Profile Redesign: Case Study
On Leading the Overhaul of Next Big Sound’s Most Popular Report
Overview
In 2014, I led the redesign of the most-used report on Next Big Sound, Profile. Profiles present data-focused summaries for the +1,200,000 entities, like artists and authors, that we track. Its traffic represents nearly 50% of all unique pageviews on Next Big Sound. The redesign contributed to a measurable increase in monthly active users and served a major role in making Next Big Sound “Wired’s 2015 Most Innovative Company in the Music Industry”.
The Problem
A number of forces converged in the summer of 2014 to inspire the team to rally around the redesign of Profile pages. Although Profiles were the most used report for the vast majority of users, we had not made any significant improvements to Profiles for nearly two years, in spite of significant quantitative and qualitative user-focused learnings. Furthermore, Profiles represented major technical debt — after transitioning the rest of Next Big Sound to Angular.js, Profiles were the last report built on Backbone.js. Finally, in 2014, we brought on our first client outside of the music industry, Macmillan Publishing. Profiles would need significant changes to support authors and books as well as artists.
The Audience
Profiles serve many masters. At the time, our business plans challenged us to serve advertising agencies alongside our traditional music clients and our new publishing customers. Within the music industry, Next Big Sound is the primary portal for music data for many different users, from major label executives to undiscovered, independent artists. Our users show a wide range of data literacy, from advanced power-users to total beginners, to even the data suspicious. Profiles are the first report type that new users encounter — and are also critical for label-side weekly reports.
For the redesign, we resolved to orient Profiles toward new users, given our goals to increase monthly active users, engage with advertising agencies, and provide an entry point for users to first engage with data. In this strategy, other Next Big Sound reports, like Graphs and Find, would focus on our expert, enterprise clients.
The Team
The Profile redesign project involved nearly every member of Next Big Sound’s twenty-person team. As lead product designer, I led the research and planned the structure of the new Profiles. As project lead, I worked closely with our visual designer, front-end engineers, back-end engineers, data engineers, data scientists, and our client services team. I coordinated communication, set goals and deadlines, made the tough calls, and generally played the role of a note-taking, gap-filling, troubleshooting cheerleader. In the words of Next Big Sound co-founder David Hoffman, I “set a new bar for what being a project lead at NBS means.”
Constraints
If design, indeed, “depends largely on constraints,” then the Profile redesign had plenty on which to depend. With the addition of Macmillan Publishing as a client, Profiles now needed to support three entirely different entities: artists, authors, and books.
Supporting these three different entities presented huge challenges, especially for information architecture and microcopy. From huge stars with hundreds of tracks and millions of fans on dozens of networks, to obscure artists with only a few hundred followers, each Profile had to address vast differences in the breadth and depth of associated data. To these constraints, our redesign project added concerns about data privacy, contractual obligations to maintain features, and team availability.
Design Process
It all began with research, both qualitative and quantitative. Research for Profile started years before the project itself — I continue to record any and all qualitative user feedback in one easily-referenced and synthesized document. All these learnings gave our project a significant head start.
I interviewed around a dozen stakeholders at Sony and Macmillan — just enough for signal on where our clients’ needs overlapped and divided. I also ran the numbers to identify our most frequent Profile visitors, then interviewed them to learn what they loved and what they needed from Next Big Sound.
Of course, I looked at the numbers themselves, too, to understand how users were using current Profile pages, especially to see which features were most used, like date range selection and demographic measurements, and which features were unused, like the Goals and Events reporting features.
From this research, I developed several use-cases that we used to evaluate design decisions throughout the process of the project. It was a lot easier and faster to align on solutions while assessing problems through the eyes of a small band checking their social statistics every day, advertising agencies comparing different artist partnership candidates, or editors writing weekly round-ups for two dozen authors.
Personally, I love simple tools, like a pen and paper for sketching, or a word document for taking notes and organizing my thoughts. I started thinking about the information architecture of Profiles in a Google Doc, structured around the questions that users ask of Profiles — “How have I been performing lately?” or “Where is my performance strongest, and where is it weakest?’ for example. Focusing on information before layout is critical.
