Image from IBM marketing page for Return to Work Advisor
Human problem
Circa first quarter 2019, our research indicated that many large organizations were responding to the pandemic by forming ad hoc teams charged with monitoring changing conditions and responding with daily decisions about workplace operations and employee welfare. Teams typically involved some combination of departmental or business unit executives, location leaders and their key operations personnel, human resources, legal counsel, and medical subject matter experts.
As a member of this team, a user needed to understand the aggregate health and availability of the workforce at each worksite in a location (such as a city), in the context of the recent epidemic trends of the surrounding county and state - or other geographical units, as appropriate to the containing country.
Business problem
To address the human problem, Product Managers envisioned a command center centralizing data and analytic insights for daily decision-making:
I had four roles:
Challenges
Pivoting quickly. The COVID-19 virus introduced societal and market uncertainty that disrupted our clients' healthcare priorities and our business as usual. However, it presented opportunities to help people to return safely to their place of work or study. Our scrappy design team pivoted from our existing projects to meet the urgency.
Time. We had to move really fast to deliver a solution, motivated by what we perceived as a limited window for having a real life impact on employee well-being for our clients.
Radical concurrency. The time constraints and unknowns had us rethinking every part of our approach for taking a product from concept to delivery. We overlapped defining the abstract and concrete parts of the design much more than usual.
Design approach
Return to Workplace Advisor product and Watson Works suite
Despite a compressed timeline, this was my most "design-led" project ever. It surprised me because it would have been easy for engineers or others to conclude there wasn't enough time for design activities to "slow down" the start of development. However, Design was able to accelerate development and overall time to value.
Watson Works design activities from abstract to concrete (Note: Many Images purposefully obscuring details)
Using mental models to think through each plane rapidly
Personas and organizational context (teams, ecosystem). Research consisted of market intelligence, Enterprise Design Thinking exercises to crowd-source team insight, and interaction with sponsor users. Product Managers and Designers collected and applied it throughout the project.
Identifying target personas, organizational context, archetypes
Personas to drive mental model, layout, and navigation design
Workshopping user needs ("mad libs") for quick validation. Design partnered with Product Managers and Data Scientists to explore user needs in adjacent markets by organizing Enterprise Design Thinking workshops. User needs statements or "hills" typically describe Who, What, and Wow. My design director and I collaborated with our data science colleagues to patent a way to accelerate identifying users' analytic intent and supporting data sources for a hill.
"Mad libs" - Adding analytics and data to "hills" from Design Thinking (patented)
Example of partially completed "mad libs" templates
Mental models and information architecture for sense-making. As we went along, I drove our alignment on concepts, terminology, and how to present them in the user interface for quick comprehension.
Mental model that ultimately guided chunking the UI into tabs
Mapping mental model to business questions analytic insights
Wrangle those Key Performance Indicators. To design Watson Works, the cross-product solution, our squad of designers collaborated with 3-5 product managers who gave us quite a list of key performance indicators (KPIs). I spent some time consolidating them and looking at them from different perspectives in Mural before holding a playback to review a straw man proposal for which KPIs to bubble up to the main display.
From 30 to 9 KPIs for initial discussion with product managers
Mapping KPIs to business questions and themes they address
Mind map close-up
Analyzing which KPIs are top-level vs. specific to a business theme
Low fidelity mockup introducing business themes as UI tabs
Medium-to-high fidelity in Mural to show where KPIs ended up
Design system and reusable team assets for the win. By recommending to engineers the Carbon Design System and Carbon Charts for this new project, I was able to save an incredible amount of back-and-forth on interaction design, information design, visual design, and branding because those detailed concerns already are well designed and nicely packaged. This freed up energy to innovate on the unique aspects, namely the COVID analytics and visualizations. Another accelerator was the time I took previously to create a chart library in Sketch specifically for health care oriented visualizations used by our team.
Using a design system let us focus on higher level concerns
Going high fidelity sooner. Usually, I like to invest in high visual fidelity only after thoroughly vetting low and medium fidelity ideas. In this case, I produced high fidelity mockups aligned with the Carbon Design System after about a week, as discussed. I then iterated 30 times, with team playbacks about every 3 days during the roughly 90 days it took us to go from kickoff to delivery. However, the iterations were fine-tuning because of the initial investment in mental models and analysis before going high fidelity It's not for every project, but going to high fidelity right away did a few things:
Human outcome
Using the Return to Workplace Advisor product, large employers were able to monitor and manage changing conditions with analytics to: identify company locations were re-entry conditions had been met; surface emerging trends and search for national and local patterns, and predict COVID-19 projections for the next two weeks. (Description paraphrased from IBM marketing page).
Our team also designed a companion employee web application, for which I provided visual design support and a sounding board for process flows and responsive design.
Employees were able to access policies, resources, and updates from a central hub from their phone, tablet, laptop, or desktop computer. They could self-report symptoms, enabling employers to assess their health status and communicate individual re-entry status and steps for accessing testing and medical care.
Responsive employee web app design from marketing materials
Business outcome
Just six weeks after the Return to Workplace Advisor product kickoff, we were able to debut this product to multiple clients for pilot programs with more than 40K unique users and 20K daily unique users providing health status data for aggregation in the command center. Ultimately, the project has earned millions of dollars in revenue. The business decided to not pursue our designs to tightly integrate multiple products into a Watson Works command center, but has offered a more loosely integrated suite.
IBM Return to Work Advisor (RTWA) has become IBM Workplace Health Advisor (WHA) and is part of the Watson Works suite.