Sophisticated Web Stats Give Editors Better Idea of Reader Interests

By: Steve Outing There are plenty of important stories making headlines today, from Iraq violence, to terrorists wreaking havoc in London and elsewhere, to Social Security overhaul attempts, to massive budget deficits and related tax cuts for the wealthy, to debate over presidential aides leaking the name of a CIA agent. Of course, keeping up with and reading about all that is tantamount to eating the vegetables in your media diet.

Many online readers, it would appear, prefer to go for the dessert. Take this March 8 story published in The Guardian's education section: "Necrophilia among ducks ruffles research feathers." Well, who wouldn't be curious enough to read an article with that provocative headline!

You probably guessed that this particular story was popular -- indeed, until the London terrorist bombings, it held up as one of the most-read stories on the U.K. paper's Web site for the year.

We know that, of course, because newspaper Web sites like The Guardian's these days have detailed traffic numbers at their disposal. Today's news editors know for a fact if sports articles are the biggest reader draw, or if articles about local crimes consistently outdraw political news. They can know how particular stories fared, and track the popularity of news topics.

If a news editor isn't savvy enough to realize that deviant ducks are going to get heavy readership, audience statistics will set him straight on that and much else.

But what should that editor do with that knowledge? Abandon his traditional news judgment as the principal news-selection tool and look foremost to Web site logfiles to guide story placement? As editors are armed with increasingly deep and fast audience tracking data, how much should they rely on them?

I put such questions to a group of online news editors in the U.S. and overseas for this column.

The advent of live stats

Not every news site has access to live stats for articles. At The Guardian's Web site, article pageview numbers weren't available "live" until very recently, reports assistant editor Neil McIntosh. Story selection "has all ridden on gut instinct thus far, rather than hard numbers from that day, or hour," he says. Prior to the stats upgrade, where Web site stats came into play was with identifiable waves of interest -- which sometimes resulted in high play of a story over several days. Editors left to their own judgment might not have played the stories in the same way or kept them highlighted so long.

McIntosh says watching the stats during the first round of London terrorist bombings altered the choice of format for the news later on. The site's news weblog saw "remarkable traffic" on July 7 following that day's bombings; the blog included minute-by-minute coverage of the situation. When another round of bomb alerts occurred two weeks later, the site went back to the minute-by-minute form, "because we knew that's what readers would most likely want."

McIntosh expects the introduction of live stats being available to editors to encourage more reaction to reader behavior. "Stats can inform decisions and help with the business of allocating resources during a crisis," he says.

Stats focus: local coverage

In Spokane, Washington, at the Spokesman-Review, Web site editors do routinely check site stats during the day for locally produced stories, though not for wire stories or celebrity coverage, says online publisher Ken Sands. His editors are mostly interested in tracking the performance of and reader reaction to stories that are "our franchise," Sands says -- recently, coverage of Spokane's mayor, a quadruple murder/kidnapping, and retiring U.S. Supreme Court justice Sandra Day O'Connor's day fishing in the area.

Sands says that on a fairly consistent basis, some 8-inch local-crime story played on an inside page of the paper will have one of the highest viewer counts on the entire site. Stories about rape or sexual assault seem to be especially strong draws for readers. A recent inside-the-newspaper article, "Landlord is charged with rape," became the most-read story on the Web site -- yet it didn't make page 1 of the print edition.

In such instances, when an inside print story is getting huge traffic, Sands says that the Web site staff may do a work-around so that it stays on the top of the page, when normally it would move down the page and be replaced with breaking news. He hopes print-side editors will take notice of reader "demand": "I'd like them to recognize the kinds of stories that get the most reader interest and modify their judgment accordingly. ... A lot of times, there are really compelling local stories that get buried in print because there's an 'important' government story that doesn't get nearly the readership."

Watch stats so you don't get dull

Another newspaper Web site that pays close attention to audience behavior as seen in site statistics is Israel's Ynetnews.com, the English-language news site of the big Hebrew-language paper Yedioth Ahronoth. Ynetnews.com managing director Alan Abbey says that some news decisions indeed are influenced by site stats. "I do not see anything wrong in providing readers with a story they will find interesting and entertaining, as long as we don't 'bury' the 'real' news to get it up there," he says.

Abbey says that the site does not drop important stories out of the mix if they are "dull," but editors do utilize three homepage "strips" (high-on-the-page promotional locations) for what he calls "Hey, Martha" or "Holy s---!" stories -- ones that typically get high reader traffic.

Do traffic stats on specific stories sometimes influence what's done with a story? Abbey cites the example of the story "IDF frowns on Dungeons & Dragons," which generated 147 "talkbacks" from readers. "We kept it high in the mix longer than we would have otherwise because of the strong interest," he says. "Was it that 'important'? Not really. Was it interesting and a 'Hey, Martha!' story? Absolutely. The editor-in-chief (of the newspaper) thought the story was silly."

