How to Stay Relevant When Personal Data is Not an Option

By Dan Willstrand May 09, 2018

Tags Publisher

Reading 2 min

Content recommendations can be found everywhere today: Netflix recommends movies that their algorithms believe you should watch, Tripadvisor automatically pushes restaurant tips when you’re on vacation, newspapers suggest articles you might like and Facebook, well, their feed is nothing else than an endless stream of personalized content recommendations.

Recommendations provides a value

The question is: Why has this recommendation frenzy become so widespread? The first, obvious, reason is that the companies giving them make money on it. They drive consumption of content and services that people would otherwise not be eager enough to find on their own. The second reason is that recommendations really have a value to people. Since the the amount of digital content has exploded, it’s quite exhausting to search actively for everything you need. Intelligent recommendations is like having a close friend do the searching for you in advance and then whisper the best suggestions in your ear.

GDPR has a big impact

Understanding how these systems work has become utterly important for media owners since the GDPR will have a direct impact on how you can use them. Let’s try to explain this simply. Recommendation engines generally function in two ways:

The first is a content-centric approach, which is how Strossle works. The engine is designed to interpret and categorize content, and match it with other content that it thinks should be relevant, based on historical, generic consumption patterns. Example: people reading articles about bananas are likely to be interested in articles about health or exercise. Not more articles about bananas. The capability to give accurate recommendations is then continuously improved through machine learning.

The second approach to recommendations is called collaborative filtering, and is more user-centric. The engine first looks at what it knows about a user, and then at what similar people like. Such systems can be effective even before the user has provided any feedback through actions (clicks). The problem with user-centric recommendations is that they need personal data, which will be much more difficult to collect as the GDPR kicks in. Saving personal data calls for an active consent, and the billion dollar question is if users are willing to give that. For example, in a recent survey by PageFair, only 3% of user said they would be willing to give consent to tracking for the purpose of advertising.

The future is content-centric

The key takeaway is that publishers and advertisers who want to continue providing their users with great recommendations, without demanding consent, should deprioritize systems based on personal data and focus on content-centric solutions.

Remember: the GDPR was designed to protect people’s privacy, which in the wake of Facebook’s data crises seems more urgent than ever. Ask yourself: do you really need all the data that third-party solution providers collect about your users? Or can you work in a more privacy-enhancing way?

Don't miss: A checklist on what you should investigate before choosing your recommendation engine

 Checklist for Selecting Content Discovery Platform Get my free checklist