Bettr.me
An AI-powered sentiment tracking and analysis app.
The Challenge
While mental health in the workplace has become a priority for all types of organizations and their employees in recent years, few useful tools exist in this space. Our challenge was to build a product that allows people to take more control over their mental well being by encouraging them to spend time in a way that leaves them fulfilled, supports their long term goals and provides their organizations with a means to enact meaningful changes to their work environments.
My Role
As the design lead from concept to first release, I was responsible for all aspects of design, including research, testing, prototyping and more. I played a large role in the product’s roadmap, built the design system and worked closely with the development team throughout build.
Team
An agile team of 5 developers, a data scientist, a product manager, a QA specialist and me.
Length
1 Year, Full-Time
First Concepts
Our journey began with a simple business plan and a few concept mockups which were leveraged to get buy in and secure funding from our primary stakeholders. While these conceptual interfaces were a far cry from what we eventually designed and built, they did paint a picture of what was possible and build confidence in the idea.
A concept mockup to sell the vision for the product. These mockups were very different from what we ultimately designed and built.
Competitor Analysis
Our initial search of the market found only a few potential direct competitors. Most of these tools focused on providing people managers with a means to monitor the morale and overall well being of their employees. Few took the approach of attaching sentiment to blocks of time, projects or types of work. Direct competitors generally fell into three categories:
Niko Niko calendars in which individuals provide anonymous daily updates on their general sentiment along with an optional short description.
Natural language processing systems which monitor one or more commonly used communication channels (typically email) and leverage AI to mine for changes in tone and sentiment.
Automated surveys which prompt individuals to answer a few questions a day or week on general morale, engagement or mood.
A direct competitor which leverages natural language processing to track employee morale.
As there were very few tools on the market which sought to meet the same goals as our product, much of our competitor analysis focused on indirect competitors, that is products whose functionality overlaps that of our product while it’s primary goals were different. Time, fitness and personal mood tracking apps were of particular interest as they provided quick and easy methods for tracking everyday happenings in people lives. An analysis of these interfaces provided ideas and inspiration as we began brainstorming solutions and potential features.
A fitness tracking app.
User Interviews
Initial user research was conducted through user interviews, stakeholder interviews and surveys with two broad groups of potential end users, people managers and their direct reports.
Employee participants came from a variety of roles and backgrounds. Our discussions focused primarily on weekly planning, tracking and personal monitoring behaviours. Of those we talked to, we found none that engaged in formal sentiment tracking on a consistent basis over a long period of time. The few individuals that did report tracking sentiment had only done so inconsistently in typically in the form of a diary or journal. While many participants saw great value in tracking their sentiment, they cited lack of time, lack of tools and difficulty tying their emotions back to particular events as key obstacles.
We also found that about half of participants consistently used some sort of tracking app on a weekly basis over a long period of time. These apps were generally focused on fitness or expenses. Common motivating factors for individuals to use these tools were to keep themselves honest/accountable and/or to provide actionable insights on their behaviour that would otherwise be difficult to uncover.
Interviews with people managers from multiple industries allowed us to focus on sentiment collection and analysis at a more organizational level. We found the most common ways in which people managers maintain a pulse of their direct reports is through regular 1-on-1s, semi-regular career development meetings and occasional company wide engagement surveys. While most interviewees felt these methods provided them with a strong understanding of their team’s general morale, there was a clear appetite for a tool that could provide data driven insights on anonymous team member sentiment. Conversations around performance, placing people in the right roles and conflict management were cited as common challenges that a sentiment focused tool could help alleviate.
Notes from a user interview with a people manager.
Personas
Leveraging codified qualitative data from user interviews, stakeholder interviews and surveys, personas were created to represent and build empathy for our target users in a succinct and engaging way. The persona below was one of two based on common goals, needs and frustrations unearthed during user interviews with people managers.
A primary persona representing potential end users who manage direct reports.
User Flow Diagrams
As we began to focus on high level features, user flow diagrams allowed us to visualize how our users would interact with the product and accomplish their goals in variety of scenarios. These diagrams came in varying levels of fidelity. The diagram below was scaled over several iterations until it became a blueprint for how users would interact with a proposed Slack integration feature.
