A 10-week research-led IA redesign for a Toronto sustainability nonprofit, led as primary researcher and designer within a four-person team.
Impact
~30% → 100%
Average task success across the three hardest tasks
3:40 → 0:38
Event registration time
2.3/5 → 4.6/5
Donation trust rating
Validation results from prototype testing against the same tasks that failed on the live site.
Challenge
OurGreenway had grown citywide. Its website still told a neighborhood-sized story. Programs were buried, Research had no home, and donations opened an unexplained York University page. People weren't struggling to read the site, but there was friction.
My Role
I led the research strategy and IA redesign as the primary researcher and designer within a four-person team, working directly with OurGreenway's CEO, Treasurer, and Operations Coordinator. I ran six research activities, including the open card sort and independently recruited usability testing, and delivered the accessibility audit.
The Solution
Problem
More than ten reports and publications were buried inside Our Work, listed individually with no browsing controls, and the Publications link in the navbar was broken outright, a critical-severity finding in the content and accessibility audit.
The Donate button redirected to a York University page with no explanation. Two of six usability participants considered leaving at that point, and one did not finish the task.
When I go to the donation page, I get redirected to York University, and it makes me feel like it's a phishing website.
Stakeholders confirmed that research was a major part of the organization's work, but not all of it was being published. The existing structure, listing everything inside a single dropdown, wasn't built to show the full scope of that work. It had nowhere to live in the navigation, and nowhere for a visitor to look.
Research
I ran each method because it could answer something the others could not: what existed, what was broken, what the organization could change, and where visitors still got stuck.
Benchmarked against Evergreen and WindReach, two Canadian nonprofits with comparable multi-program, multi-audience sites. Evergreen's top-level research hub and site-wide search became the clearest reference for consolidating OurGreenway's scattered content.
ScreamingFrog technical crawl plus manual WCAG 2.1 review. Nineteen issues surfaced across four priority levels: 3 critical, 4 high, 5 medium, 7 low, and shaped which fixes had to ship with the redesign.
Full map of navigation hierarchy, click depth, and labeling consistency. Programs took two clicks to reach versus one for Contact, a concrete signal that a high-intent action was structurally buried.
Teams calls with the Treasurer, Operations Coordinator, and CEO, run before writing a single recommendation. This is what reframed the donation redirect from a technical problem into an under-explained partnership.
Eighteen real content cards on UXtweak. The sort produced two clean, almost non-overlapping clusters, but participants could group the programs-and-research cluster without being able to name it.
Six participants were recruited independently to avoid familiarity bias and tested the redesigned prototype across the core journeys.
Insights
Card sort
The open card sort grouped 18 content cards across five participants on UXtweak, giving the team evidence for how people expected the site to be organized. Usability testing then showed Programs and Research represented different visitor intents and needed separate destinations.
Missing destination
Events never appeared as a category in the card sort, not because participants overlooked it, but because it wasn't represented as content to sort in the first place. That absence was itself the finding: Events didn't have an address, so no one could find one.
Constraint
Stakeholder interviews reframed the donation redirect from a technical problem into an under-explained partnership: York University wasn't a bug, it needed disclosure, not a costly infrastructure change.
Information Architecture
Before
After
The card sort established two strong mental clusters. Usability testing then showed where those clusters needed to split: active programs and published research served different intents, while Events needed to be introduced as a destination because the existing content model had omitted it entirely.
RAW CARD SORT OUTPUT — BEFORE USABILITY TESTING SHOWED WHERE IT NEEDED TO SPLIT
Cluster A
Who we are
Cluster B
Programs, research, and impact
Accessibility Audit
3
A broken Publications path, an unexplained donation redirect, and heading plus keyboard failures blocked critical journeys.
19
A ScreamingFrog crawl plus manual WCAG 2.1 review surfaced issues across four priority levels: 3 critical, 4 high, 5 medium, 7 low.
16
After the 3 critical issues were addressed in the final design, the remaining findings were handed to the development team by severity.
Testing
01
Six participants recruited independently to avoid familiarity bias tested the existing experience and the redesigned prototype across the core journeys.
02
I set targets for findability, donation trust, time to core content, and resolution of all three critical WCAG findings before moving into execution.
03
The hi-fi prototype was evaluated against the three hardest tasks, including event registration, with task success, completion time, and donation trust as success measures.
04
The final direction paired measurable task improvements with a WCAG remediation document prioritized for the organization's development team.
Across the three hardest tasks
~30%
100%
Average completion time
3:40
0:38
Post-task self-reported rating
2.3/5
4.6/5
Outcome
The original footer emphasized donation and sign-up actions but gave visitors no way back into the site once they reached the bottom of a page. The redesign mirrors the top-level structure, so Programs, Research, Events, and Get Involved stay reachable from anywhere, including for keyboard and screen-reader users who rely on the footer as a navigation layer.
Before
Visitors reached a collection of calls to action with no clear route back to programs, events, or research.
After
A calmer hierarchy supports discovery, repeat navigation, and keyboard or screen-reader users who rely on footer links.
Stakeholder interviews revealed that York University was not an accidental detour; it was the organization's research partner and donation processor. Instead of recommending a costly infrastructure change, I made the relationship and redirect explicit before people left the site.
Before
Participants landed on York University without enough context; one described the experience as potentially feeling like phishing.
After
The redesign makes that partnership explicit before anyone leaves the site, naming York University, explaining why the redirect happens, and confirming the payment portal is intentional.
Publications had been mixed into Our Work, listed as individual titles, and hidden behind a broken link. The new Research destination brings reports together in one browsable hub with filtering and sorting, and a clear entry point from the main navigation.
After
Reports, field notes, and in-progress research became visible in one browsable destination with filtering and sorting.
Reflection
People who already know a website are not the same as people who need to find their way around it. Recruiting independently, rather than through the client's contact list, protected the research question and exposed failures a familiar audience could easily have worked around.
Setting success criteria before design gave the team something concrete to aim for and gave validation an actual baseline to test against. Stakeholder interviews also kept the donation recommendation grounded in a real organizational constraint rather than an assumption about what the client would accept.
During validation, our client arranged an interview with an 80-year-old community stakeholder, someone the accessibility audit existed to serve, not just describe. The video-call link that worked for every other participant didn't work for her: she needed a phone walkthrough just to join, and once in, struggled with basic controls like muting. Much of the session went to helping her connect rather than gathering insight. That reframed how I read the audit's own WCAG findings: accessibility gaps often aren't abstract compliance line items.
The takeaway I carried forward: good IA gives every important intention an address, then makes that address feel trustworthy enough to enter.
