I redesigned CNET’s AI Atlas to help readers find and spend more time with the coverage, while making the page more valuable to sponsors within our engineering capacity. Jump-link usage rose from 6% to 20%, and two sponsorships were secured before launch.
Overview
AI Atlas brings CNET’s AI coverage together in one place. It was attracting visitors, but readers weren’t exploring much beyond the main article feeds. That limited the experience for readers and the value of the page to sponsors.
The problem
Most clicks went to the article feeds. The jump links, FAQs, and glossary attracted little interaction, even though they were meant to guide readers through the page. For a destination built around editorial depth and credibility, that limited both the reader experience and the sponsorship model.
The redesign needed to make useful content easier to reach, especially on smaller screens.
My role
I led the redesign, working with research, editorial, sales, and engineering. I brought together the research findings, proposed the page structure, and designed the experience. I also worked through the different priorities each team brought to the project.
Research
I combined six months of analytics with 12 unmoderated Userlytics sessions. Half the participants were actively looking for AI content; the other half were more likely to encounter it while browsing.
Visual design
Participants found the colors and imagery overstimulating, and some perceived the palette as overly feminine for the broader audience we wanted to reach. This was my first project owning branding from start to finish, and I used that feedback to guide the refresh.
A quieter palette
I removed the light purple and reduced the overall use of color to give the brand a more sophisticated, digital feel without overwhelming the content.
More people in the imagery
I introduced more human imagery to help readers connect the coverage to ways they might use AI themselves.
Motion with a purpose
I worked with our motion design team on graphics that made the hub feel more modern and helped guide readers down the page.
Product direction
Some visitors wanted to see what was new. Others arrived knowing what they were looking for. I proposed two connected pages to support those different needs.
Tradeoffs
The redesign covered two pages, a new hub and a directory, but engineering didn’t have the capacity to build both at once. Waiting for both would have delayed everything, so I chose to ship the hub first. It was less work for engineering, and fewer readers came looking for something specific, which is what the directory was for. Shipping in phases also meant we could learn from the hub while the directory was built. The cost was an incomplete experience for six months.
Results
Reflection
We knew what we wanted to improve, but I would have set clearer numerical targets for each major change. Without them, I can’t fully separate the redesign’s effect from the AI news cycle. Jump-link usage is the clearest signal because it measures the new structure directly, whatever the traffic. Targets I’d set next time:
I would also measure interactions with the prototypes earlier and bring sales into testing sooner. Securing two sponsorships before launch showed that the prototypes could help us understand commercial interest as well as usability.