When the American Mathematical Society redesigned its website, it knew which communities it wanted to reach but wasn’t sure what content would resonate with them. By combining tech savvy and in-person research, AMS built a new site that grows with its members. (Titled “A People-Driven Web Redesign” in the print edition.)
How do users fare on your website? Do they read and navigate with ease—finding answers, making discoveries, clicking everywhere you hope they will click? Probably not: According to Forrester Research, 97 percent of websites fail their users. And most fail in similar ways. Since 2005, the top failures have included basic problems like legibility, use of space, and task flow.
In most cases, association staff cannot determine audience needs internally. So when we at the American Mathematical Society decided to pursue a website redesign, we set out to learn about and correct the mistakes we made on our existing site and discover more about the users we serve.
AMS was an internet early adopter: We launched a Telnet service in 1990 and debuted our website, ams.org, in 1996. In those pioneer days, things grew quickly and organically. Yes, there was structure and navigation, and standard content regions such as headers and footers were evolving. But web development was all very seat-of-the-pants.
Flash forward a decade later. Our website had databases for dynamic content delivery and about 45,000 HTML pages, as well as images, videos, PDFs, Java applets, and MP3s. The site had become immense and unwieldy, and our users were struggling to find relevant information in the ocean of content. In an effort to communicate everything to everyone, we had created a website that met many of our members' information needs in the aggregate. But when it came to communicating effectively with specialized groups, we were at a loss. We needed another solution.
Personas: Take One
In 2009, AMS undertook an intense website redevelopment project scheduled to play out in three phases. First, we'd make cosmetic fixes to the site's look and feel and make minor tweaks to site navigation. Then we would address necessary changes to the underlying information architecture and content management. Finally, we'd reach out to AMS members and other stakeholders in the mathematics and scientific community to fine tune content delivery for our diverse constituencies.
Steps one and two were addressed incrementally over two years. Step three turned out to be a bigger challenge, so we brought in a consultant to assist in our rebuilding efforts. With guidance from AMS staff and executives, our consultant quickly determined we were a very "inwardly focused" organization—that is, we relied heavily on our internal notions of what the community needed.
We tended to focus on information silos. For instance, we assumed that people looking for books on research mathematics would probably not be interested in mathematics for the general public. While we had built an adequate navigational structure within these silos, we rarely focused on cross-pollination among them. We also lacked the facility to communicate with special segments of our user community, such as the librarians who are so important to our publishing program.
In response, we decided to employ personas: archetypal representations of user groups. A team including AMS staff from the executive director to customer service to volunteer leadership developed seven personas to represent the AMS user community: authors, faculty, librarians, media, mathematical reviewers, researchers, and students. We then conducted focus groups to confirm our choices of personas. Finally, we selected page content for each persona. In the case of librarians, for instance, many of the links were undoubtedly useful for them, but there was still a problem: Content was chosen without external input.
Personas help web developers understand the user group and present users with information customized to their needs and preferences. When you miss the mark, though, it is painfully obvious. Reexamining our librarian persona using Google Analytics, we learned that the bulk of information we had selected for the librarian community was not being accessed at all. In fact, the few links that were used were those that led the reader away from the content we'd placed there and to other areas of the site.
Clearly, it was time to rethink our approach.
Personas: Take Two
Analytics proved that something was missing in our first attempt at personas. But the data did not give us adequate reason to abandon the concept altogether.
Millions of people from all over the world access our website, sorting through the rich content with a variety of attitudes and goals. Personas should help segments of our audience quickly find the information most pertinent to their needs—they're a highly recommended approach to website design. So where did we go wrong?
In late 2011, we began a new outreach project targeting the librarian segment of our website users. Data gathered from a short email survey told us that our page links were reasonably targeted to this user group, but no one was clicking on them. We used the page to provide information that we wanted them to see, and rightly so. But more important, what did they want?
We knew math librarians looked for information like journal pricing and frequented the AMS online bookstore and research tools like MathSciNet. But we didn't know much about their preferences and habits. Who are they, really? We needed a fresh approach to outreach, one that would help us move past the guesswork and put human faces to our website users.
We started with a single librarian, simply because the opportunity for an interview arose at a mathematics conference at Cornell University. That first in-person conversation helped shape a new user survey and future conversations. Sitting in the head math librarian's office, we discussed how math librarians assist their patrons, how they access information on ams.org, and what tools might make librarians' jobs easier. We looked together at the page for librarians and navigated the website through her eyes.
Hearing from just one audience member naturally informed approaching the group as a whole. Treating users as collaborators from the start changed the picture entirely.
After this first illuminating conversation, we set out to approach the math librarian community to better tailor our website to their needs. Through this process, we learned a host of lessons that other organizations can apply to their own efforts to understand what members are looking for. These are a few of them:
Join members' community. Even time-crunched, resource-starved web content developers can join user communities for free. In our case, the websites and corresponding email listservers for the Special Libraries Association (SLA) and American Libraries Association provided free access to a wide group of potential users. These groups were gracious enough to allow us to send out a survey to their lists. (We were unsure of the etiquette, so we asked permission first. It never hurts.)
