How CIR Gauged the Success of News Interactives

    by Lindsay Green-Barber
    January 20, 2016
    Screenshot of The Center for Investigative Reporting's data apps page.

    Lindsay Green-Barber is the Director of Strategic Research at the Center for Investigative Reporting. This post was written and originally published by CIR and is republished here with permission.

    As news organizations strive to take advantage of the possibilities afforded by online platforms and to meet the demands and needs of increasingly data-savvy audiences, new forms of digital storytelling abound. A news interactive is one of these new storytelling methods, intended to take complicated data and make it accessible to media consumers. However, news interactives, or apps, require significant organizational resources, and the impact of these projects on audiences remains underexplored.

    “Why do we make news apps, and what makes an interactive successful?”

    At the Center for Investigative Reporting, we pay careful attention to the impact of the media we create and to the calculus between the costs and benefits of assets produced on various platforms. We wanted to better understand the impact of our news interactives and to identify characteristics that contribute to an interactive’s success. While based on an analysis of CIR news applications, these findings also have implications for the field.


    To begin, I asked CIR staff, “Why do we make news apps, and what makes an interactive successful?” In response, they identified three main functions of news interactives: to “dive deeper” and “explore” data, in general; to enable users to find information specifically “relevant” to one’s own life or community; and to “tell a story.”

    But, in fact, do people dive deep and explore our data apps? And does this exploration carry over to deeper exploration of other elements of our stories? Or does the audience consume an application as a standalone story?

    Screenshot from CIR's app Ebay Waffles on Guns as Debate Over Assult Weapons Continues

    Screenshot from CIR’s app Ebay’s gun-selling policies.


    What is a news interactive, anyway?

    In the Data Journalism Handbook, Scott Klein of ProPublica says a “news application is a big interactive database that tells a news story.” However, news interactives range from quick tools that illustrate simple concepts to complex data applications built upon sizable databases. For example, CIR created a simple timeline to show how “Ebay waffles on guns as debate over assault weapons continues,” while the data interactive “Do you live in one of California’s pesticide hotspots?” is based on 10 years of data from the California Department of Pesticide Regulation, including 22 million applications and 1.5 billion pounds of pesticides, and allows for deep interaction by the user.

    OK, but what is a successful news interactive?

    For the purpose of this research, I consider a news interactive to be successful when it is highly trafficked (many unique visitors), deeply explored (evidenced by the number of events or instances of interaction on the page), and results in a better-informed audience. Here I focus on the first two indicators of success and leave measuring a better-informed audience for a future project.

    I tested the following characteristics to see how they correlate to the success of news apps:

    • Design.
    • Relevance to a large number of people.
    • Have a good “10,000-foot view” and a good “5-foot view.”
    • Present a story interestingly to a broad audience, as well as a specific constituency.
    • Present data to which the audience can immediately relate.
    • Present data either not available elsewhere, or in a way not available elsewhere.
    • Load quickly for the user.
    • Quickly deployable.
    • Easy to find.
    • Presented in context.
    • Have memorable information nuggets that can be shared.
    • Have clearly written, yet intriguing, headlines.

    To test how these characteristics relate to an interactive’s success, which this study defines as traffic and clicks, I analyzed 29 CIR data applications. Each characteristic was assigned an ordinal value based on a three-point rubric (Low, Moderate, Strong). Two coders analyzed each data app. Intercoder reliability was high for six independent variables:

    • The degree of interactivity, meaning how deep a user can go in exploring data and interacting with the application.
    • Whether it was part of a full package, meaning there was an accompanying text story, video, audio and/or other content.
    • National in scope, meaning that the data are relevant to a national U.S. audience.
    • Local in scope, meaning that the interactive presents data relevant to a geographically local audience.
    • Designed for a specific constituency, meaning that the interactive presents data relevant to a particular interest group audience.
    • Unique data, meaning that the interactive presents data not easily available elsewhere.

    I first determined the correlation between each of the independent variables (calculated as the average score of two coders) and the dependent variables of Web traffic (for 90 days post-publication) and average events per session. I found that Web traffic is moderately positively correlated with applications that have higher scores for “degree of interactivity,” “full package,” and “local.” Web traffic proved to be moderately negatively correlated with “load time.” An interactive that scored high in these criteria and experienced high traffic is the previously mentioned application that explores California’s pesticide hotspots.

    Screenshot from CIR's app on California pesticide hotspots.

    Screenshot from CIR’s app on California pesticide hotspots.

    Because CIR has gone through multiple transformations over time, I also split the data into two sets: one for data applications that were hosted on the California Watch website (2011-12) and one for apps hosted on CIR and Reveal’s websites (2013-present). “Degree of interactivity” and “full package” remained moderately correlated with Web traffic for each subset. However, “local” proved to be correlated only for CIR and Reveal application traffic.

    Traffic to CIR and Reveal apps is moderately positively correlated with “design/aesthetic” and strongly negatively correlated with “load time,” while California Watch apps do not correlate with either of these variables. This suggests that users are becoming more discerning about the design of the apps they use and more accustomed to rapid load times.


    The same analysis using events per session shows that this dependent variable is correlated with “degree of interactivity” and “full package” (both at the .05 level). However, when we analyze the data as two separate sets (one for CIR and Reveal, a second for California Watch), we find that “degree of interactivity” is strongly correlated with events per session for CIR and Reveal and not at all correlated with events per session for California Watch. And on the flip side, “full package” was moderately correlated with news apps, in general, but not at all with California Watch’s or CIR or Reveal’s in particular.


    Conclusion and recommendations

    This analysis supports the following hypotheses:

    • Applications that are part of a full package will be both more highly trafficked and more deeply explored by users. This could be because there is more content pushing audiences to the interactives or because they are more heavily promoted by the organization.
    • Applications that have high degrees of interactivity will logically have more events per session, but they also have more unique visitors.
    • The aesthetics and design of applications are becoming increasingly more important to users.
    • Load time is increasingly more important to users’ likelihood of both visiting and using an app.

    While these findings are specific to CIR, the implications are valuable for the industry as a whole. These findings would be strengthened if additional newsrooms applied the same methodology to analyze their news interactives and shared their findings. If other organizations found support for the hypotheses listed above, we as a community could be more confident that this is a general phenomenon.

    Click here to view the full, unedited white paper complete with footnotes.

    Lindsay Green-Barber is the Director of Strategic Research at The Center for Investigative Reporting. CIR is one of the largest nonprofit newsrooms dedicated to investigative reporting. It recently launched a weekly national public radio show and podcast called Reveal, produced with PRX. Learn more at revealnews.org.


    Tagged: analytics cir data apps metrics reveal tracking

    Comments are closed.

  • Who We Are

    MediaShift is the premier destination for insight and analysis at the intersection of media and technology. The MediaShift network includes MediaShift, EducationShift, MetricShift and Idea Lab, as well as workshops and weekend hackathons, email newsletters, a weekly podcast and a series of DigitalEd online trainings.

    About MediaShift »
    Contact us »
    Sponsor MediaShift »
    MediaShift Newsletters »

    Follow us on Social Media