Gjettekonken
January 2026 - August 2026
NRK planned to have a good collection of special features for the front page of nrk.no for the 2026 World Cup in football. As the men's national team qualified for the first time since 1998. Our goal was to reach people with little existing knowledge of the tournament. Specifically, fans who were only interested because Norway had qualified, and who had no existing product built for them. We designed to increase their understanding of and interest in the championship, not just their engagement with it.
The product: a daily prediction game tied to each match, where users guessed match outcomes before or during play.
Building on a Previous Concept
The product builds on an earlier version of the concept, which had drawn a peak of 40,000 players for a single event. This version drew almost 280,000. Making it our most successful iteration to date. The core change this time was structural: shifting from a fully live experience to an asynchronous one. That shift raised two design problems. First, if users missed the live match, how would they find out their results? Second, how could the guessing format stay compelling enough that people wanted to predict, rather than experience it as a form?
We treated the second problem as a hard constraint: without real gamification, the format would collapse into a standard form dressed up as entertainment. So we set one requirement for ourselves early: every part of the flow had to feel like play, not data entry.
Key Features
Results delivered as a story. We needed to reveal match results without lengthening the page or adding a separate results tab. The solution was a story-format reveal that opened automatically, so we never had to explain the feature to users. Because it opened without a prompt, we built in fast, obvious controls to skip ahead or slow down, so users could set their own pace.
Visible consensus. To create a sense of playing alongside others, without turning it into a full multiplayer game, we previewed each user the average guess from all other players immediately after they submitted their own. Users could then keep their answer or change it. This built a sense of shared play while adding real stakes to the decision.
Groups. Groups is a new feature we're developing together with the minigames team. The underlying hypothesis: users in a group play more often, and the social layer adds depth to the game beyond individual scoring. This is early-stage work, minigames turned out to be the most natural entry point for groups in the wider NRK product ecosystem.
Screenrecording from NRK.no on July 1.
Designing Around Live Constraints
As mentioned, a central constraint was timing: half of the matches kicked off late at night. The previous version of the concept ran fully live, with questions posted and answered in real time during play. That model depended on a production crew we didn't have available for the World Cup. Their workflow couldn't support a concept with that level of live demand. So we redesigned the format around a "24-hour interaction day" instead of a live window. Gjettekonken was promoted across platforms throughout the day and night, and users could submit guesses at any time. Scoring reflected this: guesses submitted before kickoff earned full points, guesses submitted after kickoff earned none.
Process
Before any sketching started, we defined the game loop; the core path a user follows from starting a round to finishing it and starting the next. This came first because every other design decision depended on it, including the content itself. The match questions were written by two partners from NRK Sport, working from that same loop.
During the anchoring phase, we tied each concept decision back to NRK's broader strategy. This made it easier for stakeholders across departments to align behind the product and kept the team motivated through a long lead time: two and a half months of anchoring and one and a half months of design before development started.
Measurement & Learning
The product ran live for six weeks, with an agreement upfront that we'd adjust it in response to data. That required hour-by-hour visibility into performance, so we built a tracking dashboard in Data Studio, using BigQuery queries we wrote with help from Gemini. Working with AI tools this closely carried over into how the team built the product itself: a significant part of this project was learning to collaborate with a Claude-first developer, who was simultaneously building their own workflow for integrating Claude into development.
