The Newsonomics of the Washington Post's Reader Dashboard 1.0
First published at Nieman Journalism Lab, April 7, 2011
Don’t call them pageviews— call them pages read.
Don’t call them unique visitors — call them readers.
Welcome to The Washington Post’s new foray into understanding — and acting on — how readers actually consume digital news.
I wouldn’t quite call it a revolution. But it’s a firing shot in an effort to bring a modicum of science to the art form we editors like to believe we exercise so gracefully. It’s The Washington Post’s reader dashboard 1.0.
“Journalists like to believe that readers read every story they write,” says Raju Narisetti, one of two managing editors at the Post and head of washingtonpost.com news operations. “We’re disturbing that illusion. We’re also saying that focusing on the numbers doesn’t equal pandering.”
Readership reports at the Post, as at many other newspapers, used to be on a “need to know” basis. Now that philosophy has flipped, with the company’s management saying lots of people need to know the numbers, that business literacy is an essential part of news people’s jobs. About 120 staffers receive three hourly traffic reports, a midday performance chart, and numerous tracking reports. After the Post irons out internal reporting, hourly reports will move close to real-time. (The Post is also doing a bit of ironing with its new redesign, the introduction of which has caused all kinds of hiccups with its technology and traffic.)
The Post — with 17 million monthly visitors, growing well, and $113 million in annual digital revenues (including its Slate operation) — has long operated a model metro website. Washingtonpost.com was an early leader in pouring resources into digital, building impressive regional audience, as the site matured, under the business leadership of Caroline Little and editorial direction of Jim Brady, untethered from the Post newsroom and business mothership. The 2008 arrival of Katharine Weymouth as publisher and then Marcus Brauchli as editor set the stage for reuniting print and online. Brauchli brought in Narisetti, with whom he had worked at the Wall Street Journal. Narisetti is ushering in this new analytics era at the Post, shaking up what has been a top-performing, but often traditional, news and features staff. In the process, he is creating models of reader awareness that I think will soon be copied across newspaper companies.
We’re into an era when we can no longer play ignorant. We can decry the “content mill” methodologies of the Demand Medias, Examiners, and AOLs, but unless traditional news people understand — and apply as they see fit, working with their own long-standing news principles — data-driven knowledge of readers, they’ll lose the future. (See “The Newsonomics of 2011 Metrics to Watch“)
So let’s take a brief look at the newsonomics of the Post’s reader dashboard 1.0, what it is beginning to tell the Post, how it informs Post actions, and where it may lead in 2.0.
- The Post’s basic report forms a single dense page. It tracks the basics of pageviews and unique visitors, and breaks them down by story, multimedia, and blog, by section, and by time of day, all within an historical timeline graph. It also documents where the traffic is coming from — the referring websites — and top search terms.
- Understanding journalists’ dislike of jargon other than their own, the Post did that quick translation, calling monthly unique visitors “readers” and pageviews “pages read.” The idea: demystify foreign terms and turn them into what they are — stats any self-respecting journalist has to care about.
- Usage is tracked against goals, and these aren’t goals that grow out of dartboard exercises. The Post starts with digital revenue goals, per month and per day. Those require a certain level of impressions and pageviews, given an average cost-per-thousand ad rate. The key notion: tie together, with numbers made visible, reading and ad performance.
- The Post has merged its print and digital analytics groups, under the leadership of Laura Evans, director of marketing research and analytics. That, too, is a work-in-progress, but portends a more holistic view of the Post’s customers — and that’s especially valuable as core customer concentration comes into focus.
- Watching the logs helps direct hour-by-hour presentation decisions on the Post web and mobile sites, replacing guesswork with performance-based data. The Post noticed, for instance, that photo galleries — without sound — get better play during the day (with workplace usage), while video supplements to stories get more used in the evening. That may make commonsense, but trends pop out of data that force attention.
- Resource allocations, within the newsroom, can be better informed by data, says Narisetti. In the old, print world, as editors, we guessed or went along with our prejudices. The prejudices and guesses will remain, but the data adds support or should, at least, surface second thoughts.
Early analytics begin to tell a story of a business model in construction.
For instance, the Post now knows that less than 10 percent of its audience accounts for more than a third of its traffic, paralleling similar trends I’ve picked up at The Wall Street Journal and the Financial Times. It knows that Facebook referrals increased 238 percent in 2010 over 2009, while Twitter referrals jumped 123 percent. Following the trail, it has discerned that Twitter is best at multiplying traffic on big breaking stories, while Facebook referrals — so far converting to almost double the pageviews of Twitter referrals — will relay the trend stories better. Such knowledge can help in deciding which stories to feed most heavily to which social networks.
The Post is starting to get a handle on its core customers, as the news business increasingly groks — witness much paywall strategy — that those core readers are the building blocks of the new business, while Google-fed fly-by traffic is paling in value. Now tracked: percentages of visitors who read two-plus pages (34 percent on a recent day) and those whose sessions lasted five minutes or more (19.6 percent on that same day).
My sense is that the 1.0 dashboard is a good start, but needs quickly to meet more nuanced questions of the day. For instance, focusing on those core readers means connecting up reader profiles, derived from registration, with clickstream behavior (topics read, commercial topics researched). Version 2.0 will probably focus more on core reader behavior, because satisfying them means better ad targeting and pricing, and the ability to get them to pay for one-off digital products and subscriptions.
With the Post getting 86 percent of its unique visitors from outside its DMA (that’s high, and often runs 50 percent or greater at many metro sites), the Post will learn how these readers’ behaviors are different from in-DMA readers, and, in time, can serve them differently in story selection and presentation.
Learning how news consumption varies by platform — especially tablet and smartphone — will be key metrics to add soon.
Gathering and analyzing site data is, of course, just a beginning. Connecting that data to customer relationship systems, pay-model performance, ad targeting and performance — to name just three — is a further work in progress (see AdAge “Counting to Infinity: Why Media Companies Need More Analytics“). Consequently, we’re beginning to see a number of startups in formation to pitch their services to news companies.
Overall, it’s a recognition that data isn’t exhaust — something collected internally and warehoused, or demanded from partners and warehoused — but a new rocket fuel for intelligent decision-making.