Sometimes the best way to figure out what people want in an app to ask them directly. In late 2016, Matilde Giglio and Mayank Banerjee traveled to dozens of universities across the U.K. to ask students some probing questions about their news consumption habits. Where were existing news sites and apps failing? Where could they do better? What features are important to younger readers?
That listening tour, which added up to conversations with thousands of students, helped give birth to Compass News, an app designed to give users a simple, informative digest of the day’s news. Twice a day, Compass News pushes out six of the biggest news stories, offering short summaries in a card-like format. Because the app is designed for college-aged readers who may not be familiar with certain stories (the app’s average user is 22 years old), many stories also include a “What’s the story so far” tab, offering background information on the key events and people.
Giglio said that Compass News focuses on college-aged readers because so few existing news apps and news organizations currently do so. “We in this industry consume a lot of information and news. We pay attention to multiple outlets and check them and Twitter multiple times a day. We wanted to create a product for everyone, not for those kinds of news junkies.”
That idea has started to resonate in the U.K., where Compass News has drawn 50,000 users over the past 10 months. The company has ambitions to grow, particularly in the U.S., over the next few months. To pull that off, Compass News has joined the latest class of media startup incubator Matter, which will give it some cash and guidance on how to accelerate its business. Compass News also recently inked a $1 million funding round to help with its expansion plans.
Compass News employs six news editors to scan thousands of stories from over two dozen sources each day. (Its total staff is 12.) These curators, who have experience reporting for outlets such as BBC, the Financial Times, and Wired, focus their attention on “stories that are really worth reader’s time,” said Giglio, which tends to be those that give readers context, or take a unique stance, as opposed to those that just report the facts of what happened. “Helping people figure out what things mean is a big area of focus for us,” Giglio said.
Also important: making the app feel finite. One of the most common concerns that Compass News heard from students is that the modern news environment is often overwhelming, particularly if people find out about a story hours after it has broken. This feedback inspired Compass News’ twice-a-day editions, which are designed to give users a more digestible, newspaper-like version of the news. Another benefit: encouraging people to check the app multiple times a day. “The students we spoke to really do feel that less is more,” Giglio said. (The now-defunct Yahoo News digest and The Economist’s Espresso app, among others, have also taken versions of this approach.)
Compass News’ story selection process has become increasingly driven by artificial intelligence. For months, the company has trained an AI editor to observe the Compass News editors so as to mimic their curation decisions at scale, particularly when those humans are off the clock. Co-founder Mayank Banerjee compared the idea to the process of training a self-driving car. “You don’t teach a self-driving car to drive by encoding rules. You teach it by getting a human to drive it around by millions of miles,” he said. “It’s a learning algorithm, as opposed to a pure rules-based one.”
Ultimately, Giglio and Banerjee hope that the Compass News feature set can help it stand out where many other news apps have not. Likewise, the team thinks that the app’s arrival a time of growing skepticism about the role of Facebook and Twitter in the news ecosystem will be able to give it an edge.
“Using Facebook right now involves opening it up, using it, and looking back to realize that you wasted an hour of your time,” said Banerjee. “That’s not what we’re setting out to do.”
This article originally appeared on Nieman Lab. It was republished on IJNet with permission.