When I quit my editing job at Wired in 2012 to spend more time with my startup, Contextly was focussed on helping readers dive deeply into a story, via smart related links at the end of stories.

I was frustrated then at the state of digital publishing tools and the lack of innovation in the news media generally.

I still am.

After all, one of the benefits of digital publishing over paper publishing is the ability of readers to find stories that aren’t that day’s stories.

We found that readers *really* respond to smart related recommendations pointing them into a publisher’s archives. Readers are curious.

But one of the things we realized is that not all readers are in deep dive mode.

Readers show up to a news story with a range of motivations. Sometimes a reader getting to the end of a story is done with the topic.

The paper analog of this is turning the pages in a magazine or jumping to a different section of the paper.

So we expanded our recommendations to include an Explore section, which let readers find other new content and also surface evergreen content. (We define “evergreen” as an story that’s older than 3 weeks that’s still giving value to readers, which we can algorithmically determine based on audience behavior). We’ll also put in personalized recommendations for return readers.

This gives a far better recommendation mix than common, naive strategies like the top 5 most popular stories or the five most recent stories.

The goal then is to create pathways for readers with varying goals; from killing time to diving deeply into a specific topic.

We then tackled an even harder problem: following a story. It’s a common thing for readers to do, as evidenced by the common question: “Have you been following the story about Samsung Note 7 batteries/water poisoning in Flint/the Warriors/etc?”.

Those who subscribe to a daily newspaper are much more likely to be able to stay abreast of ongoing stories fairly passively, but one of the things that digital news (and machine learning) makes possible is letting readers expressly say they want to follow a story and then have that sent to them.

So we built the FollowUp button, which goes on every story on a publisher’s site. Readers can subscribe to follow a story, and, if we know it, we’ll send the reader an algorithmic timeline of that story from that site. Then when we detect relevant FollowUps (without using tags or categories!), we’ll notify the reader.

This is very different from being presented with a menu to choose a category like politics, and the results have been quite encouraging. In short, readers don’t just follow big national news stories, they follow stories about their local community (controversial planning decisions or local crime) or about their interests (high school lacrosse).

We used this same thinking to drive development of other features, as well.

We’re not alone, of course. The Washington Post has done a great job since the purchase of the paper by Jeff Bezos of designing “reader-first,” which was one of his explicit goals in 2013

Smart news design serves audience interests, like not missing the next story on the California drought, while building new digital experiences to fulfill those needs, like the FollowUp button does.

This also happens to cater to publisher’s interests — getting reader’s email addresses, having a new direct distribution method and getting readers to come back.

That approach is the opposite of all-too-common dark patterns like pop-ups that obscure stories as readers read them, unneeded page jumps, stats-juking photo galleries, ad units disguised as recommendation units, or infinite scroll sites attempting to mimic the Facebook feed, with none of the grace or algorithms.

Those design choices aren’t designed to satisfy readers’ goals — and in many cases, actively and intentionally impede the reader. On some sites, the design feels like an obstacle course intended to keep readers from actually getting to the end of a story, let alone finding another story on that site.

Build for your community; not against it. Readers will reward you for it.

This post is a response to the newly resurrected Carnival of Journalism blogging challenge. The prompt for this month was: Designing for audience behavior…