The path toward sustainable journalism, already challenged by a disrupted advertising business model, is also being undermined by something more unexpected – terrible data, says Tom Rosenstiel, a nonresident senior fellow at the Brookings Institution. He is the executive director of the American Press Institute and one of the nation’s most recognized thinkers on the future of media.
Analytics – another word for audience data or metrics—was supposed to offer the promise that journalists would be able to understand consumers at a deeper level. Journalism would be more connected and relevant as news people could see what audiences really wanted. Handled well, this should have helped journalists pursue what is at its core their fundamental challenge: learning how to make the significant interesting and the interesting more significant.
In his article, Tom Rosenstiel writes that a generation into the digital age, the problem associated with analytics wasn’t the one that some feared – the discovery that audiences only cared to be entertained and distracted. The bigger problem was that most web analytics were a mess. Designed for other purposes, the metrics used to understand publishing today offeed too little information that was useful to journalists or to publishers on the business side. They mostly measured the wrong things. They to a large extent also measured things that were false or illusory.
Traffic to most websites is probably over counted by more than double
As an example, he mentions the metric that was called “unique visitors”, which was not what it sounded. Unique visitors were not different people. Instead, this metric measured devices; the same person who visited a publication on a phone, a tablet, and a computer was counted as three unique visitors. If they cleaned their cookies, they were counted all over again. The traffic to most websites was probably over counted by more than double, perhaps more than triple.
Time spent per article, in contrast, could offer a sense of depth of interest in a particular piece. But by itself, it could also mean that someone stopped reading and walked away from the computer. Page views could tell a publisher how many times an individual piece of content was viewed. But views could not tell the publisher why.
Many stories reveal no patterns at all
Using conventional analytics, every story was an anecdote, says Tom Rosenstiel. Publishers would look at popular stories and say let’s do more like those. But they were largely inferring what “like those” meant. More often, the metrics revealed that the most widely viewed stories were extraordinary by definition – and fairly predictable. Rosenstiel mentions a publishing company, that examined its 50 most popular stories each year to look for patterns. Nearly all of their most popular stories were outliers, stories that revealed no patterns at all.
Knowing that a story about the championship football team or the bizarre tale involving a celebrity was viewed by a lot of users, offered little help in understanding how to make one’s coverage of government or water quality 20 percent more engaging. Nor did a page view by itself tell whether consumers found that content valuable or an annoying waste of time they were teased into viewing.
In addition to the problem that the metrics used in digital life were often crude and insufficient, there were too many of them, Rosenstiel believes. Yet, if one was to pick the wrong ones to focus on at the exclusion of others – page views, as an example, while ignoring time – it could chase one down the path of ignoring the significant in favor of trivial and hasten the irrelevance of a publication.
To journalists, the whole is always bigger than the parts
Journalists themselves had not helped, he continues. The prospect of suddenly knowing how many people read or watched a given story was frightening. To journalists, the whole was always bigger than the parts. They put the great photo on Page 1, the comics inside, or teased video of the water skiing squirrel coming up, in the hope that people would learn the important news along the way. But if they chose the other path, putting bikini clad celebrities at the top of their web page every day (and some have), they thereby forfeited their authority in favor of page views to sell banner ads that only a fraction of one percent of those viewers would ever click on.
Metric numbers in the first decade monitored the behavior of people mostly at work
Journalists also feared metrics because the numbers, particularly in the first decade of the web, looked awful, says Rosenstiel. Eye tracking studies suggested that the average web pages were looked at for just seconds, raising the possibility that digital screens were a terrible place to get people’s attention. The web itself, many theorized, could be shrinking attention spans, reducing patience and leading to a loss of deeper thinking. Maybe the only thing people wanted in the digital age, when consumers were in charge, was eye candy stories (Justin Bieber), quick hit items with no depth, or “list-icles” (7 things to know about Donald Trump’s hair).
A decade later, it turned out that many of the early hypotheses were oversimplified. For one thing, the early data on digital engagement weren’t measuring something inherent to digital screens. They were monitoring the behavior of people on desktops, mostly at work. With the advent of touch screen mobile devices, it became evident that not only could people pay attention to a screen, they would even read books on their phones.
Journalism has to understand and make use of the new world of metrics
But for journalism to understand and make use of the new world of metrics, it first had to learn how to turn bad data into good. Some publishing operations were beginning to try. They were attempting to look at the data and apply more journalism values to understanding it. Mic, for instance, was thinking about different metrics that combined equal trust. The audience tracking firm Chartbeat talked a good deal about time spent in highly engaged activity (recent scrolling). Upworthy tracks, what it called Attention Minutes, which involved both how many people looked at a piece of content and for how long and how long a particular unique visitor would spend with any of its content over time.
Efforts are made to create new, more useful analytics
Buzzfeed, one of the most sophisticated publishers when it comes to data, according to Rosenstiel, was thinking a good deal about what content people were sharing, or what it called “social lift.” If people had decided to share something with their friends, Buzzfeed valued it in a way that was measurable. The company could also triangulate that with topic to identify communities of interest.
What followed was one such effort in trying to create new, more useful analytics. It was based on work at the American Press Institute, where Rosenstiel is Executive Director, working with a talented digital team and in partnership with some 55 publications. The analytics were based on a unique data set of more than 400,000 stories from these publications and the patterns emerging about what people engage with. The work also showed that any publication, even small ones, could become a master of rather than a victim of its data.
Read the whole story here:
http://www.brookings.edu/~/media/research/files/papers/2016/02/19-media-analytics-rosenstiel/solving-journalisms-hidden-problem.pdf
