Case Study

Simplifying large-scale content navigation

Organizing navigation around how people learn

I led the navigation redesign for the O’Reilly learning platform, working with designer Shirley Jin to understand why learners struggled to find relevant content. Our research informed a structure organized around topics and learning goals, with most of our recommendations reaching the final prototype and the first phases of implementation.

The challenge was to simplify access to a large catalog while supporting different reasons for learning: exploring a subject, solving a technical problem, or practicing a skill.

The original sidebar mixed learning resources, account settings, and support links, making the navigation a focal point for improving content discovery.

The catalog reflected platform categories more than learner intent

Customer feedback and usability studies showed learners losing their way to content. Conversations with Customer Support reinforced that the existing navigation was not meeting their needs. Shirley and I reviewed this evidence together and mapped the structure to understand where it broke down.

Mapping the existing information architecture exposed a fragmented structure that emphasized content formats and platform features over the topics learners wanted to explore.

Our working hypothesis was that consolidating navigation around learners’ goals would make the catalog easier to navigate. Reducing the number of links alone would not be enough; we needed to understand which distinctions mattered to learners and which reflected the platform’s organization.

Testing where labels and expectations diverged

We used tree testing to benchmark the existing structure and card sorting to investigate how participants grouped its content. Our target audience was mid-career software engineers who used online learning platforms. Shirley and I shared study planning, execution, analysis, and documentation.

The findings pointed to four requirements for the new structure:

Support topic-first discovery. Most users began with a subject, while the existing categories emphasized formats such as books, videos, and interactive content.
Keep routes for specific learning goals. Someone looking to experiment with cloud technology might seek a hands-on format directly.
Use labels that explain the destination. Terms such as “katas” and “scenarios” made it difficult to anticipate the content behind them.
Separate account administration from learning content. Participants’ groupings supported a clearer boundary between those tasks.

Our tree-testing study established a baseline for how participants found learning resources and completed common tasks within the existing navigation. View the tree-testing report.

Finding assigned learning materials was not straightforward: participants reached Assignments with a 51% success rate, but only a 24% directness score.

Participants took 28 different paths to find a saved book. Many chose Workbooks, suggesting that the label competed with the places where saved content actually lived.

Most participants found All Events, but the low directness score showed that the route through Live Events was less obvious than the destination itself.

An open-ended technical problem revealed several competing paths, with participants turning to Answers, Resource Centers, and Scenarios to look for help.

Only about a third of participants reached Sandboxes when asked where to practice code safely, highlighting a gap between learners’ goals and the platform’s labels.

The card-sort analysis surfaced potential groupings across learning resources, helping us compare participants’ mental models with the existing structure. View the card-sorting report.

The similarity matrix showed how often participants grouped items together. Books and Audiobooks, for example, were paired by 61% of participants.

The clustering analysis brought account, preferences, support, and administrative items together, supporting a clearer separation between managing an account and finding learning content.

Supporting several paths without multiplying top-level choices

I used these findings to shape a navigation model with distinct routes for topics, search, and learning activities. The proposed sitemap consolidated content formats under Browse and separated profile settings and administration from discovery.

The proposed sitemap consolidated learning resources under Browse, elevated Topics and Search, and separated profile settings and administration from content discovery.

We defined the redesign around a small set of product decisions:

Keep topic browsing close to search so learners can move between exploration and a specific query.
Replace obscure labels with concrete descriptions of topics, formats, and actions.
Group related formats rather than giving each one a separate top-level entry.
Simplify the structure while preserving access to content and key platform capabilities.

We translated the research into requirements and user stories focused on clear labels, orientation, returning to previous learning, and access to search.

Using limited space to force clearer priorities

We chose a horizontal top bar to give content more page width and constrain the number of primary navigation items. That constraint required us to group capabilities meaningfully instead of giving every feature its own link.

The design paired a Topics menu with search and a Start Learning menu organized around actions. It supported learners who knew the subject they wanted to explore as well as those who knew the kind of activity they wanted to undertake.

The homepage concept replaced the vertical sidebar with a compact horizontal navigation, giving learning content more room while keeping key entry points visible.

The Topics menu offered a subject-first route into the catalog, helping learners begin with what they wanted to learn rather than a particular content format.

The search dropdown paired direct search with shortcuts to content formats, supporting learners who knew what they wanted and those who preferred to browse.

Start Learning grouped resources around actions such as attending an event, practicing code, and preparing for certification. Consolidating these options helped accommodate teams’ visibility needs without adding more top-level links.

The trade-off was visibility: teams wanted their capabilities represented in the primary navigation. Grouping related options preserved routes to those capabilities while limiting the number of choices competing for attention at the top level.

Balancing discovery with existing customer and technical needs

Product Marketing was hesitant to remove format-based landing pages because it used those pages to direct customers to content. This made the redesign a cross-functional decision about access to the catalog, beyond the navigation layout itself.

Product Design Director James Touhey, VP Julie Baron, and Senior VP of Product Adam Witwer helped align expectations across Marketing, Editorial, and Customer Support. Their involvement helped us work through resistance to changes in the platform’s information architecture.

Engineering also raised concerns about whether the existing implementation stack could support the changes.

Moving the proposed structure into early implementation

The direction received support from executives, Product, Marketing, Customer Support, and Sales. Most of our information architecture recommendations made it into the final prototype and the first phases of implementation.

The refined navigation brought Topics and Start Learning together beside Search and restored the full O’Reilly wordmark for clearer brand recognition.

The documented outcome is a revised navigation model and progress into implementation. This case study does not include post-launch evidence of faster discovery or increased engagement, so those remain intended benefits rather than measured results.

Explore the click-through prototype.

Next Case