Housi - Replacing a builder app with a real product
A subscription living app and website, with map, search and booking, built to replace an app assembled in a no-code tool.
Subscription living needs a product of its own
Housi created a way of living without the burden of long leases: the apartment is subscribed to as a service. A product like that lives or dies on the experience of finding the place, understanding what is included and closing the deal.
The app that existed had been assembled in a no-code builder. It worked as a shop window, but it could not carry the product the company wanted to offer.
The challenge
- Replace without losing identity. The new app had to take over the same name on the stores, which means coordinating ownership and identifier transfer while the old product is still published. Build before the API existed. At the start there was no data interface of its own. Development began consuming a legacy feed, and Housi's API was only ready months into the project. The map as the main interface. Searching for a property in a list is one thing; searching on a map, with price on the pin and filtering by area, is quite another.
Start with research, not with screens
Before designing anything, the work began by looking at what already existed: a heuristic review of the app in use and a comparison against sector references, in Brazil and abroad.
Then came the part usually skipped: usability research with real people on the prototype, with a defined script, recorded sessions, tabulated observations, a report and an insight matrix to prioritize what to change.
That changes product decisions in concrete ways. Interface adjustments stopped being opinion and started having observation behind them, including changes to elements that looked obvious in the design and were not in use.
On top of that came the high fidelity design of all three fronts: app, website and admin panel.
What was built
iOS and Android app, from a single base. Registration and sign-in, password reset, profile editing, a four screen onboarding with custom illustrations, property search, filters, favourites, property detail, help and settings.
The map. Properties shown as pins, each displaying the rent, filtered according to zoom level. The search field completes the address as the person types.
Website, in React, reusing the data handling layer already written for the app. Less duplicated code and the same behaviour on both ends.
Admin panel, for the operation to manage content.
Deliveries reached the client as successive builds from the first weeks, not as a single handover at the end.
Integrating with data that is still being born
A good share of the technical effort went into something that never shows on screen: making the product work while the data source was still being built on the other side.
The route was to consume the site's legacy feed first and, once the dedicated API was ready, migrate to it. That let the app move forward in parallel instead of waiting.
When the API arrived, the format differences that always appear in integration came with it: geolocation coordinates with an unexpected decimal separator, dates outside the expected time zone standard and monetary values in a divergent format. Each was negotiated item by item with the client's team, and the decision was to fix at the source wherever possible, rather than piling conversions into the app.
One small detail that costs time in every project like this: the feed only responded with a specific request header, something documented nowhere.
Stack and practices
- React Native with Expo, one base for iOS and Android. React on the website, reusing the app's data handling. Google Maps for the property map, with markers, price on the pin and zoom filtering, plus address autocomplete in search. Firebase for staging hosting and dynamic image storage. The client's REST API, with an earlier stage consuming the legacy feed. Usability research with recorded sessions, a report and an insight matrix. Sprints with impediments logged and chased daily until resolved, including those depending on the client.
Is your product stuck in a no-code tool?
No-code builders are good at the stage of validating the idea. The limit shows up when the product needs an experience of its own, integration with your data and performance on the main screen. X-Apps makes that transition, with research before design and builds shipping from the first weeks.