Skip to content
All guides

· 8 min read

How to build a marketplace app: lessons from Rehnaa

How to build a marketplace app, from a rental marketplace we built: two kinds of user, trust, map search, chat, one API for app, web and admin, and testing.

Blueprint-style cover reading: Building a rental marketplace

How to build a marketplace app comes down to a handful of decisions made early: who the two sides are and what each sees first, how one side comes to trust the other, how people search, where they talk, and how many screens share one backend. We learned most of these on Rehnaa, a rental marketplace for Pakistan where tenants search rooms, hostels, homes, shops and offices on a map and chat with verified landlords. This post turns that build into lessons you can use on your own marketplace.

Some context first. Rehnaa, formerly SAWAAR, started as an Expo mobile app with a Next.js admin panel and an Express and MongoDB API. We built rehnaa.pk, the web version of the app, on that same API.

What a marketplace MVP needs, and what can wait

Build in the first versionLeave for later
Separate sign-in paths for each sideReviews and ratings on both sides
A listing flow the supply side can finish on a phoneBulk upload and listing imports
Search with filters, a list view and a mapSaved searches and alerts
Direct chat between the two sidesPayments and deposits inside the product
A simple dashboard for the supply sideDetailed analytics for the supply side
An admin panel to check and remove listingsAutomated moderation

The right-hand column matters, but it does not decide whether the market works. Our guide on how to scope an MVP covers how to make these cuts. For how the build itself is planned, see how long it takes to build an MVP, week by week.

Lesson 1: a marketplace is two products sharing a database

Tenants and landlords want different things from the same data. A tenant wants a place. A landlord wants enquiries from suitable tenants.

On Rehnaa, the role is set at sign-in, and it decides where you land. Tenants go to the home feed. Landlords go to their dashboard. That one routing decision makes each side feel the product was built for them, and it keeps landlord tools out of the tenant's way. The same split works outside rentals: MealMentor, a meal-planning platform we built, has one portal for chefs and another for users.

Practical points from this:

  • Decide the roles before design starts. Every screen belongs to one side, the other, or both. Write it down in the screen map.
  • Give each role its own home screen. Do not make one side dig through menus built for the other.
  • Allow for people who are both. Some landlords are also tenants somewhere else. Decide early whether one account can switch roles or needs two.

Lesson 2: make supply easy to add

A marketplace with no listings has nothing to show the demand side. So the supply side's main task, posting a listing, has to be short and forgiving.

Rehnaa's listing flow is a 4-step wizard, and drafts are saved locally as the landlord types. If the phone locks, the browser closes or the connection drops halfway through, the work is still there when they come back. Splitting a long form into steps also means each screen fits on a phone without endless scrolling.

If you are building a marketplace, time how long it takes a real person on the supply side to post their first listing on a phone. Then cut steps until that number is comfortable.

Lesson 3: trust is a feature, not a policy

Rehnaa's promise to tenants is that they talk to verified landlords. In a rental market, where a tenant may be sending a deposit to a stranger, that promise matters more than any filter.

Trust in a marketplace usually rests on a few parts working together:

  • Sign-in people recognise. Rehnaa supports email or phone login, Google sign-in through Clerk, and password reset by one-time code. Phone sign-in matters in markets where many people use a phone number before an email address.
  • A verification step for the side that carries the risk. Decide what you check, who checks it, and what a verified badge means in plain words.
  • An admin panel that can act. Someone on your team needs to review, hide and remove listings and accounts without asking a developer.

You do not need automated identity checks on day one. A person checking documents by hand is often the right start, as long as the rule is clear and the admin panel can apply it.

Lesson 4: search is location first

People rarely search for a rental by keyword. They search for a place near something: a city, a neighbourhood, a university, an office. Rehnaa lets tenants search by city or by university, then narrow the results with filters and sorting in a list view or a map view.

