The Dramatic Benefits of a New Information Architecture

Written by Lindsay DeVellis, Senior Strategist

Many clients call Brooklyn United with the same need. They want a better user experience for their audiences, and a website that is easy to navigate.

When users describe a website as difficult to use, the issue often goes much deeper than the visual design. One of the first things we look at, regardless of whether our client is a global brand, institution, or early stage startup, is the website’s underlying information architecture — how the website’s content and pages are organized. A good information architecture (IA) uses scalable hierarchy and familiar nomenclature to make content easy to find. When organizations shift and grow without revisiting their IA, tensions sprout between content and its organization. Pages end up balkanized — created for niche reasons and with little content or connection to the overall site, located in hard-to-find places, saddled with obscure names, and deployed in such numbers they fragment the potential user journey to a point of little return.

This was the case with our client, the Onassis Foundation USA (Onassis), a cultural and educational institution dedicated to promoting Greek culture throughout the Americas. In 2015, they reopened to the public after a 3-year renovation period with a new vision and vibrant purpose. Their free, high-quality programming and events are the lifeblood of the foundation and its key means of connecting with and expanding its audience. Through qualitative and quantitative research, we learned their existing information architecture was opaque to users, hindering users’ ability to discover all the exhibitions, events, and media that Onassis had to offer. We set out to fix this before tackling any of the visual design. Informed by the breadth of Onassis’s programming, user needs, and online engagement, we prioritized the task-based needs of users, and effectively improved navigability by 116%.

Here’s how we did it.

Step 1. Research

As with any task strategists undertake, our process began with research. Onassis had an established audience of Manhattan intellectuals, and was able to provide existing research on this audience going back several years. They also wanted to expand into a younger, more diverse demographic. Thus, we had two different audiences we needed to understand: one already aware of Onassis and their mission, and another that required introduction and education.

We reviewed the audience research Onassis provided us, and spoke to a few individuals who were engaged with the brand. We didn’t have as direct access to their new target audience, so we referenced industry reports and existing market research in comparative spaces. The Tate, for example, has strong documentation online around their digital strategy and audience insights. We also used a modest Facebook advertising budget* to survey audiences with like interests, and reached out to culturally active New Yorkers for interviews.

*Facebook advertising is quite cost-effective and offers niche targeting options; it can be a great way to recruit users if you can find $100 to promote a post or two!

Our research confirmed that digitally savvy, young adults are very goal-oriented when online. Though they may like to kill time on entertainment sites like Facebook and YouTube, when it comes to other tasks, they’re serious information seekers.

They want to accomplish tasks, solve problems, and find answers to questions as quickly as possible.

This insight bolstered our initial strategy for a navigation that prioritized specific tasks, rather than reflecting Onassis’s internal structure.

Step 2. Test Existing IA

Our research evolved from gaining insights around audience motivation to watching specific audience behavior. By testing the existing IA, we’re able to identify what’s working well, and what’s not. It also allowed us to set a benchmark for our work moving forward.

There are a few different ways to test a site’s IA, and we’ve found great success with tree tests, a text-based study that asks individuals what word they’d click on to complete a particular web task. To set up this test, we used site analytics and initial audience surveying to outline priority tasks for the site. By using real audience data to determine the test questions, we ensure our test aligns with user needs.

We ran a tree test with the tasks above to see where users would expect to find information in the existing Onassis site map. The biggest insight we gained through this exercise was this:

Users generally didn’t know where to look first for events.

Any task related to events took users the longest to complete, at about 16 seconds. Pathways of event-related tasks showed that many people clicked through the entire primary navigation before realizing events were listed under the section labeled after the events’ physical location. It was also not always clear to users how Onassis categorized their events and exhibitions. For example, for an event where a famous filmmaker will discuss Ancient Greek literature at the New York Public Library, users selected 5 different pages: “Lectures & Panels,” “Talks,” “Film,” “Special Events,” and “Sponsored Events,” based on what they thought was most identifying about the event.

This identified a core issue with having a different landing page for each event type.

Overall, participants completed the above 8 tasks at an average success rate of 30%: plenty of room to improve.

Step 3. Card Sort

Our next step was to see how the target audience would group and categorize various content types on their own. To do this, we ran another IA activity known as a card sort. The card sort was comprised of 50–60 different content types, each representing a single “card.” We asked test participants to sort these cards into groups that made the most sense to them. They then labeled each group, illustrating their individual instincts for how information might be organized.

The results of a card sort can be parsed in many ways. We looked at instances where over 50% of the participants agreed on card pairings, and consolidated those pairings into 10 content groups. These groups became the basis for creating a new site map tuned to audience expectations about content.

Among other helpful insights, the card sort revealed a simple solution to one of the IA’s challenges: how to distinguish the Onassis Foundation USA from its parent foundation in Greece, the Onassis Foundation. Shining light on this relationship was important for audiences to understand the brand’s history and esteem, but proved problematic in the existing IA.

Through the card sort, we learned that 70% of participants placed the Onassis Foundation and Onassis Foundation USA content cards in the same group, providing justification to place the brand information together on a single page. This opened up a lot of room in the primary navigation for us to provide a pathway to priority tasks, like finding events, exhibits, and visitor information.

Step 4. Create New Site Map

Armed with user data, we were finally ready to create a new version of the Onassis site map. This required translating the content-oriented groups above into formal pages.

We also used the insights from our tree test to supplement the card sort groups. For the new site map, we transitioned from organizing the IA around Onassis’ different brands, to organizing the IA around similar content types:

  • We oriented the primary level of navigation around priority user tasks
  • We placed all information about the Foundation’s hierarchy and different brands in a single place, under “About”
  • We made sure events could be found in a single place, rather than being segmented into different pages based on event type
  • We chose language that was used by Onassis’s target audience, rather than language used internally by the organization

Step 5. Test, Revise, and Test Again

We put 2 variations of a new site map to the test to validate our work. We repeated the initial tree test and used the same 8 tasks, in addition to 2 new tasks. Our objective was to find out if audiences had an easier time locating specific programs, if each program had its own place in the primary navigation.

By A/B testing the site maps in this way, we were able to clearly define a benchmark and surpass it. When Onassis came to us, their IA performed at a success rate of 30%. Both new versions of the site map easily doubled user success rates compared to the old.

The top-performing IA had a success rate of 65%.

This second test showed much improvement, and validated the majority of our work around consolidating programming under common nomenclature. We also found additional areas for improvement. For example, though the research objective was not aimed towards the “Multimedia” section, we also learned that “Multimedia” was a more successful and straightforward page title than “Learn Online,” which had been used in previous testing rounds.

In Conclusion

Before we restructured Onassis’ information architecture, traffic to the home page and “About” page made up over half of all pageviews. Users were spending more time reading about the organization to develop a cursory understanding of their offering, rather than engaging with their content and finding things to do — activities more likely to build a meaningful connection between the foundation and their audiences. Now, consumption of content is much more diverse, with Onassis’s spring exhibition accounting for the most pageviews at 17%, “Visit” on par with the home page at 10%, and so on.

Restructuring Onassis’ IA was instrumental in enabling users to explore their content and engage with their brand. If you’re looking to increase website engagement and create an all-around better user experience, you should always look critically at your information architecture first. The most beautiful design and the most powerful content aren’t worth much if people can’t find them.