Improving Developer Experience: Clear Linux OS Installer redesign process

Hey everyone,
As a step towards transparency with the community, we are sharing the results of the internal studies that led to the redesign of Clear Linux Installer.
We believe this will give insights to everyone involved in the discussions happening on this topic, and we hope to keep learning from you all.
Let’s get to it.

•••

A crucial aspect of the developer experience is the OS installer. Our journey to improve and come up with the current GUI started with the first version of it. Our team (made up of UX designers and Clear Linux developers) ran a study on this first version of the installer and implemented a new one based on study results.

first-version
The version of Clear Linux OS installer tested (vs 0.7.0)


The goal of the study was to gather qualitative data from developers that were not familiar with the Clear Linux OS installer so that we could better understand the fluidity of the installer workflow, the consistency in the interface, and user comprehension of the required tasks.

••••••••••••••••••••••••

Methodology
A usability test is a method of evaluating a product by testing it on users. Nowadays it is mostly applied in an informal/guerrilla-like method in an environment controlled in a minimal way. As most user researches the statistical significance is not taken into account. Instead, it’s used as a method of broadening perspective and inspiring design. Usability testing is conducted with a group of potential or current users. They are asked to complete a series of typical tasks. Usually, tests are recorded and analyzed to identify areas of product improvement. It is known that 5 users are a sufficient sample size to inform about the performance of the interface [1] [2].

Participant profile
We recruited six participants who had attended an internal Clear Linux OS training workshop. Each participant received a USB drive with a Clear Linux OS image to install it on a provided laptop.

Participants gave their impressions while going through each installation step. After finishing the installation, participants responded to ten statements regarding how they felt about their experience with the installer. The statements were presented through a System Usability Scale (SUS) survey, using a rating scale from Strongly agree to Strongly disagree.

Study process
The team did an initial pilot test with a single team member to refine the test flow. The team then performed the study over two weeks with six evaluation sessions. In the observation room, screenshots of the Clear Linux OS installer were posted around the room, and team members used sticky notes to document interesting patterns, participants’ comments, and design issues. After each session, the team placed the sticky notes on the appropriate screenshots.


Notes taken during usability tests.


Summary of SUS ratings

By the end of the six evaluation sessions, the notes allowed team members to see which screens in the installation process were causing issues. The team then came together to grouping common problems throughout the whole installation process.

••••••••••••••••••••••••

Findings and Recommendations

Feedback gathered from the usability test within the clusters found during a team workshop.


With the feedback provided by the participants, we identified improvements for the installer’s consistency and to facilitate better understanding by the user about what is required to complete the installation process. The study was valuable in identifying three areas that needed more attention: component focus, visual consistency, and content.

Component Focus

25
Color pattern for selected items on the original version of the installer. Changing between screens caused some users to misunderstand selected options.


Many participants were confused about what component was in focus on a given screen. Their confusion was especially apparent in the setup of a new partition when participants needed to select a partition type and then specify the partition size.

We developed a specific color palette to provide contrast and guidance in the interface, making it clear to the user what is or is not selected, and what is or is not available. We also recommended two navigation patterns: keyboard arrows to select options inside the main component of the screen (lists in general), and hotkeys to navigate through tabs and buttons.

Visual consistency
The second area of concern was visual consistency. Some components were displayed in different ways from one screen to another, making it difficult for participants to know what to expect when interacting with a component.

We designed a style guide focusing on the main components to define the patterns that the interface should follow. These patterns were created using color, shape, text, and behavior to decrease the user’s cognitive workload while navigating through the screens.


Style guide created to support visual consistency for the Text User Interface (TUI).


Content
Last but not least, we tackled installer screen content. We found some participants were unsure of what to do with on-screen options or actions.

first-version
The original version of the installer home screen with advanced options. Not easy to find out what’s inside.


We recommended the addition of text hints for the most important options, to help users understand the options and defaults, as well as visual indicators of edited options.

after

The current implementation of the TUI. Hints give the user an idea of what to expect inside each option and when the default state has changed.


The UX team also incorporated the following changes in the design present in the latest implementation:

  • Items in the Menu should be presented in tabs and separated into required and optional configurations.
  • Navigation components (components used to navigate from one screen to the other) should be different from content components (components used to input data to configure the Clear Linux OS installation).
  • Screen titles should be the same as listed in the Menu (e.g., clicking in “Bundle selection” leads the user to the “Bundle selection” screen.
  • Unify partition setup in one screen: The user sees current partitions in available disks and can choose to edit partitions manually or to create default partitions in a given disk.
  • Items that still required user input should be highlighted to facilitate the process.
  • Provide a review page to check details before starting the installation process.
  • While installing, the installation screen should display progress to make sure the user is aware of what’s happening.

•••

From Text User Interface to the Graphical User Interface

Using the information architecture of the improved TUI and the findings from our study, a new GUI installer was developed. The user flow and interactions designed for the TUI during this process could be implemented and enhanced for the GUI.

A GUI grants a broader range of interactions and possibilities. The implementation of visual assets supports users through iconography to identify their path and guide them through the install process.

GUI implemented in the latest release. Mouse support and other capabilities are available now.


Future improvements are in progress, and we are already working on more details regarding the look and feel of the interface to match the branding style guide, making sure users will have the same experience on different touchpoints with Clear Linux OS.

Future designs. Branding alignment and accessibility


••••••••••••••••••••••••

Further actions

It’s been a long journey from the initial investigation to the current implementation. As a rolling release OS, we believe that actively listening to people using the product is the key to creating a dynamic environment that will push innovation. With the feedback gathered from the study, we were able to identify and improve the major pain points for the user and improve the overall user experience during installation. Now, with the community involved, the possibilities are massive. Our goal is to consolidate findings in our forum discussions and shape the product accordingly, sharing our results, usability tests, and everything else that will empower transparency with the community.

••••••••••••••••••••••••

References:

[1] How Many Test Users in a Usability Study?

[2] Why You Only Need to Test with 5 Users

[3] Usability 101: Introduction to Usability

[4] Rating The Severity Of Usability Problems

[5] Severity Ratings for Usability Problems

[6] A Practical Guide to Usability Testing

[7] Report Usability Issues In A User By Problem Matrix

••••••••••••••••••••••••

Team:
UX: Beatriz Palmeiro, Fabricio Novak, Marina Strano
Core development: Leandro Dorileo, Mark Horn, Reagan Lopez
Gnome prototype: Ikey Doherty

7 Likes

https://community.clearlinux.org/t/how-can-we-improve-our-installation-experience/116/67