To maximise value from data, news organisations must create systems that facilitate staff to access and understand it, as well as a cultural environment that allows teams to act on that data.

During the recent INMA Smart Data Master Class, media leaders from News UK, Financial Times, Condé Nast, Mediahuis Belgium, Süddeutsche Zeitung, and NZZ shared how their companies are building structures that support a culture of data and using that data to create better experiences for their audiences.

News UK creates a single customer view

As director of data strategy at News UK, Will Sach is tasked with looking at how the company can gain value from its data.

In recent years, the company has invested in a data programme of new assets, tools, and capabilities that allow for growth by increasing volume and value, Sach said: “To do this, we really need to first understand who our customers are, how we can best respond to them.”

And that led to developing the single customer view, or SCV, to be able to create a unified data profile for users. Sach said they hypothesised that data “becomes disproportionately more valuable when it connects with other data.” By creating the SCV, they hoped to improve their understanding of customers, their experience, and their overall value to the business.

“Our data vision really was about how to move away from siloed product interaction to connecting all those devices, those products, those propositions to specific customers,” Sach said. “And when we’ve got that, how do we then use that information to grow our known and active user base?”

Several data and business challenges stood in their path, including missing data, low scalability of data systems, inefficient marketing activities, an incomplete picture of customers’ activities, and, as a result, very little understanding of who their best customers were, Sach said:

“That leads to this big question about how much value were we leaving on the table [by] simply not understanding our customers, not understanding our need to be able to respond to them.”

Financial Times defines metrics that matter

Metrics That Matter (MTM) is a company-wide initiative that “gives all departments a shared framework and common language so the right numbers are used to achieve the FT’s goals,” McKinley Hyden, head of the analytics business team at the Financial Times, said. “It’s almost like a translation service.”

The team laid out some guiding principles in the beginning, including the desire to “reduce focus and pressure on the North Star” by not trying to force one number to work for everyone. One key guiding principle was to use existing metrics wherever possible.

“The FT is an intellectual place that’s prone to overthinking, and with every new initiative someone wants a completely bespoke metric,” Hyden said. “That makes it difficult to make decisions across a portfolio of projects and initiatives, and to streamline decision making.”

The MTM objectives are grouped into four main categories, a critical one being focused on maintaining processes and governance to ensure the framework stays relevant.

Hyden admitted that that objective “is the bit that always gets a bit of a yawn, but it’s so important.” One way the FT is ensuring that the framework is maintained is by creating a data catalogue so anyone can search for a metric or data point to find out its history, provenance, details, and quality, Hyden said: “It’s a hugely ambitious project.”

Condé Nast finds — and shares — its data story

Not only do news organisations need a good understanding of data itself, they also need to recognise how the definition of data may be vastly different depending on who you ask and what department they work in, according to Mehul Shah, vice president of enterprise data and analytics at Condé Nast:

“Data has a very kind of raw, meaningless version of it, but when you put that into the right context and when you organise it in a much better way, then it becomes a powerful asset.”

Getting teams to access the right data and right tools, making the right data available to the right people, and changing the data culture in an organisation is the ultimate goal. To do any of this, Shah says an audit of what that means for a specific organisation is needed since each company is different.

The final piece of the data puzzle from Shah is being able to turn the data into a story to communicate to executive leadership. The data story should be focused on big picture opportunities for the business.

“We started talking to as many people as possible, creating as many use cases as possible, and then started thinking in order to solve all of this and potentially future proofing ourselves, what we should be thinking,” Shah said. “Unless you’re creating business value, nobody’s interested in what you’re doing.”

Mediahuis Belgium creates internal innovation to get data to newsrooms

Yves Van Dooren, data science business partner newsrooms for Mediahuis Belgium, said journalists don’t need more analytics; they need more empathy.

The company adopted a design-thinking mentality centered around empathy, going into the newsrooms and interviewing journalists and editors to find out what they thought of the current dashboards.

“Most of the time people just answered, ‘I just don’t understand those screens, it’s way too complicated. Just give me a list. I don't need all those numbers,’” Van Dooren said. “And I just couldn’t understand why people were so negative.”

But this feedback led to the creation of a series of prototypes to meet journalists’ needs, resulting in a newsletter that would provide that same information so they wouldn’t have to log in to a dashboard to check story performance.

“Empathy was the key and empathy is something very specific,” Van Dooren said. “It’s not just knowing how people feel about something, but feeling how they feel. And this really is a key element that you need to learn.”

Süddeutsche Zeitung uses personalised paywalls

When news media companies talk about personalisation, content is typically the first thing that comes to mind.

But Carmen Heger, head of data at Süddeutsche Zeitung in Germany, said personalisation means so much more: “There are multiple scenarios and methods for personalisation that go beyond content recommendations.”

Süddeutsche Zeitung is using personalisation at various touchpoints to increase conversions, subscriptions, and retention. Heger presented four scenarios to demonstrate how SZ uses personalisation, one of which is focused on subscription type.

The subscription model asks what type of subscription package is right for each user. News media companies offer different packages, but SZ wanted to know how to determine which offer to show. Initially, it showed a more expensive offer to those with a higher propensity to purchase and a less expensive offering to those less likely to purchase.

In its second version, however, SZ used a two-stage approach to personalise the offer, using an XGBoost algorithm to compute data on known users.

“Where we have historic data for those, we compute this once per day, and we basically predict how likely a user is going to subscribe if we show them one of these four [offers],” Heger said. “It finds the most suitable offer according to the user behaviour and increases conversion.”

NZZ uses data to create personalised news feed options

Personalising content is the name of the game for media company NZZ in Zurich, Switzerland. They’ve been at it for about six years and lean on a combination of data and machine learning to build successful and responsible news recommendations.

It’s all about balance, Cristina Kadar, the company’s data science and machine learning product owner, said. It’s important to collect data and insights but also think critically about applying it in a way that makes sense for their brand.

For its daily newspaper, NZZ revamped the “next read” section. This was a section of the newspaper that was underperforming so they developed a fully algorithmic driven “next read” section on their article pages. Over the course of a year, the team did several A/B tests with both algorithms and feature improvements.

At the end, they created three different feeds: One sources articles from the same author and fed those to the user, another provides a topic feed that sourced articles with the same topic the user was reading. The third is a user feed where they went heavy into one-to-one personalisation.

After the first stage of the revamp, NZZ saw a 48% increase in weekly average click-through rate and a 61% increase in average completion ratio. After the second stage of the revamp, which gave the changes a little more time, NZZ saw final results of 73% increase in click-through rate and 78% increase in completion ratio.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .