Redesign a working taxi app so that a stranger can book, ride and pay without being taught, while keeping every kind of trip the taxi trade already offers.
The app had a job before it had a design problem. PideTaxi's cars are licensed taxis with meters, regulated fares and a radio dispatch behind them. The competition was apps that had none of that and looked better. So the face-lift could not become a copy. A taxi is hailed from the street. A taxi is paid in cash, or on a card terminal in the back seat, or on a company account when the passenger is on the clock. A taxi will take someone who has not decided where they are going yet. None of that exists in a ride-hailing app, and all of it had to survive the redesign. I started from iOS, and it became the reference the Android work was translated from.
Decision
The payer is not always the rider
A ride-hailing app has one payer: the account holder, with a card on file. A taxi has never worked that way, and the passengers this app serves did not either. The payment sheet lists them all at the same level. Cash. The card terminal in the back of the cab. A saved card. And a subscriber account, which is a company paying for its employee's journey, with its own name and its own billing details kept apart from the passenger's. That last one is the design problem in miniature. The person in the seat is not the person who pays, and the screen has to serve the one who is present without hiding the one who is not. So subscriber accounts are chosen like any other method, and the invoice goes where it belongs.
Decision
Know who is coming before they arrive
The wait is where a taxi app is judged, and this screen shows a person before it shows a car: photo, name, rating, and a call button, above a route drawn to where the passenger is standing, with the minutes left. None of that is decoration. A licensed taxi comes with a licence, a driver and a meter, and those are the things a ride-hailing app cannot promise. Putting the driver's identity on screen first is how the app makes the taxi's own argument for it.
Decision
The door is short, and it does not lie
Sign-up is one thing per screen. Email, then a name, then a password, then the code that proves the phone is real, and each step has its expired state and its error state, because a verification code that arrives late is not an edge case on a phone that just landed in a foreign country. What the app does not do is pretend. If there is no taxi near the passenger, the screen says so. The searching state has a no taxi outcome designed into it, rather than a spinner that runs until the passenger gives up.
Result · The iOS app shipped as version 4.0.1 and became the reference for the Android build. What it demonstrably produced: a booking that accepts every way a taxi is paid for, a wait that shows the passenger who is coming, and a sign-up short enough to finish on a kerb. The app is on the App Store today. No figures from the redesign were shared with me, and none are claimed. Next chapter: Android, the same product on a platform with its own grammar.
Read more