I also love to go really, really wide on initial wireframes and explore many different solutions to design problems. “What if Profile was just a bunch of tables?” “What if Profile featured nothing but auto-magically generated text?” “How many different navigation patterns can I explore?” After generating these wide solutions, I love to toss them in front of the team and a handful of users to help close on what sticks, what doesn’t, and especially, what prompts provocative questions.
After synthesizing this feedback, and refining the wireframes, my role switched from Project Designer to Project Lead for the Profile redesign. More accurately, I was serving in both roles for several weeks. While I continued to refine the wireframes, our visual designer joined the project and added some beautiful visuals to the process. Meanwhile, I coordinated work with our front-end and back-end engineers to validate our plans, identify edge cases, and scope development work. Mostly, I made sure that communication happened, meetings were scheduled, notes taken, deadlines set, and everyone stayed positive and focused.
Finally, while supporting the team building Profiles, I worked with our client services team to develop and execute our launch plan. We ran an informal beta test to uncover issues and refine the page as our work finished. We made sure that our users were fully aware that change was coming, whether through in-app messages or emails. When we flipped the switch, we coordinated social media announcements, reached out to the press, and made sure we were prepared for feedback from our users — which was overwhelmingly positive, thankfully!
Of course, a major project like our Profile redesign is never really finished. We continue to learn from action-focused metrics to understand what’s worked well, and what hasn’t. In 2015, we pushed multiple, incremental improvements to Profiles, based on user feedback, that made the report even stronger and more useful.
Outcome
Simply put, our design project dramatically simplified Profile pages for our users — it’s important to know when to say “no” to new features, after all. From seven tabbed pages, Profiles became one long, scroll-focused page. We removed unused controls and instead presented smart defaults for our users. Left on the cutting room floor after research validated their obsolescence: key metrics, a graphing tool, the events stream, ranking information, similar artists, an event-performance analysis module, and a goal-monitor tool.
While aggressively pruning unused tools, we did say “yes” to a number of other new features. Profiles introduced more data science into the Next Big Sound product than any other project before or since. Based on our user’s constant need for more context, we created Reach, Engagement, Metric Trend, and Artist Stage metrics for every entity, including natural language explanations of those metrics. We facilitated easy one-to-one comparisons between entities, incorporated Wikipedia-based descriptions, and grouped data to mimic our users’ terminology to help orient users learning about new entities. We added more song-focused information and elevated and greatly expanded the available audience demographic information to provide more of the data our users need. New links to preset graphs and on-page source attributions increased navigation to other pages while limiting needless clicks and visits to others. Finally, never discounting the importance of helping our users share data, we built first-class PDF exports for those all-important printed Profiles. A 20% increase in unique pageviews, on a weekly year-to-year comparison to our old Profile pages, speaks to the success of these new features.
Lessons Learned
It’s hard to capture everything that I learned from this massive project. Primarily, I learned so, so much about management and leadership skills — the critical virtues of patience and endurance, the challenges and rewards of taking care of a team, and the value of consistent, clear, and constant communication. As I’ve written, I learned how to be the very best troubleshooting, note-taking cheerleader that I could be! We always have more ideas than time and resources to realize them, and I learned how to gracefully say “No.” It also a powerful experience to be trusted to safely guide my teammates through months of complex, exciting work.
In retrospect, I wish we had created an incremental, learning-and-validation-focused plan for redesigning Profiles. By bottling all our changes for months, then releasing them all at once, we missed many opportunities to learn. For example, the one-to-one comparisons feature, though validated by research, remains underused on Profiles — with early learnings, we could have improved the design, or delayed creating the feature. One critical metric — load times — remain slow on the page. I wish we had created more opportunities to understand the performance impact of putting all our metrics on one page.
Finally, I learned that introducing statistics and data science into the product is both simply not easy and is entirely worth the effort. The data science metrics — Reach, Engagement, Trend and Artist Stage — remain the most-clicked elements on Profiles. Though these metrics took months of work to create, they have since saved years in time for reporting and research in the product. Communicating advanced statistical ideas to most users — even intermediate ones, like standard deviations — is a real challenge that excites me to this day, whether through smart visualizations or natural language annotations.
Interested in seeing more? Want to see our work in action? Profiles are available for all Next Big Sound users — sign up or sign in and check out one of your favorite bands!
