The first time I knew I made the right decision to attend General Assembly was five days before my UXDI course started. I had been riding a rollercoaster of emotions the whole month leading up to this moment, whipping wildly between the high of starting something new and the low of leaving an industry I called home for seven years. In the general anxiety of a major life change, I suppose I was looking for a sign I was doing the right thing. I found it in the welcome email from our instructors:
“It is our pleasure to welcome you to UXDI October 2018 (aka UXDI Gordian)!”
Seeing those words, my myth-loving heart thrilled. If my teachers valued the metaphor of the Gordian Knot (and meaningful metaphors in general), they were my kind of people, and I was going to be OK.
Like many Greek tales, the story of the Gordian Knot starts with a legendary task: unraveling an impossible knot. Many had tried and all had failed, but eventually, news of the Gordian Knot reached Alexander the Great. An oracle prophesied that anyone who could undo the Knot was destined to become the ruler of all Asia, so Alexander decided he needed to succeed.
When Alexander the Great attempted to untangle the Knot, he failed the same way people before him did. But then he did something different. He stepped back from the problem, drew his sword, and cut clean through the Knot. With a new perspective and a little creativity, the Gordian Knot was solved.
I’ve leaned on this story throughout the course (despite its problematic justification for conquering Asia), and I certainly needed it for P3.
CLIMBING THE MOUNTAIN
GA’s third project (P3) is commonly referred to as “the Mountain.” It comes at the midpoint of the course, arriving too quickly after two stressful, strenuous projects. Sleep-deprived and slightly delirious, we kicked off our lessons and started climbing.
The focus of P3 is Information Architecture (IA), which Wikipedia says is “the art and science of organizing and labeling websites…to support usability and findability.” To test our IA skills, we would be examining the St. Jude Children’s Research Hospital website as a case study and evaluating it using several different methods.
The deliverables for our previous projects had been assigned in a linear fashion, so each new part of the project built on what we had just completed. But the deliverables for P3—a current sitemap, a revised sitemap, a task flow, a user flow, a competitive/comparative feature analysis, a closed card sort, and an open card sort—were all due together, and as we’d soon find out, they were horribly entangled with each other.
EXISTING SITE MAP
I made a plan of action to work through the deliverables, starting with the existing site map. It felt like a solid foundation, as the research I’d do for it would inform how to tackle the rest of the evaluations.
Site maps have a logical structure based on hierarchy. Like family trees, they show the relationship of pages to each other, “parent” to “child.” For me, they also brought back pleasant memories of biology and taxonomy, where Latin names told a story of discovery and relationships between species.
But the St. Jude site was a much different beast than the straightforward e-commerce site we’d mapped in class. The website was complex and sprawling. Two sections were completely entwined together (“Get Involved” and “Ways to Give”) in a way that didn’t make sense to me. It was difficult to get a clear sense of hierarchy from the navigation, which meant I had to map the site by diving into the interior pages and checking out the URLs.
I felt shaky about my foundation, so it was hard to move forward with the next part of the project. But I knew that, with our one-week deadline, getting stuck for too long on any piece of the project meant failure. I pushed on out of necessity, resolving to double back when I could.
COMPETITIVE/COMPARATIVE FEATURE ANALYSIS
After the snarl of the site map, it was a relief to work on the competitive/comparative feature analysis. The answers came more easily, as I was somewhat prepared by my previous sales and marketing experience.
I picked three competitors for St. Jude fairly quickly, selecting non-profit research hospitals with a national reach (two of which specifically treat children): Shriners Hospitals for Children, Mayo Clinic, and Children’s National Health System.
The comparator was harder to pin down. I couldn’t figure out an indirect competitor to a research hospital, so I looked at the other piece of St. Jude’s business model: their fundraising. Then it hit me with the force of a Sarah McLachlan song played over a sad commercial: the ASPCA.
Once I had my competitors nailed down, I picked out ten site features, laid out a comparison chart, and finalized this deliverable with confidence.
