When GDPR came along in 2018, it was the first time users’ wishes about their data privacy were taken seriously. This had an immediate impact on the Internet ecosystem, according to Ian Hocking, vice president of digital at South China Morning Post in Hong Kong.

“We saw companies closing almost overnight because they could no longer keep up with the governance for doing business in Europe,” Hocking told INMA members at the Asia/Pacific News Media Summit on Friday. “Publishers and marketers had to try to keep step with it, and build tools and services that enabled them to ensure data privacy.”

GDPR regulations were really guidance, and Internet companies really should uphold a standard when it came to data privacy, he told summit attendees. This however, led to different interpretations and ways of handling privacy.

“That meant that, after that initial jump, it kind of stalled. It didn’t really have the teeth to create a significant and profound change to the way that people experienced the Internet.”

Big Tech platforms, however, did create that kind of change on a worldwide scale. When platforms such as Google and browsers such as Firefox and Chrome began doing away with third-party cookies, that was a move Internet companies and publishers couldn’t ignore. So they started a new push towards user privacy.

“Platforms really do have a global impact,” Hocking said. 

In early 2019, South China Morning Post (SCMP) decided it wanted to take hold of the situation by building a better framework to understand its users. It planned to use that framework across the business to create better outcomes for both the company and its partners.

“We moved to first-party cookies, and we built a profile that was based on those IDs that were first-party to us,” Hocking said. “And we moved to only using data created on our platform. We managed to make that switch in about four months, and the biggest thing we did in order to enable that was we moved from a traditional DMP to a CDMP — one that enabled us to use first-party cookies and create profiles around that.”

At the same time, SCMP decided to build its own first-party data platform, called Lighthouse, which is at the terminal of every person who works at the company.

Google’s recent decision to delay its third-party cookie removal will mark a profound impact on a lot of the ID alternatives on the market, Hocking said.

Key benefits of building a self-service data platform

Hocking shared the main reasons SCMP decided to build Lighthouse:

• All SCMP staff have full access, with no chance of damaging the data or DMP.

• It’s self-service, with increased speed and efficiency for staff and clients.

• Consistent presentation of data — everyone sees the same data and talks about it in the same way.

• Control and measurement — they know how the platform is being used and how to improve it.

• Bespoke intuitive UI that was made to present SCMP data with context.

• It’s modular and proprietary, and the company is able to move quickly to provide new services.

• It presents a unique market offering that creates a reason for direct conversation with marketers.

• SCMP sees a greater return on investment for clients. Campaigns are led by insight and optimised to generate stronger ROI and build better partnerships.

“We aren’t alone in this,” Hocking said of building an in-house data platform. “In fact, 68% of those surveyed [by INMA] said they were building their own first-party solution, which is amazing and speaks to how seriously the industry is taking this.”

This means there are likely to be many more publisher walled gardens in the near future, as they take control over their own first-party data, he said.

What is the SCMP platform Lighthouse designed to do?

In the past, SCMP used third-party cookies, while today the company uses first-party cookies, allowing them to build much larger and more complete user profiles.

In the future, he said SCMP will build out authenticated login ID to further profile users with first-party data. This will allow the company to really understand how to build a better value exchange for customers.

He shared the type of first-party data available in Lighthouse:

• Preference: A long-term attribute unlikely to change.

• Opinion: Current way of thinking, which might change based on circumstances.

• Sentiment: The positive, neutral, or negative feeling about an article.

• Intent: Declared action to do something.

• Behavioural: A predictable, usual way of being.

• Interest: Declared passion for something.

“When we put all of that together in a profile, we start to build what we think looks much more like a person,” Hocking said. “I think publishers that have a great, ongoing relationship with audiences to really benefit and present something interesting [comes from] building out not just an ID with an attribute attached to it, but a broad profile that really talks to them as individuals.”

This type of data includes psychographics, is altitudinal and behavioural, and builds out the profile in the context of a network. SCMP’s Lighthouse launched in June 2020 and was the first publisher data platform in Asia.

“It’s designed for us to create a space to understand our users,” Hocking said. It’s usual for people to consider such a platform as a way of driving clicks, SCMP teams do not think of it that way. Rather, it’s about identification and building robust profiles for actionable results, such as great ad personalisation, he said.

“You get to ad value as a publisher by demonstrating to a client where you can find them prospects. They want to understand how you can find them new customers, and by having really great in-depth profiles that aren’t just a data point, you can do that very well.”

This can’t be accomplished nearly as well by creating probabilistic audiences through third-party data. For example, a user might read one article about food, and then get put into a probable audience segment of foodies, and perhaps served a lot of content that he might not be interested in. The data is anonymised and presented as an aggregate segment.

“This usually has a very low on-target percentage,” Hocking said. “As a buyer, it’s going to be very difficult to understand what that is. There’s a lot of obfuscation that happens.”

There is usually only 30%-45% confidence in such a segment, with the user IDs typically hashed, anonymised, aggregated, and presented as one big segment for usage. This makes it difficult for ad clients to understand the data they are buying.

An effective first-party data platform, on the other hand, provides much better, more precise segmentation, with an 80% or above confidence in prediction.

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