Jayson Cheung

Knowledge Consultant |
UX Designer

University of Toronto Career Learning Network Redesigned

- User Experience Design -

School Group Project

Project Role

Low-Fidelity Designer (Balsamiq Mockups)
High-Fidelity Designer (Figma)

Current User Experience

The Career Learning Network (CLN) is a job board developed by the University of Toronto for its students and faculty to find jobs suited to their interests. The current interface for the CLN presents difficulties to its users, especially beginners and novices. Our group conducted a contextual inquiry study with 5 participants to observe how users interact with the interface uninterrupted. Users were experiencing obstacles at the beginning of their job search, often unable to locate the jobs module, or were unable to submit a query.

Working Towards An Initial Understanding

We conducted the study with novice and experienced users of the interface. We found that experienced users had a mental model and workflow that that did not match the conceptual model of the tool. Experienced users would use the advanced search tool and submit a search, but omit parameters, to gain access to the jobs listing page. In a subsequent coding of the participant transcripts and affinity diagram, these problems in user workflow emerged as consistent obstacles:

-Hard Time Finding List of All Jobs Available
-Too Many Criteria to Fill Out (Advanced Search)
-Hard Time Narrowing to Relevant Jobs
-Interface has other elements that distract from workflow

In conjunction with the affinity diagram, our group performed a thematic analysis on the coded transcripts of each of the 5 participants in the study. We analyzed the codes of each transcript to interpret the experience of each user at each stage of the workflow. Our group discovered that the users perceived that:

-CLN is difficult to begin to use
-Navigation is slow and hesitant
-Advanced Search is burdensome to use
-Search workflow is not coherent or streamlined
-CLN provides unique value to University of Toronto students

Our group used participatory design and user-centered design to develop an alternative interface for the CLN. We conducted a participatory design study with 6 participants to induce elements that users wish to incorporate or modify about the CLN. Most of our participants emphasized controls for search results filtering and refinement. Each study that was conducted was accompanied with an ethics protocol, consent form, and conducted with voluntary participation from participants.

Our group then proceeded to create a low-fidelity clickable prototype using Balsamiq Mockups 3. This allowed our group to make changes without committment to the design and test our design using focus groups. Our initial ideas attempted to target the obstacles to the user workflow discovered through the affinity diagram and thematic analysis. We attempted to simplify the login landing page by removing local navigation and opting for a global navigation in the form of a hamburger menu. Modules were instead converted into links that were accessible through the global menu or landing/home page. Instead of a separate page for the advanced search criteria input, we changed to an overlay window so that users would interrupt their workflow by re-entering their main search query. Our group also attempted to provide preview information about jobs on the listing page level using tags as the first level of information that a user encounters, along with a job's title and organization.

Refining our Ideas

We conducted a focus group with 3 participants to evaluate user feedback on our initial prototype for the CLN redesign. The biggest issue idenfied by users in this test was the overabundance of secondary information that was presented to participants. Participants identified that there were an excessive amount of tags used to describe a job posting and many of them were not relevant to their specific job search. Participants did not encounter difficulty when interacting with the overlaid search criteria filter. Participants commended that the interface is less cluttered than the current CLN interface, but suggested that more simplicity could reduce cognitive burden even further.

We iterated our prototype according to the feedback received in our first focus group session. Our group created a high-fidelity clickable prototype on Figma and conducted a second focus group with 2 participants. We compared our prototype with the current CLN interface on three tasks:

-Find a full-time job posting related to 'Project Manager' within the 'Community and Social Services' job industry -Find a full-time job posting related to 'Project Manager' within the 'Construction' job industry

For both of our participants, the time needed to complete the tasks was significantly reduced on our prototype versus the current CLN website. Most significantly, we demonstrated that the current filtering menu on the CLN is fragmented, as we intentionally targeted 'job industry', as it required the user to navigate to the separate 'Advanced Search' page to modify. With our prototype, users can easily modify their search without interrupting their workflow to navigate to a separate page, then backtrack to continue their search.

Our group created a third iteration for our prototype to further reduce interruptions in workflow. We adapted a preview window within the jobs listing page so that users are able to access the information within a job posting without navigating away from their search results. A third focus group was conducted with 8 participants to evaluate task completion between our second and third iterations. Participants were tasked with finding information about job postings using both versions of the CLN prototype interface.

Overall, the individual task completion time was lower for participants using the window-view representation (third iteration) of the job listing page than the table-view representation (second iteration). On average, participants using the window-view were 21.75 seconds faster than participants using the table-view. Within our repeated measures, variance was lower within the window-view group than the table-view group. A Shaprio-Wilk test discovered that the window-view group had a p-value of 0.1152 and the table-view group had a p-value of 0.587. However, since our sample is very samll, our group retained a false null hypothesis, as the data we collected was non-parametric (not normally distributed).

Disclaimer: This project was developed for a class assignment with a group of three other students

External Resources Used:
Laptop Mockup by Danielle Buerli

- Project Date -
JAN - APR 2018

Back to Top