Fresh off of my feature analysis, I was excited for my task and user flows.
Flows appealed to younger version of me—the kid who made complex legends for her treasure maps and obsessed over flowchart quizzes in teen magazines. I also loved their sequential, detailed logic, which felt like the adult version of the PB&J test that rocked my world in elementary school.
Having spent plenty of time investigating the St. Jude site (and the rabbit holes I could go down to find information), I picked a task that was relatively straightforward: “Find resources to help you plan a fundraiser at your office for your coworkers”
Unfortunately, “relatively straightforward” on the St. Jude site meant a roundabout path for resources, heading off the site and into my Gmail account before being directed back to the site. The user flow got even more interesting, as there were several ways to get to the form that was essential to obtaining the fundraising information.
With my flows finished, I was still puzzled by the overlap between the “Get Involved” and “Ways to Give” sections. I hoped card sorting could illuminate whether this jumble was as big a problem for other users as it seemed to me.
CLOSED AND OPEN CARD SORTING
Card sorting was the deliverable I was most worried about, though it turned out to be the most enlightening and the most fascinating.
When I began prepping my sort, I was skeptical that five different people, all with their own mental models, could give me usable trends. But this step definitely turned out to be one of those “trust the process” moments.
Picking representative cards for such a large website was a little tough, but I stuck to about thirty cards. For the closed sort, users would place the cards the way they’d expect to find them under the six existing primary navigation categories. For the open sort, users would group the cards and name their new categories themselves.
Since I didn’t have ready access to medical professionals for my card sorts, I picked people who had jobs that involve categorization, prioritization, and looking for themes (editors, literary agents, and a product manager). For the open sort, I prioritized people who were decisive and weren’t rushed for time, as I thought that open sorting could be more stressful than closed sorting.
Once I completed all six sorts, I synthesized them into a large Excel chart. The bright red areas in both sorts (where all three users placed the cards incorrectly) were ones I wanted to pay attention to.
As displayed by the color-coding, the closed sort users mostly matched the cards under About Me, Care & Treatment, Research, and Training to their original places on the St. Jude site. Open sort users categorized their cards similarly to the existing site, with two major changes: enmeshing “Get Involved” with “Ways to Give” and eliminating “Research” completely.
REVISED SITE MAP
Card sorting had given me data for my revised site map, but now I had to decide what to do with that data.
Based on the closed sort success rate of users matching cards to the original categories, I kept the existing primary navigation largely the same. Still, I knew I had to address the “Get Involved” and “Ways to Give” sections.
Since users strongly intertwined “Get Involved” and “Ways to Give” in both the closed and open sort, I combined them in my revised site map. I took the name for the new category (“How to Help”) from one of my users. I also accepted my other open users’ subcategories of “Donations,” “Fundraisers,” and “Other Ways to Get Involved,” as they all made distinctions between ways to give.
I also chose to move “Affiliate Clinics” from “Care and Treatment” to the “About Us” section, as all six users miscategorized it.
One change I decided not to make was to the “Research” section, which was not represented in any user-made open sort categories. However, none of my card sort users were in medical or scientific fields, which is the segment of users the Research section serves. I think asking users in the medical field to name categories would have yielded different results, so in the absence of real data, I left the category as is.
UNTYING THE KNOT
After I turned in all of my deliverables, I was both relieved and elated to have climbed the P3 Mountain. (I had emerged mostly unscathed, aside from a slight St. Jude website Stockholm Syndrome.) I learned so much about the art and science of IA, and I surprised myself with how much I could accomplish using the marriage of both my soft and hard skills.
While P3 was a tangled mess at times, I kept the Gordian Knot mindset with me: remembering to step back from the immediate problem and look at something seemingly impossible from a different perspective. Though I have so much more to learn (and I wouldn’t trust myself with a sword), I now know that I can use patience, creative thinking, and the right UX tools to unravel even the knottiest problems.