Most marketplace ideas fail in ways that were knowable before a single line of code was written. The supply side would not have bothered. The demand side already had a workaround they liked. The take rate the economics required was higher than either side would tolerate. Every one of those facts can be discovered in weeks, for almost nothing, by talking to people and running transactions by hand. This post is the validation sequence we run with founders to discover them.
One framing note before the tests. Validation is not about proving your idea is good. It is about generating evidence either way, as cheaply as possible, with kill criteria you set in advance. If you go in looking for confirmation, you will find it, and it will be expensive.
What Validation Actually Means
For a marketplace, validation means demonstrating three things with evidence rather than opinion. That the supply side will show up and list. That the demand side will show up and pay. And that the exchange between them supports a fee that funds the platform. Miss any one of the three and the marketplace does not work, no matter how strong the other two look.
Notice what is absent from that list: surveys about whether people like the idea, waitlist signups, and encouraging conversations with friends. None of those cost the respondent anything, so none of them are evidence. The tests below are ordered so that each one asks somebody to give up something real: time, effort, or money.
You Are Validating Two Businesses, Not One
A marketplace is two customer acquisition problems joined by a transaction. The suppliers and the buyers have different needs, different alternatives, and different reasons to ignore you. Each side needs its own validation, and the side founders skip is almost always supply.
That is backwards. As we argued in supply comes before demand, supply is the side that determines whether early demand converts, and it is usually the side with less obvious incentive to join. Validate it first.
Test One: Supply Interviews
Talk to fifteen to twenty potential suppliers in your niche. Not a survey, actual conversations. You are listening for three things: how they get customers today, what that costs them in money and annoyance, and what they would need to see before listing on a new platform.
The pass signal is not politeness. It is specificity. A supplier who starts negotiating details with you, asking what the fee would be, how payouts work, who handles disputes, is treating the platform as real. A supplier who says “sounds great, let me know when it launches” is being kind. Count only the first group.
Test Two: The Demand Signal
Now the buyer side. The cheapest honest test is a landing page that describes the offer as if it were live, plus a small amount of targeted traffic. But the metric is not signups. It is what people do when asked for something concrete: a deposit, a booking request with real dates, a phone number for a callback about a specific need.
Even better is intercepting demand where it already exists: the Facebook groups, subreddits, and classifieds where your future buyers currently solve the problem. If demand exists, it is already flowing somewhere, badly. Your job in validation is to find that flow and confirm the pain, not to conjure demand from nothing.
Test Three: The Manual Transaction
This is the test that matters most, and the one most founders skip because it does not feel like building a startup. Broker five to ten real transactions yourself, by hand, with no platform. Match a real supplier to a real buyer, coordinate the exchange over text and e-transfer, and take your intended fee, or at least state it.
The manual transaction test converts every assumption in your deck into observed behaviour. You learn whether matching is actually hard, where the friction lives, what both sides worry about at the moment of commitment, and whether anyone flinches at the fee. It is also the single best preparation for the cold start that follows, because you will have done by hand exactly what the platform must do at scale.
If you cannot make the transaction happen manually, with all your founder energy focused on five customers, software will not make it happen either.The uncomfortable rule
Test Four: The Economics Check
With real numbers from the manual transactions, run the arithmetic. Average transaction value, times realistic frequency, times the take rate nobody flinched at. That is your revenue per active user. Set it against what acquiring each side will plausibly cost and what serving the transaction requires.
Marketplace economics fail in predictable places: order values too low for the take rate to matter, purchase frequency too rare for retention to compound, or a category where the natural fee is thin because the platform adds little beyond the introduction. Our take-rate post covers how to reason about the fee side of this equation.
Kill Criteria: When to Walk Away
| Test | Healthy signal | Kill signal |
|---|---|---|
| Supply interviews | Half negotiate specifics unprompted | Polite interest, zero follow-through on asks |
| Demand signal | Strangers commit dates, deposits, or details | Signups that never answer a follow-up |
| Manual transactions | Both sides return for a second exchange | You cannot close five with full founder effort |
| Economics | Fee tolerated and unit math closes | Take rate must exceed what anyone accepted |
Set these thresholds before you start, in writing. Validation without pre-committed kill criteria is theatre, because every ambiguous result will be read as encouragement. Walking away from a bad idea after four weeks of testing is a success. The failure is discovering the same facts after a year and a build budget.
This sequence is exactly what our pre-development sprint runs with idea-stage founders, with the added benefit of people who have watched dozens of these tests play out and know which signals lie. If you want that judgment in the room while you decide, marketplace consulting is the place to start.
