fizy, Turkcell, 2019 to 2020

Making fizy yours, on one design system

Listeners wanted fizy to look the way they liked. Themes were the biggest step toward that, but the app had no shared rules to carry them. I mapped the app, turned a component library into a design system and rebuilt the themes on a color logic.

Role
Senior Product Designer
Team
A lead PM, a tech PM, a product owner, developers and an external agency
Timeline
2019 to 2020
Scope
Mobile app themes and design system
My role
Mapped the live app, built the design system and its color logic, and took the themes project from roadmap to release.
Problem
Listeners wanted to choose how fizy looks, but the app had no system that themes could be built on.
Impact
14 themes shipped on deadline, all built on one color logic in the design system.

Overview

fizy is Turkcell’s music, podcast and video service, with more than a million monthly listeners. In 2019 the product team made personalization a main goal for the mobile app, and themes were its biggest piece: listeners would choose the colors and look of the app. Before themes could ship, the app needed a system that could hold them.

Problem

The same look for everyone

fizy looked the same for every listener. A more personal and more colorful app had been one of the most common requests for years.

→ Shapes Decisions 03

No file of the live product

There was no documentation of the current flows. Finding the latest version of a screen meant digging through folders or asking the agency.

→ Shapes Decisions 01

The same part, built many ways

Lists, headers and cards looked different from screen to screen, and colors had no defined roles. Every change was made by hand.

→ Shapes Decisions 01 and 02

Theme concepts that could not be built

The agency drew its themes on outdated screens and recolored parts with no logic, so they would not hold up across the twenty or so themes planned.

→ Shapes Decisions 02 and 03

Outcome

14 themes, one system

Monthly listeners

1M+

on one design language

Themes shipped

14

on deadline, all on one color logic

Approach and rationale

Fix the base, then build on it

Repainting the agency screens would have meant twenty or so one-off files that nobody could maintain. I proposed a different order: understand the product first, build the guide, rework the palettes on it, then apply and hand over theme by theme.

  • Step 01

    Product analysis

    Map every live and approved flow, and list which parts repeat and which have drifted.

    → Decisions 01
  • Step 02

    Design guide

    Rebuild the foundations and components as a library the themes could plug into.

    → Decisions 01
  • Step 03

    Rework the palettes

    Check each agency palette against the live app and give every color a role.

    → Decisions 02
  • Step 04

    Apply colors to the guide

    Swap the role colors through the library and test each theme on real flows.

    → Decisions 03
  • Step 05

    Hand over to development

    Send approved themes one at a time, so the build moves while design continues.

    → Decisions 03

Execution

Making it buildable, on a deadline

I was a design consultant, first through IBM iX and then Amaris. Design joined the themes project late, so the first job was a plan the whole team could follow.

Drew the roadmap

I drew a roadmap for the project and brought the product team into it, so every step had an owner and a date.

Set standards with the developers

I agreed the product standards with the software and product teams first, so what I designed was what they could build.

Kept every stakeholder in the loop

I ran communication from the first review to release, across the product, business and development teams and the agency.

Held the deadline

I managed the design timeline against the roadmap, so the themes reached release on the agreed date.

Decisions 01

One system under every theme

Responds to: No file of the live product, and the same part built many ways

Map of every live and approved fizy flow in Miro, broken into atoms, molecules and organisms

I had started a component library when I joined, and we used it in new designs. This project turned it into a real system.

I mapped every live and approved flow in Miro, since no file of them existed, and broke the screens into atoms, molecules and organisms to see which parts repeated and which had drifted.

Then I rebuilt the foundations as a symbol library, with the lists, menus, header, nav bar, mini player and playlist cards on top.

  • Color

    One palette with clear rules for text, backgrounds and actions.

  • Type

    One font family, down from two on mobile and three on web.

  • Spacing and radius

    Defined steps, so screens and components line up.

  • Icons

    An icon guide that cut the time spent searching for icons.

  • Components

    Buttons, cards and frameworks, with every state documented.

fizy design system sheets: colour guide, headline styles, list elements, navigation, playlist cards and icons

Decisions 02

Colors with roles, not paint

Responds to: Theme concepts that could not be built

The agency concepts recolored parts with no logic. A white item turned blue, the yellow action button changed color, and the black header and the yellow button were given the same color. I gave each color in the system a role instead. A theme became a new value for each role, not a repainted screen.

Live app

Watch screen in the live app
  1. Colors, components and areas each have a defined job on this screen. Every new item follows the same logic.

Agency concept

Watch screen in the agency concept
  1. The concept was meant to go beyond recoloring areas, with colors built on a clear logic that could be developed. It did not do that.

Decisions 03

Themes listeners choose

Responds to: Theme concepts that could not be built

Starting over would have cost the deadline, so we kept the agency’s palettes and rebuilt each one on the color roles. Every theme was tested on real flows before it shipped.

  1. Copy the main file

    A copy of the main design file for each theme, named after it.

  2. Swap the role colors

    New values go into the color symbols, so every screen updates at once.

  3. Check every flow

    Each flow is reviewed, and any color that is hard to read or use is fixed.

Headline and card style sheets in five themes, each with its own role colors
  • Homepage in the Original themePlaylist in the Original themePlayer in the Original theme

    Original

  • Homepage in the Rock Star themePlaylist in the Rock Star themePlayer in the Rock Star theme

    Rock Star

  • Homepage in the Coffee Love themePlaylist in the Coffee Love themePlayer in the Coffee Love theme

    Coffee Love

  • Homepage in the Love Story themePlaylist in the Love Story themePlayer in the Love Story theme

    Love Story

  • Homepage in the Sounds Green themePlaylist in the Sounds Green themePlayer in the Sounds Green theme

    Sounds Green

  • Homepage in the Party Time themePlaylist in the Party Time themePlayer in the Party Time theme

    Party Time

  • Homepage in the Sunset Dream themePlaylist in the Sunset Dream themePlayer in the Sunset Dream theme

    Sunset Dream

  • Homepage in the Deep Waters themePlaylist in the Deep Waters themePlayer in the Deep Waters theme

    Deep Waters

Learnings

What I learned

  • Know the product before the project lands

    We joined the themes project late, and missing documentation cost us time. With the flows already mapped, design could have stepped in from the first brief.

  • Bring design into the brief

    The agency concepts were made without the live product in view. Design in the brief from day one would have saved a full round of rework.

  • A system pays off on the next project

    The color roles made each new theme a swap, not a redesign. The same base later carried the web player too.

Hiring a senior product designer,
or want to talk design systems?