Grand Erie Client project + paid implementation · 2025
Finding help shouldn’t be the hard part.
UX researchProduct design · Product analytics
Researching and redesigning how small business owners find local support, from card sorting to a live resource toolkit.
A small business owner often has to be their own HR department. Grand Erie’s resources could help, but finding the right page meant navigating unfamiliar labels and nested menus. Our work began with local businesses and continued through prototype testing and paid implementation.
Original study outputs · 2025
One route got clearer.
Find the job board. Same task, two navigation trees.
B / Existing navigationThe job board sat below Grand Erie Jobs → Find Jobs. Reported success: 69%; mean time: 11.39 seconds.
Existing: the job board was nested under Grand Erie Jobs → Find Jobs. Small, separate test groups. The +31 percentage-point difference is descriptive, not a live-site uplift.
The finding we couldn’t ignore
The toolkit was still hard to find.
The job-board route became easier to find in the revised tree. The HR Toolkit still performed poorly. An overall success story would have hidden the problem we most needed to solve.
The unresolved route / HR ToolkitPeople still took wrong turns toward Our Team and Employer Tools. The toolkit’s entry point and language needed another iteration.
The analysis / Four tasks
Look at the whole result.
Compare the existing and revised navigation. The job board improved in the test; finding the toolkit remained a problem.
Reported percentage that found the correct destination. Higher is better.
ExistingRevised
Task success: existing vs revised navigation
Task
Existing
Revised
Career pathways
23%
50%
HR Toolkit
8%
13%
Job-search guidance
0%
25%
Job board
69%
100%
Source: WPBGE navigation restructure report, 2025. Small, separate test groups; reported rates are rounded. Descriptive findings, not a statistically significant effect or a production benchmark.
The work behind the interface
My contribution
I worked on user research, card sorting, tree testing, and interpreting the findings with the team. We translated those findings into navigation and prototype decisions. After the course project, I joined Alex Capolongo and Joshua Yang as a paid UX team to help implement the toolkit. The research and website were collaborative; the study results below describe the team’s work.
Evidence → decision → interface
How the research changed the work.
A reconstructed process map from the project reports. Open each step for the evidence and decision.
11 of 24 starts reached completion; median 7:14. Category agreement, a similarity matrix and dendrograms helped the team interpret how business owners grouped resources.
Design implication Use the grouping evidence to refine categories, while treating abandonment as a limitation.
02Test the routesFour tasks → three measures+
Success rose in the revised tree, but the HR Toolkit took longer: 11.25s → 18.98s. Job-search guidance also took longer. Directness alone did not tell us whether someone reached the correct answer.
Design implication Inspect each task and its wrong turns rather than collapsing the study into one improvement score.
03Make the action clearUsability sessions → prototype changes+
Buried funding resources, unclear labels and images mistaken for buttons pointed to concrete interface problems.
Design implication Refine labels, reduce nesting, strengthen entry points and make card interactions consistent.
04Bring it to the siteDesign specifications → WordPress / Divi+
Following the class project, Fahad joined Alex Capolongo and Joshua Yang as a paid UX implementation team. The Business Resource Toolkit launched in 2025.
Design implication Carry the research into implementation constraints, then identify what still needs testing.
What shaped the work
Let business owners shape the categories.
Contextual inquiry and affinity synthesis informed a hybrid card sort. The report records 11 completions from 24 starts, with a median completion time of 7 minutes 14 seconds. We examined category agreement, a similarity matrix, and dendrograms to understand where people grouped resources together.
Why this decision
The categories needed to reflect how business owners look for help. Inconsistent grouping and unfinished sessions were useful signals, not results to hide.
Compare task success, paths, and time together.
We compared existing and revised navigation across four tree-test tasks. Job-board success was reported as 69% in the existing tree and 100% in the revised tree. HR Toolkit success remained low: 8% and 13%. We also examined directness and completion time rather than treating one rate as the whole result.
Why this decision
These small, separate test groups offered descriptive evidence for iteration. They did not establish a statistically significant change or a live-site conversion improvement.
Connect the analysis to the interface.
Usability sessions exposed buried funding resources, unclear toolkit labels, and images that looked clickable but were not. The design work focused on clearer categories, less nesting, stronger entry points, and consistent card interactions. Implementation had to fit WordPress and Divi.
Why this decision
The analysis mattered when it changed the route someone could take and the action they could recognize.
Outcome & limits
From research to a live toolkit.
The Business Resource Toolkit launched in 2025. Laurier’s published account documents my involvement in card sorting, prototype testing with local businesses, and the paid implementation team. The research contributed to decisions about labels, navigation and resource discovery; the experimental task rates are reported separately from the production launch.
The tree tests used small, separate groups with uneven sample sizes. Abandonment and instruction confusion were noted in the report. No statistical significance, causal effect, or production conversion uplift is claimed. This case uses aggregate findings; participant records and full private reports are not published. Selected original aggregate screenshots accompany the findings.
Existing → revised; HR Toolkit and job-guidance tasks were slower
Card-sort completion
11 of 24
Median completion time 7:14; study participation, not product conversion
Proposed · What I’d explore next
See where the search for help breaks down.
Instrument resource-finding tasks, then use a staff-reviewed maintenance assistant to flag stale links and unclear labels. This would extend the research into ongoing product measurement.
Consented task events
SQL / pandas analysis
Staff-reviewed content fixes
How I’d evaluate it
Measure task completion and dead ends; search reformulations; stale links over time. Retest changed labels with business owners before treating an improvement as established.
Engineering approach & boundaries
Proposed technology: minimized, consented event data analysed in SQL / pandas, plus the WordPress REST API for a draft-only content audit. Keep personal search content out of analytics. Staff approve every suggested edit. This is future work; the original project used descriptive research analysis and WordPress / Divi, not this pipeline.