CHI4Good is a day of service which occurs annually before the ACM SIG-CHI Conference. During this day-long event, UX researchers, practitioners, and developers come together to donate their talents to a non-profit organization.

For the 2017 CHI4Good, I volunteered to work with the Out Boulder County group. Out Boulder County educates, advocates, and provides services to Boulder County's Lesbian Gay Bisexual Transgender and Queer (LGBTQ) communities.

Out Boulder County asked us to improve their website so that it was easier to navigate and find information about their programs. They also wanted to find ways to improve the donations and human support which their organization received.

I worked with another UX researcher to improve the information architecture of the site, while several very talented UX researchers, developers and designers worked on improving the overall look, feel, and navigation of the site. We did this through rapid, guerrilla UX research, employing card sorting, reverse-card sorts, expert classification of themes.

Our final deliverable was a new information hierarchy which could be used on the navigation menus of the site to improve the discovery of programs and resources on the site. It also provided improved discoverability of features related to making donations to the organization.

From start to finish, this project took 8 hours with an interdisciplinary team of six. The process we used to develop the information architecture is detailed below.

A quick survey of the site revealed that many of the menus contained a large assortment of items (>10). There were also a large number of menu headings (eight), which made us wonder whether some of these topics could be condensed. Our group had difficulty finding items within each of the subheadings, fueling speculation that the topics of the headings themselves might not be intuitive.

Of primary importance, we found that the "support" button leading to the donation page was difficult to locate and separated from the main navigation area. This separation of the button, along with its red hue (which tended to blend in with the purple background on some computers), could lead to the button being missed.

Once users clicked on the "support" button, we noted that there were several additional layers of links before users could get to a place where they could donate. Donation pages were also crowded with text, making the call to action difficult to see.

With this information, the other UX Researcher and I decided to conduct a card sort in attempt simplify the information architecture of the site to resources easier to find. In particular, we hoped that an improved information architecture would help users find the donation section in a way that also exposed them to the good work and projects of Out Boulder County. We hoped that associating a call to action (donation) with the direct impact of that action would provide a motivating factor for driving donations.

Card Sort

At this point we only had six hours remaining. In most user studies I would never rely on testing with our own staff, but due to our time constraints and our own lack of familiarity of the site, I utilized some of our group members for our first pass at a card sort.

To do the sort, we tore out cards on notebook paper with each of the major topic items on the site. We then had members of our volunteer group sort the items into groups and provide names for each grouping. Our goal here was to obtain a quick intuition of the way that individuals might think of the natural groupings of the categories and the natural language that they used when describing these groupings.

There seemed to be substantial convergence in the groupings and grouping titles that participants chose, which led to some optimism that this might be the case in the population in general. This would simplify our decisions in choosing clusters and grouping titles to improve the information architecture later on.

Guerrilla Card Sort

With about four hours remaining, we pulled in passers-by in the convention center to be subjects in our card sort.

We provided our card sorters with a shuffled deck of cards showing the topic areas and then allowed them to sort them into groups. We then provided them with blank cards in which they could write the tile of these groups.

These sorts were similar in many ways to our original card sorts, but we did notice that participants thought about several categories in ways which we did not anticipate.

For example, a participant noted that "following" the organization on social media like Facebook or Twitter could be seen as a type of social support. He thought that the ability to follow the organization might be clumped into the same category as donations.

This helped to broaden our thinking into the ways in which category elements related to the psychological constructs of social proof and validity might be placed into proximity to areas for donation on the site to further motivate donation behavior.

Expert Reverse-Card Sort

With less than two hours until the end of the event, we had one of the volunteers who worked at Out Boulder County itself conduct an expert reverse card-sort. The card sort was "expert" in the sense that the individual was familiar with the domain area and the users who would be visiting the site.

To conduct the reverse card-sort, we did a quick "rough and ready" evaluation of the most frequently generated groupings in the regular card sort and the titles generated for those categories. Then, the other UX researcher and I looked at the titles that users had given to those groupings and standardized those grouping names through discussion.

We provided these grouping titles in a randomly shuffled stack which were laid out on the table. We asked our expert participant to sort the topic areas contained in another shuffled stack into these categories.

The goal here was to validate that the participant would anticipate certain topic areas within the titles which had been previously generated by users, and that these groupings roughly corresponded to those in the card sort. This would provide a very rough validation that users familiar with the topic areas would be able to find their desired topic quickly within the site.

This reverse sort took roughly 30 minutes, with another 30 minutes of follow-up discussion, which helped us to clarify some of the necessary groupings and topic titles which would be used on our final recommendations.

This is a snapshot of the final categories and subcategories we settled on after the card sort and reverse sort.

Note: The insight from our users that "following" on social media could and should be seen as a form of support did lead us to place this within a broader "Support Us" title. This title also contained the link for donations.

We worked with several talented UX designers in our group to determine how this information architecture could manifest itself within multiple ideas they had for compelling layouts. Our designers then mocked up in several interactive examples in Axure RP and as static wireframes in Sketch. These were provided as deliverables to Out Boulder County as models for potential website implementations for developers.

From start to finish, this project lasted approximately eight hours.