While Ynetnews.com editors often don't have "real-time" stats in order to help them make deadline news-judgment decisions, Abbey tells his staff to create a mix of important news and interesting stuff. "We want our readers to have an entertaining, if not enjoyable experience -- to come for the sizzle and stay for the steak," he says. And that, of course, means paying attention to what readers are saying by way of site article stats.

The e-mail metric

Pageview stats for article pages and reader comments on articles aren't the only metrics worth pondering and reacting to, of course. Another important data point: most e-mailed articles. Sites such as OrlandoSentinel.com think that's important enough information that they include the list of most e-mailed stories on the homepage. The Orlando site puts this information in the upper right side, an area of the page dubbed internally as the "You Rail."

OrlandoSentinel.com editor Anthony Moor says that if he had available real-time data on most-clicked news or most-searched terms, he'd want to include that in the You Rail, too. "We recognize that the Web is a powerful place for the interplay between users and editors," he says, and such homepage features play to that.

Moor is an editor who respects and responds to the audience desires as demonstrated in site statistics, "but we are not slaves to this information, and it does not trump traditional measures of importance that news editors rely on. ... Stats are just another, new set of data that we can mine to understand better the relationship between our users and the news. ... [My editors] know that they must apply standard journalistic values to what they learn before taking action."

Beware misleading stats

Watching and interpreting Web sites stats cannot be done in a vacuum. "Editors must consider not just how much traffic certain features receive, but where that traffic comes from," warns Meredith Artley, editor of the International Herald Tribune's Web site in Paris. Sometimes, a big traffic surge on a story will come from the piece getting picked up by Google News or the Drudge Report, or being mentioned on Slashdot. So the big traffic may not point to pure popularity, but rather to the existence of a bridge or link that leads readers from one site to another, she says.

Artley also pays attention to feedback from readers. "If we see that a story is having a big impact based on e-mail received, then we don't rush to take it off key places on the site," she says. "This happens frequently with hard news and opinion pieces."

Guidance for news editors

"As I like to say, the best thing about the Web is how much information you have about how people use the medium, and the worst thing about the Web is how much information you have about how people use the medium," says Jim Brady, executive editor of Washingtonpost.com. "That information can be used cautiously or recklessly, in my view, and you have to be careful about who has access to the real-time usage data."

Brady wants his editors to be looking at usage data with the goal of building a long-term knowledge base about what works and what doesn't on the site -- applying it to news judgments over time, not reacting knee-jerk to the latest stats. "We don't want them updating the page every 10 minutes based on usage data," he says. "I don't think we want to turn our homepage into a popularity contest, but we should certainly need to have a sense of what types of content drives traffic.

"I don't mean to imply that we never jack up a story's promotion based on traffic; we do. But we do not do it on a daily bases," he says.

A great way to handle a story that does drive a lot of traffic, says Brady, is to extend the life of the story not just by promoting it longer, but by adding a Live Online follow-up discussion, creating a photo gallery or video piece to go with it, re-running the story in a feature slot on subsequent days, or buying search terms related to the story to make it more visible outside Washingtonpost.com.

Dayparting assistance

Having a deep knowledge base on reader behavior and desires is especially useful when applying the concept of dayparting -- that is, changing the homepage content based on time of day and/or the week. "Tracking peak usage periods matched to content consumption gives us interesting opportunities to remake our pages and shuffle components around," says Gerry Barker, general manager of the Columbus Dispatch's Web site.

"Conventional" editing wisdom in the newspaper business isn't set to an ongoing 24-hour newsdesk mentality; it's set to a once-daily cycle, points out Gil Asakawa, executive producer of DenverPost.com. But closely examining reader statistics can aid in making dayparting content decisions.

The shifting pendulum

So what can we conclude from all this thoughtful pondering about site statistics and the traditional role of editors in making editorial placement decisions?

Well, it's well established that news media continues to evolve toward greater interactivity -- more connection and communication between editor/reporter and reader. Web site usage statistics down to the article level are not new, though sophistication is increasing. More online news editors have real-time stats available to help guide editorial decisions; Web site user-registration databases with detailed demographic information on site users can be mapped to article pages, showing what groups are gravitating to what stories.

Editors should give up some control over the editorial decision-making process to demonstrated audience demand. Not to do so -- to stick to a past where only editors knew what was best for readers -- is hardly a wise course.

Yet the dangers in shifting the pendulum too far away from editors' judgment and more to reacting to what readers demonstrate that they want to read with their behavior are obvious. Already we have a mainstream press that sometimes gets caught up in stories that are minor in the grand scheme of things -- the Michael Jackson trial, missing young white women, etc. -- giving more space and emphasis to them while downplaying less readable and more boring coverage of socially important topics.

After conducting interviews for this article, I'm somewhat assured that editors' judgment remains tantamount -- at least for now -- and that most online news editors are treating site statistics with caution.

But the next time you see an oddball story played high on a news Web site, staying there longer than normal, and maybe even with follow-up coverage in the days ahead, you're probably seeing the effects of editors knowing exactly what people are reading.

Let's hope that online news as a result doesn't become as ratings-driven as TV and radio. That would be a shame.