Ellucian · 2024
The site was organized like the org chart. Buyers aren't.
Turning buyer intent and behavioral data into a new information architecture for a large higher-ed software portfolio, then testing it before a trade-show deadline.
- My role
- UX strategy lead. Research synthesis, IA and taxonomy, wireframes, prototype testing and conversion planning.
- Team
- The brand and visual design team carried the campaign into the pages; engineering built in parallel.
- Constraint
- A fixed trade-show launch meant visual design and development had to run alongside the IA work.
- Result
- +12% demo-start intent (design-stage estimate)
Organize by institutional outcome and buyer intent, with several valid paths, instead of by product line.
- +12%Demo-start intentDesign-stage estimate
- 2.3×Users finding product content within two clicksDesign-stage estimate
- −18%Homepage bounce rateDesign-stage estimate
Design-stage estimates, not reported post-launch results.
Evidence
Ellucian had strong products, useful resources and real proof. It was also fragmented: presidents, CIOs, enrollment, finance and HR leaders arrived with different jobs and had to assemble the story themselves. Labels mirrored internal structure and important pages sat too deep.
I synthesized Google Analytics, Crazy Egg heatmaps, client research, and form and download data to see where people entered, stalled and converted. Then we audited the category page by page to find where everyone sounded the same.
Strategic choice
Overlaying our traffic and conversion data on the twelve recurring page types showed what people actually did. The taxonomy was built on that, not on the product catalog.
Enterprise buyers compare, check security and proof, share links and come back on another device. So the IA offered paths for direct buyers, explorers, researchers and implementation teams. Wherever someone landed, they could tell where they were.
Testing and tradeoffs
Moderated sessions tested confidence, language and credibility; automated tests checked task completion and path clarity. People liked the student-centered story but wanted to know sooner where they fit. We tightened labels, moved proof up, grouped products more clearly and matched calls to action to intent.
The content model that came out of it is simple: top pages frame the problem, mid-level pages explain the ecosystem, product pages give specifics, resources build confidence.
Execution
Running design and development in parallel only held because we stayed system-first: reusable modules and tokens, documented handoff, accessibility checks and regression review. UAT started with 18 participants and later expanded to a 19K-subject sample.
With the taxonomy settled, we wireframed every template the IA needed, desktop and mobile, as reusable modules. The brand and visual design team carried the campaign into those pages. Lottie microanimations, from the visual team, explained platform connectivity and data movement where static copy struggled.
Limits
- The metrics above are projections from testing, not measured post-launch results.
- Visual design and campaign art belong to the brand team.