What the New York Times found out by mapping audiences’ reading paths

Digital Innovation at the NY Times

Most media industry professionals have, hopefully, spent some time pouring over the digital innovation report from the New York Times. Throughout it, technology being at odds with “the way things have always been done” at traditional publishers becomes a recurring theme.

A summary of the Package Mapper story

One example: Director of News Analytics, James Robinson’s description of the “Package Mapper,” and how they determined that providing relevant content, regardless of section or news desk, creates a better experience for their readers.

In an article on Nieman Journalism Lab, Robinson walks through how and why the Times‘ technology team built the (questionably named) “Package Mapper.” In summary: the Package Mapper tracks the flow of traffic to and from articles with a certain tags, like  “Golden Globes” or “Olympics,” within the Times‘ site. Each URL is depicted by a circle showing the number of visitors, and a color representing a section of the newspaper. The Mapper also shows directional readership, so that looking at this data in a map format, you can quickly see which stories brought in the most readers, taking into account that they most likely are not coming directly from the homepage anymore, and what those readers do (or don’t do) next.

Image  from Nieman Journalism Lab

Robinson makes a fairly ground-breaking discovery here, namely that the Times wasn’t providing readers with ways to read additional articles about the same topics, even though those articles existed on the website. “Why not?” He answers his own question,

“As it turns out, our CMS does a great job of suggesting related articles within a given section, but not within a particular topic — those promos must be hand-coded by producers. And unfortunately, the producers for these pieces were working on different desks, with little coordination between them.”

In other words, the Times was creating dead ends out of their own stories.

The real problem with dead ends for digital publishers

The media industry constantly harps on how hard it is to attract readers attention in a world where there’s so much noise. Companies, including the Times, spend millions of dollars and devote whole departments to attracting readers through SEO, social media promotion, mobile optimization, and more. From a business perspective, the most efficient use of that money, or as a marketer would call it, the the highest ROI, would be to make sure once on the site, readers were given the best options to continue reading. And the Times was failing at that.

Using this insight about their audience, Robinson’s team began to implement changes, first through manual monitoring, a phone call and producers making code changes, and later through a browser bookmarklet for editors. The screenshots of reader maps show that making these changes did encourage readers to visit the additional stories.

Image  from Nieman Journalism Lab

Two challenges stopped this from happening sooner. First, the institutional problem of thinking about content grouped by editorial selections in news desks versus how the users actually read that content. And second, the technology available to fix the problem, which arguably is mostly a function of the Times not using their meta-data in the most effective way. Joshua Lasky, in a great piece on Medium, (“It’s the meta-data, stupid”) also notes how this problem inhibits the Times and other publishers’ abilities to maximize the content they’re producing.

Is there hope for publishers without a Package Mapper?

If the New York Times  is just now figuring this out, that’s bad news for the rest of media, right? Wrong. Newer media companies, unburdened by existing organizational structures,  have leapfrogged this problem and use a variety of ways to make sure their readers have access to the stories they’re most likely to want to read next. For example, Cheezburger.com (full disclosure: a Parse.ly client) mapped their user journey  by  examining  meta-data, and then implemented article recommendations to make sure their readers didn’t end up at dead-ends. It increased their KPIs across the board and was part of their eventual path to profitability. You can read the full case study on what they did here: http://bit.ly/Cheezburger-Parsely.

As publishers have more access to how their audience actually reads their content, they’re finding out their previously held assumptions don’t hold ground anymore. Kudos to the NYTimes for identifying ways in which they can improve this – but we have to ask, if it’s taking them this amount of effort to make this small change, will they be able to address other issues, like reader loyalty, quickly enough to compete with digital first properties?