The Newsonomics of “Little Data,” Data Scientists, and Conversion Specialists
First published at Nieman Journalism Lab
OSLO — Arthur Sulzberger surprised some people recently when asked what he would do differently in the digital transition, given hindsight.
It wasn’t an off-the-cuff comment. Each FTE is precious at The New York Times and at every other newspaper company these days, and the Times is indeed spending more than most on engineers and data scientists to build out its knowledge about its business.
It’s the biggest companies that are making the biggest investment in analytics; the gulf between the national/global and local yawns wider in the digital age. The Financial Times, long a leader in analytics, is making new pushes, and Schibsted, arguably the most digitally advanced big news company in the world, is investing substantially in customer intelligence.
Though its recent announcement of still another major global partnership — doubling down in Asia and Latin America — caught notice in the industry, it’s its huge push in analytics that should be getting more attention.
Schibsted is part of a quiet revolution in consumer intelligence. The future goes to the smartest, and the shift is spawning job titles never imagined in the news trade — like news data scientists and conversion specialists. And the politicization of data use — with all its public policy, privacy, and anti-competition implications — grows every week. Last week, Sen. Ed Markey questioned recent announcements by Microsoft and Google that they could track digital customers across web, smartphone, and tablet platforms. Such tracking will unlock a whole new level of consumer insight — and ad targeting. It is the Googles, Microsofts, Facebooks, Twitters, and Yahoos who have built their own data advantages, based both on huge usage and investment in analytics.
News companies — even the biggest and smartest — are playing catchup. They look to mine the deep data they have about their own visitors, from through-the-day news usage to shopping and classified buying behavior. They hope to use that to build competitive positioning against the web behemoths.
We might have thought that progressive Schibsted would be farther along in the data sciences (“Looking to Europe for news-industry innovation: Schibsted’s stunning classifieds and services business”). Long-time Schibsted strategist Sverre Munck says that despite the company’s great successes and acute reading of changing consumer behavior, it still felt like it didn’t know what it needed to know.
“Our analytics were haphazard, ad hoc, case-by-case,” says Munck, Schibsted’s soon-to-be-retired executive vice president.
Where Schibsted believes its unique strategies are right — separating out digital businesses from print, growing whole new global classifieds businesses, investing in financial consumer transaction sites — it invests heavily. It’s now doing that in analytics.
Edoardo Jacucci, a McKinsey grad, has been hired to lead Advanced Data Analytics. Putting together an initial team of five, Jacucci’s mandate is to grow quickly to 20 and then to 30 by the end of next year.
Jacucci is all about actionable analytics. He ticks off how his team will directly impact the business, using the analogy of funnel traffic flow — you know, all the various sources of web traffic that flow into the top of news company’s wide-topped funnel, with too little of it converting to paying digital customers:
- Predictive analytics
- Recommendations engines
- Community building
- Conversion optimization
- Reduction of churn
As with the FT, which paved this road in newspapering, the Schibsted analytics operation is the polar opposite of the traditional newspaper research function. It’s not about creating reports and PowerPoints. “It’s about turning data into useful insights until you get tangible results in the business,” Jacucci told me in Oslo, the company’s headquarters. “We want to set up a solution delivery team. Given a business question, we need to turn it into a solution quickly.”
This new news science is all about how we properly mate humans and data. Much of data science would be without value without this simple truth, as summed up by Jacucci: “It always starts with the right business question.”
Schibsted’s ADA crew will be based in three cities: hometown Oslo, Stockholm (where it has strong operations), and Barcelona. The Barcelona hub will work with Schibsted properties in southern Europe and globally.
The “global” part isn’t aspirational. Its just-announced partnership with Telenor, a major European telecom company, and Singapore Press Holdings, aims to grow its footprint in both Asia and Latin America. The two new companies formed out of the joint venture will exploit classified opportunities in Latin America and Asia — and provide new competition to publishers in those populous markets.
Schibsted already operates in 29 countries. From France to Argentina to Indonesia, classifieds has been its growth engine. The results are impressive and industry-leading: Classifieds supply 25 percent of the overall company’s revenues and 52 percent of its EBITDA, with operating profit growing 10 percent annually.
Given that success and Internet orientation, the digital businesses now drives about 45 percent of Schibsted’s revenues — and 61 percent of its profits . Among its peer companies, it is closest to that magic crossover point (“The Newsonomics of Crossover“), even as it struggles in the old business, growing news revenues just 2 percent a year. (Still, that positivegrowth puts it at the head of its class.)
Farther west, the FT’s analytic capacity steadily builds. I’ve been tracking the company’s leadership for five years (“The FT as a News Company of the Future“), from the time it first replaced its small traditional research team. Now, it funds 30 staffers, 20 on its analytics side — making use of customer data — and another 10 “data experts” who develop and maintain analytical systems and databases, says Tom Betts, head of data analytics.
