I joined Pakistan's premiere music streaming service a few days after its public launch. I've curated playlists, written liner notes for original releases, designed launch posters, coordinated partnerships with the Citizens Archive of Pakistan, among a host of other roles as the small team needed. In 2016 I moved from designing UX part time to taking over the product team entirely – leading a team of developers, designers and support staff in their movement from a rag-tag on-fire dev team to a world-class product team.
At its launch the company was a combination of freelancers and music nerds that banded together behind the dream of representing a pop culture they grew up with. Often we found internal processes – technical and human – left us chasing deadlines. Every new release required multiple people. So I built our design to optimize this toolchain.
We reduced the design requirements we had built for custom artwork, added more flexibility to the interface, and over time built a new stack that gave our curators the ability to create entirely new canvases and structure without arbitrary technical restrictions.
Seemingly high product standards were stifling new production. Simplification of process required simpler, more elegant UI. New stylistic components were created, and older ones lightened. Textures and light typography that made for an unusable graphic style were replaced with concrete color and legible type.
We started by live prototyping a UI that opposed the current in almost every way – giving us a new design space in which to aim.
From prototype I started rebuilding our front end nearly from scratch, focusing on accessibility, legibility and scalability. This meant building the Browse product that converted our previous multi-man operation into a production process that required minutes and one person to create an entire new curation surface.
This took us over a year and a half to actually put into production, as we slowly built the pieces not just of this new product – but of a product team that understood the importance of software performance, legibility, and scalability.
Converting our design language was not just about visuals – it was about building a company culture that understood what it means to be a world-class product company. The Pakistani software industry is nascent. But more than its inexperience it is hampered by never having built the tools to cater to its own customer base. A large fraction of Pakistani software companies develop for foreign clients, and are hence used to receiving detailed requirements and building without added attention to polish.
Patari draws an audience of local and diaspora users that are attracted to the product precisely because it delivers the kind of representation that other technology does not. Hence the very basis of our company relied on customer understanding. We had no user testing infrastructure, our ability to react quickly to market changes and new insights was poor, and we had no resilience in an industry with high employee turnover.
So our entire product restructuring was built around internal documentation – of development and design tools. We trained inexperienced developers to move from responding to requirements to eliciting and building specs themselves. Graphic artists were trained to become UX designers, that sketched, prototyped, then redlined. Our product became more reliable, less vulnerable to the team changing (and the team liked it better anyway), and more polished.
Like most Pakistani software, Patari is built in English. Most of Pakistan does not speak English, so this makes little sense. But we did it this way because we had design precedent and tools to get English software built with a small team. Building Urdu software means fighting consistent battles with programming environments, design tools, and guidelines from big companies that are Latin-normative.
Urdu is even worse represented in computing than other languages, and it is not an exaggeration to say that every part of the interface needs retouching. From developing terminology for common interface elements, to changing line heights to make Urdu text legible at the same point sizes, even to the hacks we need to make sure the text we write isn't garbled between our toolchain. And we also have to find ways to convert a huge database of metadata into a different script. Not just us, no one in the industry has these tools. We are slowly attempting to build these tools for Urdu, and hopefully in more local languages.