A detailed user flow diagram which visualizes how users will interact with a Slack integration.
Feature Ideation
When problems were clearly defined, I ran frequent brainstorming sessions which allowed our team to ideate on a wide range of solutions before prototyping, testing and refining our designs. The example below shows the results of one such session focused on a sentiment entry mechanisms. As you can see from the options presented, not all ideas are created equal, but every idea deserves a fighting chance!
The results of an brainstorming session on potential sentiment entry mechanisms.
User Testing
Regular user testing of prototypes uncovered flaws and allowed us to rapidly iterate on designs before they reached development. A structured approach to user testing forced us to maintain objectivity and consistency across tests even when faced with limited time and/or resources. In some cases, early user testing completely changed the direction of major features. The test report below for example outlines several issues with a ‘sync from calendar’ feature. The testing ultimately forcing us to rethink the feature and focus our efforts on integrations with existing calendar apps instead.
The results from a set of user testing session in which a proposed feature was tested.
Team Collaboration
As we were working in a highly agile environment, close collaboration between development, design, quality assurance and product management was necessary in order to ensure the team was focused on features that would provide the most value to the end user. Borrowing a few pages from Jeff Gothelf and Josh Seiden book Lean UX, I played a large role in designing and instituting a process which defined how multiple features should be designed, built and tested in a two week sprint structure.
A process diagram illustrating the staggered feature design and development sprint cycle followed by our team.
A process diagram illustrating the complete life cycle of a feature from design to release.
High Fidelity Prototype
Since most user testing was conducted with wireframe prototypes, the primary purpose of hifi prototypes was to create a pixel perfect design which, along with user stories and a robust design system, included all of the specifications developers needed to implement the solution. Nonetheless, we often put these hifi prototypes through another round of testing whenever time permitted.
Since we built integrations with various apps (screenshots below) to give users an easy way to enter sentiment into the system, the main purpose of the web app was to view data and insights related to previously entered sentiment. The dashboard served this purpose by allowing managers to view their team’s sentiment on an aggregate level and individuals to review their own entries in a variety of visualizations.
Filters allows users to quickly zero in on particular subjects, time periods and tags. In this view, the user has changed the time period to view the last 30 days of sentiment for their team.
While users could enter sentiment into the system through this Entries page, most used the various plugins and integrations that we developed for sentiment entry. The main purpose of this page was to provide an easily navigable history of sentiment entry.
Admin settings allowed super users to add or remove users, create and manage groups and edit default subject lists.
Notification settings allowed users to subscribe to digests and set up reminders through integrations with the apps they used most including Slack, Outlook and Google Calendar.
Extensions
Findings from initial user research suggested that the success of the product hinged on making it easy for users to connect sentiment to things that were happening throughout their day in real-time. Thus, we built several extensions which gave users the opportunity to log sentiment where ever they work, including in their favourite calendar apps and browsers.
Slack App sentiment entry
Browser extension sentiment entry
Calendar extensions sentiment entry
Email Templates
I designed customizable email templates to provide a personalized view of one’s sentiment data or the data of direct reports. These were meant to encourage further engagement with a steady stream of actionable insights delivered straight to one’s inbox. Of course, users could enable or disable these emails through their personal settings.
Personal weekly summary email.
Manager weekly summary email.
Design System
Our design system was built from the ground up to create a unique visual identity, ensure easy scalability, force consistency and reduce ambiguity in the design to development handoff. Given time constraints and our agile approach, the system was developed and evolved over time as our designs matured. It remains a living document to this day.
Colours
Component blueprint for a custom date filter.
Fonts and Icons
Component blueprints for control components.
Component blueprint for dashboard filters.
Outcomes
Bettr.me is being leveraged by many organizations, both large and small, across a range of industries to improve work environments, and by individuals to place a greater focus on their mental well being. The Covid-19 pandemic has resulted in even more interest in the tool as organizations adapt to remote working environments.
“Bettr.me showed me how much happier I am when I learn while I work. Now it helps me plan time each week to read a white paper, watch a demo video, or run a learning session to share some knowledge.”
- Olivia, Engineering