The New York Times is also building on a solid analytics basis. When Marc Frons moved up to the chief information officer role last year, he created a Business Intelligence team, pulling from disparate groups as well as building out. BI now counts around 30 developers, data scientists, and managers, more than double the numbers of a year ago. In addition, the Times has a separate Customer Insight Group, reporting to marketing, with another two dozen staffers, mainly analysts. There are more data scientists in other departments.
In Frons’ view, it’s all about implementing a single vision of the value of the data, organizing it properly for accessibility — and then decentralizing: “You want to decentralize the ability to use the data.”
We can trace a lot of the fast-growing appreciation of data value to the Times’ planning and execution of its paywall. “The paywall feeds customer data, not just anonymous users. We’re moving up the value chain,” says Frons. Now as CEO Mark Thompson talks about “working the engagement curve,” expect that whatever products tumble out next spring will be based significantly on what the Times has learned over the past several years.
At the FT, Tom Betts has seen the fruits of such data analysis, both intuitive and counter-intuitive. The perhaps intuitive, proven: “In excess of half of allsubscriber consumption now happens via a smartphone or tablet. These new channels have driven significant increases in consumption, but this change is also driving peaks in consumption where we have not historically had them. Add into the mix the global nature of our audience and distilling the needs of our customers becomes challenging.” So the understanding helped drive one of the newest FT products, fastFT, which provides fast-paced market-moving news and insight 24 hours a day. (On the other hand, in print, the FT last week announced it would be cutting down to a single global print edition for the entire world, with editor Lionel Barber saying in the future, “our print product will derive from the web offering — not vice versa.”)
The counterintuitive: From data log analysis, the FT learned that staggering numbers of new digital subscribers didn’t register for free before they became paying for a subscription — they came to our digital products with little or no apparent digital footprint and went straight into a purchase. “This insight destroyed the concept of a traditional, linear ‘sales funnel,’” says Betts. “Prior to that point, we’d focused our subscription marketing on those users that had registered and we deployed our data intelligence — like developing propensity models to determine who to communicate with and about what — via email to existing registered users. This allowed us to extend the audience we could reach with our marketing to those that had not yet registered.”
We can distill four lessons out of this major, if almost subterranean, data science movement:
- The mystique of the data scientist. It might seem funny that “scientists” are entering the grungy trade of newspapering. Two-thirds of the staff that Edoardo Jacucci is hiring at Schibsted are data scientists; how does he describe the role? Two of three have PhDs, and all at least have a master’s. They may be statisticians or engineers, but importantly they “can reason at a meta level.” Marc Frons provides an analogy that’s useful for those who’ve been in the web business for a while: “‘Data scientist’ looks a lot like a ‘webmaster’ in 1993.” In other words, it’s a term to describe a set of skills — comfort with data, rigorous analysis, ability to direct machine learning exercises, and more — that most people in your company don’t possess.
- What’s the faith of a “conversion specialist”? Surprisingly agnostic. The web has produced hundreds of thousands of people with e-commerce experience, storytelling chops, and programming skills. It’s some blend of those three — with that intangible of “understanding the business” — that makes a good conversion specialist. Conversion, as in turning visitors into paying subscribers and occasional readers and advertisers into regulars, among many other targets. A third of the new Schibsted ADA staff will be “conversion specialists.
- Three is the magic number. Each of the three companies noted here organizes their analytics staff differently, but there’s an essential common denominator to make it work. These expanded analytics teams must work closely and collegially with both IT staff — who can make things happen — and business management. Building core competence in that teamwork is what will separate success from failure.
- Little Data trumps Big Data. Big Data we know. Technology companies advertising on Bay Area freeways and in tech-aware airports cite their ability to tame it. It’s the bogeyman that seemingly only the biggest companies can possibly take on. Little Data, though, breaks apart Big Data, just as anti-procrastination experts urge us to break big projects into smaller tasks.Explains the FT’s Tom Betts: “I think Little Data is an enormously valuable concept. There is so much hype that surrounds ‘Big Data’ and the suggestion is that there is a correlation between size of data set and business value, which I just do not think is true. You are just as likely to find key audience insights in small data sets as you are large ones.“There is an obsession with building tools to do things as fast as you possibly can, and much of the “Big Data” technology is focussed on facilitating that. I think scale is an important topic in data analytics — the sooner we can act on signals and triggers we see in our data the greater our chances of success, especially in marketing. But that is just one side of the story. For me, empowering others to use data to make better business decisions, however small, is more important.”
Done right, this news science revolution offers the promise of faster cycle time — creating new products and services much faster than the glacial speed that is the news industry’s legacy. As I’ve advocated “selling more stuff” as one key strategy to revenue growth, I’m often asked: Sell what? The data guys have their own answers. Properly implemented, they could change the nature and speed of product development.
“We want to insanely accelerate innovation in the business by implementing data-driven features,” says Jacucci. “We have to turn customer analytics into features on the website. That will give us the right to call ourselves a digital company.”