When joining a user community, remember that you are entering a multidimensional conversation. Every group is different. The community we joined is active: Its members help each other find rare publications; they express heartfelt well wishes upon retirement; they congratulate each other on success with students. And they do all this as virtual colleagues. We would never have known these personal details from analytics or survey data. Observing attitudes and personalities proved useful in assessing how to present content not for just for a user group but for a group of people.
Determine member preferences. Ask them: What are you looking for? How do you find what you're looking for? And what have you not found? Assume nothing. Some behaviors are guided by personal preference; individuals develop habits that web design can accommodate but cannot precisely anticipate.
For example, librarians we surveyed were split between bookmarking their frequently visited pages and navigating from the AMS homepage. And general web-usage data indicated that many task-driven users either prefer or resort to a search engine like Google or a website's internal search tools. To accomplish specific tasks, users generally choose one of these options according to preference and website usability. They navigate or search to the content they need to accomplish certain tasks.
Still, while users tend to be task-driven, serendipity plays a key role in web activity. Librarians found both task-oriented resources and unexpected items useful. It was clear that our persona page needed to present a range of dynamic content, even if the most task-oriented links were already bookmarked by frequent users.
Engage in active listening. Don't just ask questions; engage in conversation. Some librarians replied to our survey with questions that we could never have anticipated, that we didn't know the answers to, and that we didn't know were relevant. Answering them significantly informed web content. Much of the content we now offer on our "Information for Librarians" page resulted from entirely unanticipated survey responses. Following up with individual respondents helped clarify many of their points and fostered an ongoing give and take.
Since the initial survey, we've received occasional unsolicited suggestions and questions from librarians. Not only have these users helped us build a persona page geared toward their information needs, but they have also helped us become a more engaged and responsive resource provider—a human face—for ams.org.
Conduct multifaceted outreach. Whenever possible, give users multiple ways to connect with you. Some seemingly disinterested users of our website eventually provided generous feedback over the phone, while others preferred providing written comments through email. Allowing for personal preference facilitated successful communication.
Also look for opportunities to engage in person. Networking doesn't necessarily require a huge time investment. It means incorporating your audience into your usual activities. At a January mathematics conference, we were welcomed into a meeting of librarians. Several attendees remembered the email exchanges on the listservers and invited us to an evening reception. When we were asked, "Will you be attending this summer's SLA conference?" we realized that something new had occurred: We'd become a part of our user community.
Construct a picture of the group's information needs. Through our outreach efforts, we confirmed that our persona page for librarians could be much better tailored to their needs.
First, we assembled a list of specific group feedback: requests for additional page content, likes and dislikes, navigation and browsing habits, social media preferences, and so on. The list provided a glimpse into their daily routines. For example, librarians frequently seek printable material to post on the library bulletin board such as posters, article recommendations, and other dynamic content.
Second, we developed a general profile of the math librarian group. This subjective, observation-based profile left us with a picture of a group made up of people who are communicative, passionate about their work, helpful to each other, and concerned with keeping up with issues and news in their field.
Create a guided experience. After getting to know math librarians personally and processing their feedback, we had a sound understanding of the group's information needs. We now had a big pile of focused content to share with our librarians. The key was to break out of the former organizational scheme and present a guided user experience.
Focusing on potential audience actions, we sorted the page content into categories, offering both quick action items and "serendipity" items—like printable posters and recommended articles—for reading, printing and sharing. Serendipity items would be dynamic (perhaps changing as often as quarterly), creating a place worth returning to. Finally, we added a "librarian highlight" section where we would regularly profile math librarians and their home institutions. We hoped the page would grow from a flat list of links to a community-oriented home base.
Success and Beyond
In February 2012, we released the librarian persona page (www.ams.org/publications/librarian) with revised content, visual resources, article suggestions, and a librarian highlight. Unique views to the page are up 13 percent, and users spend more time there than they did a year ago. We've been hearing good responses from the community. We can confidently report this project as a success.
But the process can't stop there. In addition to continued evaluation using Google Analytics, we'll regularly seek audience feedback and promote dialogue between our organization and audience. The web is a living, evolving entity, and the persona must be also. Over the next year, we'll continue developing the six remaining persona pages using the same process.
While our process worked well with librarians and is easily applicable to other situations, all user groups are different. The librarians we approached were exceptionally communicative and saw themselves as collaborators in the project. Our advice? Focus on each user group's key differences and personality traits early on, work with their unique habits and preferences, and adapt to their needs and attitudes.
Our users-turned-collaborators now have content tailored to their needs as people, and we have an approach to personas that we can implement across our website. As a result, the ams.org user experience will continue to get better.