The map does a lot of the work. It is a full-screen OpenStreetMap search with a radius slider from 5 to 50 km and a "search this area" button, so a tenant can drag the map to a part of town and see what is there. Two things we would tell anyone adding a map:

  • Offer both a list and a map. Some people think in streets, others in prices. Let them switch without losing their filters.
  • Read the map provider's terms before launch. If you use OpenStreetMap's public tiles, its tile usage policy asks for clear attribution and warns that heavy use can be blocked. Plan for a paid or self-hosted tile service once traffic grows.

Lesson 5: keep the conversation inside the product

On Rehnaa, tenants and landlords talk through real-time chat built on Socket.IO. On a phone, a conversation opens full screen with a back arrow, the way it does in the app.

Chat inside the product does three jobs. The landlord sees every enquiry in one place, your team can see the market working, and there is a record when something goes wrong. If people swap phone numbers in the first message and leave, you lose all three.

Real-time chat is one of the larger pieces of a marketplace build. If budget is tight, an enquiry form that notifies the other side can come first, with chat added once you know people use it.

Lesson 6: give the supply side a reason to come back

Landlords on Rehnaa get a dashboard with overviews for the last 7 days, the last 30 days and all time, a list of inquiries, and tenant matches with a match score and the reasons behind it.

The reasons matter. A score alone asks the landlord to trust a number. A score with "matches your area" or "fits your budget" explains itself. Whatever your marketplace sells, the supply side should open the product and see, in one screen, whether it is working for them.

Lesson 7: one API for the app, the web and the admin panel

Rehnaa runs three clients on one Express and MongoDB API: the Expo mobile app, the Next.js admin panel and the web version we built. The web version has 20 routes that mirror the app for both sides: the home feed, search, property details, messages, profile, the landlord dashboard, listings and inquiries.

Because the API already existed, the web build was mostly a front-end job. That is the payoff of one backend: a listing posted from a phone shows up on the web and in the admin panel with no syncing.

If you are deciding where to start, our post on whether to build a web app or mobile app first covers the trade-offs. Whichever you pick, design the API so the second client is easy to add.

Lesson 8: most marketplace users are on a phone

Below 1024 pixels, rehnaa.pk switches to the mobile app's layout, bottom tab bar included. Someone who has used the app finds the same screens in the same places in their browser. Someone who has not still gets a layout built for a thumb rather than a shrunken desktop page.

Design every screen for a phone first. In a marketplace, both sides often act on the move.

Lesson 9: test both sides, end to end

Every marketplace feature has two ends. A listing has to look right to the landlord who posts it and the tenant who finds it.

The Rehnaa web version has 174 Playwright tests across desktop, mobile and responsive runs. It also has a demo mode that runs the whole site on fixtures with no network, which makes it possible to show and test every screen without touching live data or waiting on the API.

You do not need that many tests in week one. You do need the main journeys covered for both sides before launch: sign up, post a listing, search, open a listing, send a message, reply.

Lesson 10: keep the brand easy to change

When SAWAAR became Rehnaa, we moved the product to the new name with a logo system drawn in code, new app icons and a new link-preview card. Names change more often than founders expect. Keep the name, logo and colours in one place in the code, so a rename is a small job.

How to build a marketplace app: a short checklist

  1. Write down both sides and the first screen each one sees.
  2. Time the supply side's first listing on a phone, and shorten it.
  3. Decide what "verified" means and who checks it.
  4. Build search around location, with a list and a map.
  5. Keep conversations inside the product.
  6. Give the supply side one screen that shows whether it is working.
  7. Put every client on one API.
  8. Test the main journeys for both sides before launch.

Marketplaces are one of the kinds of product on our web app development page, next to SaaS platforms, online stores and internal tools, and we have a separate page on marketplace development. If you are planning a marketplace of your own, our MVP development process starts with a blueprint week and a fixed price, or you can tell us what you are building and we will come back with a plan.

  • Marketplaces
  • Case study
  • Web apps

Building a first version of your product?

We scope it in a week and launch it in 8–12, for a fixed price. Tell us what you want to build.