Pros and Cons of Measuring Web Traffic
By Jim BradyThe availability of real-time audience and traffic information has been having a major impact on digital journalism for more than a decade. The question that’s more frequently debated is whether having this information has had a positive or negative effect on the craft of journalism and how we serve our readers.
Like most issues relating to online journalism, this question is often framed in black-or-white terms. Many purists complain that having this information turns news home pages into popularity contests and whittles away at one of the tenets of journalism: our insistence on telling readers what they need to know, not merely what they want to know. Many news digerati believe this traffic data should be monitored minute to minute, and be a driving force in determining what's being promoted on news sites. I think both positions are wrong. The answer—as it usually does—lies somewhere in the middle.
The availability of a new tool is not, in and of itself, a good or bad thing. It's how the tool is used. If you refuse to use traffic information out of some misguided desire to be pure, that’s a tremendous mistake. If you are updating your home page based purely on what’s getting the most traffic at a given time, that’s an equally big mistake.
The amount of information available to those of us who have run large web sites is amazing. We know, at any point in the day, how much traffic our web sites are getting, how much traffic has gone to every piece of content, how many unique visitors have come to the site, which domains outside visitors are coming from, how long readers are staying on our sites, how many pages they're hitting, how many clicks every link on our home pages are getting, and much more.
True digital professionals have a deeper understanding of what impacts traffic patterns, which prevents them from jumping to the wrong conclusions. When I was executive editor at washingtonpost.com, my philosophy was simple: We should watch the numbers closely on a daily basis, but we should not let real-time, hourly information drive our editorial strategy. When a Washington Post story would get a link from the Drudge Report, Google News or Yahoo's home page, it would often drive hundreds of thousands of page views to that article. While I appreciated those page views greatly, I knew that traffic to that one article did not suggest a trend, since that traffic came almost exclusively from one link on one site. Conversely, traffic to advice columnist Carolyn Hax's weekly live discussion didn't drive the kind of traffic a Drudge link could, but we knew that visitors to Hax's chat—and to most of our live discussions—were incredibly loyal users of our site. So we focused on live discussions as an important part of our overall strategy, despite the fact that traffic to that feature alone might not to justify that decision.
In most ways, usage metrics are an invaluable tool. If analyzed properly, they can signal trends that sites can take advantage of. One example involved the Post’s Pulitzer Prize-winning fashion columnist, Robin Givhan. A few years ago, she wrote a column about then-Vice President Dick Cheney wearing what she deemed an inappropriate outfit during a somber ceremony at Auschwitz. That column went viral. So did her next column, on then-Secretary of State Condoleezza Rice wearing a skirt and boots at a U.S. Air Force base in Germany. The popularity of these two columns tipped us off to the fact there seemed to be an appetite for columns that explored the intersection of celebrity and fashion. So we made a concerted effort to promote Robin’s subsequent columns on that topic, and not surprisingly, they also did well.
At the end of the day, part of every online editor’s job is to build loyalty and get readers to come back to your site more often. To do that, you have to balance what gets traffic with what you're best at. If you’re lucky, those will be the same thing. But they usually aren’t. At washingtonpost.com—even with the educated, affluent audience we have—I was pretty confident that a story about the celebrity meltdown du jour would get more traffic than our story on President Obama’s current thinking on the Department of Agriculture. But The Washington Post’s bread and butter is coverage of the federal government, politics, diplomacy, national security, local news and local sports, not national entertainment news. To put it another way: If The Washington Post decided to promote stories on its home page based purely on traffic potential, what makes it unique would quickly evaporate. So any analysis of traffic also has to keep this in mind.
Just about everything about journalism has changed in the past five years, but there’s one that I don't think has: I believe people still want The Washington Post, the New York Times, the Wall Street Journal and other papers to play a gatekeeper role. If we start letting real-time traffic drive our home-page promotion, for example, we're on the path to becoming Digg. What Digg does is terrific, but it's not what newspapers should be doing. There was nothing in our traffic history to suggest that stories about military veterans were of particular interest to our readers. But when Dana Priest and Anne Hull uncovered the poor conditions at Walter Reed Army Hospital, the story went global in hours. That kind of journalism will be increasingly at risk if we get too caught up in the race for page views.
There’s a great scene in the film “Platoon,” where Willem Dafoe explains to Vietnam newbie Charlie Sheen what gear he needs and doesn’t need from the heavy backpack he’s been issued. The same logic needs to be applied to metrics and news Web sites. A good editor has to know what information is valuable and actionable, and what information is tainted and thus useless. After all, as the saying goes, a good craftsman never blames his tools.
