Directory or transacting marketplace: decide first
A directory lists providers and lets people contact them. The deal happens off the platform, by phone or message, and you earn from listing fees, ads or nothing yet. A transacting marketplace takes the payment, holds the record of the deal and earns a fee on it. The second is a much bigger product, because it owns the money and the problems that come with it.
| Question | Directory | Transacting marketplace |
|---|---|---|
| Where is the deal agreed? | Off the app | Inside the app |
| Who takes the payment? | The provider | The platform, then pays the provider |
| How do you earn? | Listing fees, featured spots, ads | A fee per transaction |
| What does it prove? | Supply exists and buyers look | Buyers will pay through you |
| Admin needed | Approve and hide listings | Listings, orders, refunds, payouts, disputes |
| Main risk | People take the deal elsewhere | Operational load and payment edge cases |
Neither is wrong. If you do not yet know whether providers will list, start with a directory, or even a website. If you already have supply and want to know whether buyers will pay through the platform, the transaction is the thing to test.
One transaction, traced from listing to payout
Here is a single home-cleaning booking, step by step, with the screen or system each step needs. Every row is part of the minimum; drop one and the transaction breaks.
| Step | Who acts | What the app needs |
|---|---|---|
| 1. Cleaner signs up | Seller | Account, profile, payout details collected by the payment provider |
| 2. Cleaner is approved | Your team | Admin review queue with approve and reject |
| 3. Cleaner lists a service | Seller | Listing form: service, area, price, photos |
| 4. Customer finds it | Buyer | Category list, one or two filters, listing detail |
| 5. Customer requests a slot and pays | Buyer | Date picker, checkout, payment held |
| 6. Cleaner accepts | Seller | Push notification, accept or decline screen |
| 7. Job happens | Both | Order status screen with address and time |
| 8. Job marked complete | Seller or buyer | Complete button, confirmation to the other side |
| 9. Payout released | System | Payment provider transfers the seller's share minus your fee |
| 10. Something goes wrong | Either side | Report a problem, admin refund action |
Count the roles: buyer, seller, your team and the payment system. Step 10 is the one founders leave out, and it is the step that decides whether people trust the platform with a second booking.
The feature list for a testable first version
Must have
- Sign-up for both sides, with the seller flow collecting payout details
- One listing type with fixed fields
- Browse and a simple search or filter for one category
- Checkout with the payment held or captured at a defined moment
- Accept, decline and complete actions with notifications
- Order history for both sides
- Admin: approve sellers, hide listings, view orders, issue a refund
- A report-a-problem path that reaches a person
Can wait for version two
- Open chat before purchase (order-linked messages are often enough)
- Detailed reviews and seller ranking
- Promoted listings and seller subscriptions
- Saved searches and recommendation feeds
- Multiple categories, cities or currencies
- A full web version for buyers
Ratings sit on the line. A single star rating after a completed order is cheap and helps trust; a review system with replies, moderation and ranking is a feature of its own.
What the first version should tell you
Define the test before you build, or every result will look encouraging. Useful questions for a first marketplace release:
- How many invited sellers finish a listing without help?
- Of buyers who open a listing, how many start checkout, and how many finish it?
- How many accepted orders get marked complete without a dispute?
- Do buyers come back for a second transaction through the app, or go direct to the seller?
The last question is the one a directory cannot answer. If buyers keep going direct, the fee model needs a rethink before you add features.
From feature list to a scoped quote
Bring the transaction trace and the two checklists to a quote. They answer most of the questions a studio would otherwise ask in discovery. For a breakdown of where effort goes in a marketplace build, read what drives marketplace app cost. Our marketplace MVP page shows the kind of first version we scope, and the quote form takes about four minutes.
Frequently asked questions
What is the smallest marketplace MVP worth building?
One that completes a single transaction from listing to payout, with an admin refund action. Anything smaller is a directory or a lead form.
Is a directory a bad first step?
No. A directory is a sensible way to test whether supply exists, but it cannot show whether buyers will pay through your platform.
Do I need in-app chat in the first version?
Usually not. Messages attached to an order, or contact details shared after payment, cover most first versions at a fraction of the effort.
Should reviews be in the MVP?
A simple rating after a completed order is worth including. A full review system with replies and moderation can wait.
Can the admin side be a web page instead of an app?
Yes, and it often should be. Your team will manage orders and sellers more comfortably on a desktop screen.
Written and reviewed by the DFY Games team. Published ; facts checked . Prices and store rules change: check the linked sources before you rely on a figure.