solving the chicken and egg problem.
Before anything is built, talk to twenty potential sellers and ask what would make them list with you rather than somewhere they already trust. Their hesitation is the product requirement, and it rarely turns out to be the interface.
Use four lenses to understand the opportunity
Supply
Find out who the sellers are, what they use today, and what would make them list with you rather than somewhere they already trust. Twenty conversations before any build will tell you whether the harder side can be recruited at all.
Demand
Describe who the buyers are, what they search for, and whether they would return without paid acquisition. A marketplace that only works while advertising is running is a channel rather than a platform, and the economics differ completely.
Trust
Decide what a buyer risks on a bad transaction and who absorbs that loss while the platform is small. Trust is bought with your own money in the first year, and the budget for it belongs in the plan rather than in the first dispute.
Operations
Work out who handles a dispute at midnight in year one and what authority they have to settle it. Marketplace operations are the cost that scales with transactions, and the commission has to cover it before the model works at all.
Assess the current state without buying software first
You can test most of this without software. Match a handful of buyers and sellers by hand, complete the transactions manually and record what each one cost you in time. That number is the operational cost the commission has to cover.
Decide what useful measurement looks like
The measure that separates a marketplace from a directory is repeat transactions per buyer. One-off usage means people found what they needed and left, which is a useful service but not a platform with defensible economics.
Track the two sides separately. A healthy total can hide a seller base that is churning as fast as it is recruited, and that pattern only becomes visible when each side is counted on its own.
When software is likely to help
Build when the manual version is working and the operational load is the thing stopping it from growing. At that point software removes a bottleneck you have already proved exists, and the requirements are written by a year of doing it by hand.
Building first, in the hope that supply and demand will arrive, is the common and expensive mistake. A platform with no participants is indistinguishable from a platform with a bad product, and it is impossible to tell which one you have.
For a narrower commerce model worth comparing before committing to a marketplace, see When Does a Loyalty App Make Sense for Your Business?.
Start with a bounded improvement
Pick one category, one city or one customer segment and make the transaction work there before widening. A marketplace that is dense in a small area is useful; the same number of participants spread across a country is a directory nobody returns to.
Separate a structural constraint from a temporary problem
Test whether the demand is structural before committing. Does the awkward transaction still happen in a quiet month, outside the seasonal peak, and when the one enthusiastic seller is unavailable? A pattern that depends on one participant is a relationship rather than a market.
Lay it out in a grid: high and low demand on one axis, plentiful and scarce supply on the other. Check how supply, demand, trust and operations behave in each square. The square where supply is scarce and demand is high is the one that will decide the business.
If the opportunity is real, write it as a claim that could fail: this many sellers will list within a month, and this share of buyers will return. Then run the manual version and see. A marketplace justified by market size rather than by a claim is justified by nothing.
Risks to make visible
| Lens | Common risk | Early response |
|---|---|---|
| Supply | Sellers will not leave what they trust | Ask twenty of them what would change that, before building anything. |
| Demand | Buyers only arrive with advertising | Measure repeat transactions per buyer rather than total visits. |
| Trust | A bad transaction lands on the buyer | Decide who absorbs the loss while the platform is small, and budget for it. |
| Operations | Operations cost more than the commission | Run ten transactions by hand and measure the time each one took. |
A 30-day validation plan: Online marketplace business model
Days 1 to 5: approach twenty sellers with nothing built. Describe the proposition and ask twenty potential sellers whether they would list, what they use today, and what would have to be true. Their hesitations are the product requirements. A marketplace that cannot recruit supply on a conversation will not recruit it with a better interface either. Ask them what commission they would accept, because the number they say out loud is usually lower than the one the business plan assumes.
Days 6 to 15: match five transactions by hand. Find buyers for a handful of those sellers yourself, over the phone if necessary, and complete the transactions manually. You learn the trust problem, the pricing problem and the dispute problem in two weeks, at the cost of your own time rather than a platform. Note how much of your own time each transaction took, since that is the operational cost the platform will later have to absorb or automate.
Days 16 to 30: decide which side you will pay for. Look at which side was harder to recruit and commit to subsidising it. Write down what that costs per participant and how long you can sustain it. A marketplace that tries to grow both sides evenly runs out of money while still being too thin to be useful to either. Set a limit on how long you will subsidise before the other side has to stand on its own, as an open-ended subsidy is simply a slow way to run out of money.
A practical worksheet: Online marketplace business model
Marketplaces fail on the side that is harder to recruit, and founders usually assume it is the other one. Fill these rows before writing a line of code, and be specific about who you can reach in the first month.
| Decision area | What to record |
|---|---|
| Supply | Who the sellers are, what they use today, and what would make them list here instead of somewhere they already trust |
| Demand | Who the buyers are, what they search for, and whether they would come back without paid acquisition |
| Trust | What a buyer risks on a bad transaction, and who absorbs that loss while the platform is small |
| Operations | Who handles a dispute at midnight in year one, and what authority they have to settle it |
| Measurement | Repeat transactions per buyer, which separates a marketplace from a directory people visit once |
Decide which side you will subsidise and say so out loud. Trying to grow both sides evenly with the same budget is the most common way a marketplace spends a year being too thin on both to be useful to anyone.
Related guides
- The same trade-off from the other side appears in E-Commerce App Development: A Practical Planning Guide.
- If the budget conversation comes first, start with Mobile Loyalty App Development: Features, Cost Drivers, and Roadmap.
When the commerce decisions in When Does an Online Marketplace Business Model Make Sense? get signed off, Discuss Your Commerce Platform.
Ali Boran Gazel