But a recent three-year stint as the Washington bureau chief for Salon.com left me brooding about the unanticipated side effects of this kind of reader sovereignty on the web. Salon boasted an admirable commitment to investigative reporting and probably devoted as high a percentage of its gross revenues to aggressively covering the 2008 presidential campaign as any publication in America, with the exception of Politico. As the rare stand-alone news Web site that is a publicly traded corporation, Salon’s precarious economic condition has been well-documented. But despite a tight editorial budget, its writers and editors aspired to journalistic excellence.
So what was the problem? Part of it was ideological. Salon has always been a left-of-center magazine reflecting its roots in San Francisco. But it also received constant feedback from the marketplace to become shriller. As an early adapter of the practice of allowing readers to post their own letters on the site, Salon inadvertently created its own left-wing shock troops who castigated any writer who deviated from the simplistic “Bush is a war criminal” and “all Republicans are evil” party line. Salon’s editors gamely tried various schemes to elevate the tenor of the letters threads. But this chorus of ideological invective had a pernicious effect on all but the most thick-skinned writers.
The larger difficulty—and one not stressed enough in the discussions of Internet journalism—is the tyranny that comes from real-time readership numbers. The dark side of reader power is not unique to Salon or to political Web sites. It casts a shadow not only on Internet-based publications, but also raises questions about the viability of proposed economic models like micro-payments to sustain quality newspapers.
I came to Salon with three decades experience in print journalism. Like almost all newspaper reporters and magazine writers, I had spent my career bathed in the narcissistic fantasy that every person who subscribed to a publication for which I worked eagerly devoured every word that I wrote. I blithely assumed that my personal audience was the circulation times three for pass-along readers.
Granted it was a fool’s paradise. But never once in my career was I lectured about readership numbers. If my editors had shared focus-group or polling information with me, the data would have been vague and out-of-date. There was no way to determine how many readers skipped immediately to the sports section after yawning through the first paragraph of my latest newspaper column.
But Salon—like every major publication on the web—receives real-time data about how many computer users have a particular article on their screens at that moment. In theory, this rich bounty of information should make it easier for an online publication to satisfy its readership. In reality, the message from readers was that they craved what they already knew and demanded constant reinforcement of their pre-existing attitudes. Economic survival depended on higher traffic, so there was a built-in incentive to billboard articles that were the most popular with readers.
Most well-designed online publications (Salon, Slate, The Huffington Post) have layouts that highlight five, six or maybe eight stories at a time. Since many readers refuse to scroll down an entire web page searching for a story that intrigues them, the prime real estate on an online publication’s home page is more valuable than a newspaper front page or a magazine cover. (A quasi-invisible story is not apt to attract links and that lessens the odds that it will pop up on a search engine).
What this meant in practice at Salon was that an article might have as few as six hours to prove itself with readers before it was yanked out of a position of prominence. Needless to say, the shelf life of an article that grappled with a campaign issue like health care was about equal to butter in an un-refrigerated dairy case. Rants about Fox News or Sarah Palin often received many times the readership of investigative articles on the Iraq war or densely reported political pieces about swing voters in Indiana. During September 2008—the second most heavily trafficked month in Salon’s history—the dozen leading stories in terms of readership included: “The Sarah Palin pity party,” “What’s the different between Palin and Muslim fundamentalists? Lipstick,” “The pastor who clashed with Palin” and “Sarah Palin’s wasteful ways.”
As a result, Salon's internal culture and stray comments by editors worked to discourage writing about important topics (campaign-issue analysis, non-war-related foreign news) because they invariably earned lower readership numbers. In a newspaper, a great lead anecdote might lure readers into an article on a non-sexy topic. But on the web, all the salesmanship has to be provided by the headline. Clever headline writing is rarely an option online, since heads and decks have to be written to fit the irony-free sensibilities of search-engine algorithms.
Of course, this obsession with traffic figures does not mean that all news Web sites will turn their coverage into a paparazzi patrol on the lookout for the latest about Britney, Lindsay and Paris. What I do fear is that worship of real-time readership numbers will send online journalism to the frivolous fringes of serious topics. Emotion invariably trumps substance. You can always gin up more traffic sniping about what Chris Matthews just said about Tim Geithner than by discussing the details of the Treasury Department’s latest bank rescue plan.
My experience at Salon has whetted my appetite for online journalism. I recognize that we are in the midst of a technological transition akin to the one that was launched when a German printer named Gutenberg threw his first book party. But what Internet journalism requires are self-confident editors (and owners) who can resist the blandishments of quick-react readership statistics and allow laudable stories time to build their own audience. Otherwise, we will all—reporters and readers alike—find ourselves stuck in heavy traffic with nothing but fluff to read.