The original TexasTribune.com site had been built in 2009 without a content strategy, without a style guide, and (perhaps most sadly of all) without any user research whatsoever. in 2015 I created and began executing the plan to redesign the entire site, while simultaneously improving our process.
At the time, we didn’t have a dedicated UX person, or any user testing groups. While we were addressing that, we started with Google Analytics; all-knowing Google provided detailed enough demographic material to start creating personas, and gave us target users to interview.
We found that our main audience was what we deemed the “Tribune guru”; audiences we wanted to expand were fly-by millenials, social browsers, and mobile-only readers.
Using quantitative data, we also found we needed to reexamine our assumptions about readers clicking on our social shares and related stories.
We were actively pursuing a younger demographic, and so I conducted user interviews with students. I also did a heuristic study using research from the Engaging News Project, and found the focus group data our marketing department had pulled.
Finally, we planned an iterative article page rollout to make sure our current users were comfortable with the changes we made. After beta testing various, we would show the new page as a clickable option. Eventually, the new page would become the default, with an option to switch back to the old page, and a survey allowing feedback.
Content, and Getting Buy-In
In keeping with best practices — some studies show that sudden, large redesigns can cause as much as a 50% drop in traffic — we started with the main article page. Inspired by Kevin Hoffman’s “Co-Design, Not Redesign” philosophy, and by his ideas designing meetings, I had to get internal consensus on content and understanding what communicate what we didn’t know about our current readers.
I started with some simple content prioritization exercises. My first step was interviewing internal stakeholders to make a requirements document. The next was running card sorts to get us focused internally on the content really needed to be on the page. The good news was that everybody agreed we needed to pare things down.
In the meantime, I ran targeted content audits. Working with sponsorship, I identified the most performant units; together, we reduced the number of total ad units from over 200 to under 20. I also worked with news apps to sunset news application pages that were no longer relevant, and removed over 50 high-maintenance pages, as well as an entire content type based on a $300/month service we no longer used.
After conducting a style audit and finding twelve separate styles for headlines, we decided that it was best to start fresh, rather than trying to use what was there. Moving forward, it was imperative that we develop a better way of putting content and design together. We needed to get on board with a style guide, and modular, predictable styles that would let us build faster.
The article page redesign was the first step of establishing a living and modular style guide. I’d already overseen the launch of a separate site that was our first foray into modular styles. In the interim, I’d also overseen the development of a style guide for our festival site. The lead developer worked with the designer to create a style guide for the new article page itself. The editors, in particular, were enthusiastic, because they’d always wanted the ability to use story templates in the CMS; this was first step in that process.
Finally, there was the question of serving our internal users. Understanding our internal users’ pain points had been a long-running project of mine; in this project, we started to think about how our editors would be able to use the new styles to build custom pages from within the current interface.
By the time I left, we had a beta testing group of expert users (also known as site “superfans”) as well as a program to find non-users via Mechanical Turk. Since I left before the launch, I can’t speak to the metrics; I feel confident, however, that at a bare minimum, the new system will serve the Tribune’s current users better and allow our internal users much faster development. The main result I wanted was to get our organization to embrace design thinking and iterative development, and on that front, I was quite satisfied. By the time I left, we’d gotten stakeholders to collaborate with us on writing user stories; moved from waterfall development with Photoshop to agile development with sketches and wireframes; and most importantly, gotten the relevant stakeholders on board with good development practices.