During Wednesday’s INMA Virtual World Congress module, “Smart Data In the Shadow of the Cookiepocalypse,” five news media executives working deep in the topic of data shared their experiences and success:

Amedia, Norway 

Amedia is in a nice position with 2.4 million of the country’s 4.1 million population reading its content — and 80% of those are logged-in users. That number was 5% in 2014. 

“I would argue that data capture is critical in a modern media company and forms the basis of effective business development in journalism, subscriptions, and advertising,” Executive Vice President Pål Nedregotten said. “Any positioning by third-party between the media and our consumers threatens this.”

Amedia also has its own platform, aID, which uses some first-party data and some modeled data (based on the data of logged-in users). Additional efforts include: 

• Using a dashboard to keep track of genders in sources. Classifying 660,000 across 20 categories over 21 months, the team found only one category where women were quoted more often than men: society. 

• Creating a “churn squad” in October. This helped with a 37% growth in digital subscribers by the end of 2019. 

• Looking at advertising data from two sources: documentation and targeting. The latter often offers incorrect data. Providing advertisers with correct, documented data helped increase digital advertising growth by 7% by the end of 2019. 

The Washington Post, United States 

Based on the idea that the currency of the post-cookie world is high-quality content, Washington Post used RED, its Research Experimentation and Development group, to run tests based on three hypotheses, explained Jarrod Dicker, vice president of commercial:

1. Machine-learned contextual adjacency, when applied to high-quality content, will boost ad performance significantly … even without a cookie.
2. Readers of brand-aligned journalism can yield better results than third-party pixels, especially for attracting net-new customers.
3. High-quality journalism can, itself, be used as a performance driver, especially on platforms omitting contextual adjacency.

Using individualised experiments, they came to the following conclusions on each hypothesis:

1. Ads that ran without any behavioural context from a cookie had an overall 25% higher click-thru rate.
2. Ads performed better and reduced the amount of recall when they were content-based rather than cookie-based.
3. Content had twice the click-thru rate and the lowest cost-per-click cost.

“This is really saying … content is the future fuel of the Web, and content is the currency in both insight targeting and creative,” Dicker said. “We’ve proven that is true even before the cookie has been deprecated.” 

Financial Times, London

Tom Betts, chief data officer at the Financial Times, gave an example of why first-party data is important, especially in the time of a crisis. 

Capitalising on the news agenda is huge right now, and the more you can identify the needs of new audiences, the better results you can get. News spikes right now present an opportunity to genuinely attract new audiences. 

A yellow bar across Financial Times’ Web site advertises a COVID-19 themed newsletter and the mechanics that make it work — from registration collection to marketing — helps build a data footprint about new audiences so FT can make more informed predictions and decisions about how to engage them. 

The company didn’t just launch a newsletter, Betts said, but a whole new customer journey called “Try Before You Buy.” 

JP/Politikens Hus, Denmark 

Denmark is an Apple nation, with more than 50% of browsing traffic going through Safari. Due to this, the new ITP has had immense consequences for the Danish market. Now, no one wants to buy iOS campaigns and programmatic platforms have taken this tracking disparity into account.  

“Obviously that is a major problem because we want to monetise our whole inventory,” said Thomas Lue Lytzen, head of product development and insights at JP/Politikens Hus. 

While some try to find loopholes, Lytzen said that isn’t the right approach because it is an ecosystem that has no future. That future, in Politikens’ opinion, is that any cookie-based audience buying will be single-site, and unified standards across Web sites can expand reach. The company decided to build its own first-party data platform, Relevance, to address the cookiepocolyse in the Danish market. 

Relevance is not only used for advertising but also for content recommendations and subscription conversion efforts. It is a huge contextual motor, Lytzen said, that helps organise content into segments and create standardisation. The company also created its own audience segments and then opened a first-party to first-party collaboration with advertisers using models with probabilistic matching to avoid privacy issues. 

Politiken saw a 70% increase in audience segment sizes by utilising this first-party data. 

Aller Media, Norway

Like The Washington Post’s Zeus and JP/PolitikensHus’ Relevance, Aller Media created its own personalisation platform, Xavier.

Over the past three years, Aller Media has been on a zigzag journey through GDPR, tech, data science, KPI targets, organisation, competence, culture, and strategy. The editorial staff and commercial department at Aller Media have the same needs, said Director of Data Camilla Fuglem:

1. To present the right message to the right user, at the right time.
2. To not waste valuable space on content that doesn’t have value.

Xavier personalisation tool has three important capabilities:

• Speed.
• The models.
• Flexibility.

“Within 50 milliseconds of when a user has entered one of our pages, Xavier has run its model and is presenting pixel-perfect recommendations,” Fuglem said. “Depending on the user’s behaviour and how the models are tuned, Xavier will have pulled articles that the user has already read and selected the best articles to present based on popularity and other models.”

Xavier can also perform A/B testing. “Having full control means we can quickly change and adapt when we have new needs occur and new sites or products are deployed, or we need to use new models.” 

Xavier is used for editorial content: The system strikes a balance between automation and manual control. Personalisation is something Aller believes the editorial side needs to take part in, by defining what kind of content Xavier can use in its recommendations and not show content that hasn’t been approved by the editors. Single articles can be added or removed from this “safe pool” as needed. This builds trust and control.

It can also be used for commercial ads: “For us, the content marketing is integrated in the CMS,” Fuglem said. “We saw that when a user has seen a [piece of] content three times, there is a very good chance they will click on it the fourth time. So then we asked ourselves, what should we show them? And we started to define some rules.”

Fuglem said personalisation has been an interesting journey that has seen results, including:

• Commercial + editorial + development department = success.
• Ownership of data resulted in added value for Aller  Media, its users, and its advertisers.
• Gives the ability to do what they want with their pixels.

Xavier personalisation is now widely used across most of Aller Media’s titles. 

The Congress continues on Friday with a module on the content-to-commerce revolution. Register here for individual sessions or the entire Congress (the latter includes access to this and previous sessions).

Banner image courtesy of Clem Onojeghuo on Unsplash.

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