tv2oj.dk — Redesigning regional news for how people actually read

The brief came from leadership. As a public service broadcaster, TV2 Østjylland needed to reach a younger audience. Our readership skewed older, and younger people were on TikTok and Instagram, not on our site.

Working next to the newsroom, I saw a problem the brief didn't mention. Journalists also shot, edited, exported and uploaded their own images and video, in five formats: 16:9, 9:16, 1:1, 4:3 and 4:5. Journalists estimated around 30 minutes per extra format.

So the challenge became: make the site feel natural to people coming from social media, without adding work to the newsroom.

Our team sat with the journalists every day, and we owned the CMS interface they published with, so the feedback was constant and direct. One moment made it clear. Some of our crop defaults were inherited and some were mine, all chosen to look like social media ratios, and close-up portraits were cropped so people lost their heads.

Nobody had asked us for a redesign, but we knew something had to change. I suggested a one-week design competition: each of us built a prototype, coded or not, and presented it at the end of the week.

Two of the prototypes kept a grid with many crops, close to the existing site. Mine was the only feed: one story per row, each shown in the format it needed instead of forced into a crop. The team voted, and we went with the feed.

Two formats instead of five. I cut the formats journalists had to produce to 16:9 and 9:16, standardised in the CMS. A single-column feed doesn't need a different crop for every slot, so the other three formats only added work.

A single-column feed on every screen. Each story gets its own space, and readers take in one thing at a time. A feed reduces editorial control over the desktop hierarchy, so I added configurable lead story treatments and a data-driven right column.

References from news and social video. From The Verge I took a centred feed with a right column whose content changes as you scroll; during the municipal elections, it showed where to find polls and local results. For video, I took the vertical format and feel of short-video platforms like TikTok, because that's where the younger audience already was.

Icon navigation. This was the most contested decision internally. I took a high-fidelity interactive prototype to leadership, including the CEO, so they could try it on a real phone instead of judging static screens.

A design system the brief didn't ask for. Tokens, a Figma component library and a documented spacing and type scale, because a feed stays coherent only if the parts are shared.

I built the prototype in HTML, CSS and JavaScript, pulling live stories from a public news API. Every test used real, unpredictable images instead of two perfect ones, so we could see where crops broke.

Two rounds of usability testing confirmed the direction: readers navigated by interest, not recency, and the grid felt wrong on mobile. Testing also exposed problems with colours, UX writing and the site's information architecture, and those findings sharpened the direction.

After launch, we validated the design with real usage: Plausible events on the UI areas we had open questions about, and Microsoft Clarity heatmaps showing where people actually interacted. That data guided the refinements, including more iterations on the 9:16 crop.

On medietorvet.dk, Syddansk Universitet — Center for Journalistik — trains 600+ journalism students on the same platform.

Youth reach grew overall, mainly through the social media strategy. I don't have site-level age data, so I can't claim the redesign itself reached younger readers.

Me, lead designer and frontend developer: design direction and final design decisions, the coded prototypes, the design system.

Lærke, frontend developer: frontend build, challenging my decisions, support in user testing.

Frederikke, project manager: project management, the user interviews, and challenging my decisions.

Tokens first. They went in midway through the first iteration, and it cost two weeks when components began to drift.

Audience data from day one. Plausible and Clarity showed how people used the site, but not who they were, so I couldn't prove the redesign reached younger readers.

Time the formats. I'd time a few journalists publishing a story, so the cost of each format is measured, not estimated.

The SDU adoption was the test I hadn't designed for. A university either builds on what you made, or they don't. They did.