The native advertising jungle is a place for a wide variety of widgets and formats. There is one factor that affects all areas and that is at the heart of the industry: knowledge. Theodore Meynard, Machine Learning Engineer at plista, and his team have developed the widget, “Knowledge Tree”, as part of the Xaxis accelerator program, Xcellerate 2019, which addresses this factor to solve one of the most urgent problems in the native advertising market: cookie fatigue. In this interview, the programmer reveals how advertisers, publishers, and users benefit from the “Knowledge Tree.”

How did you come up with the idea for the Knowledge Tree?

I had the original idea during a hackathon with the DPA last November. One of their teams used a DPA-API to view content-related articles. I immediately saw the potential for native advertising. We don’t even need an API, we can use text processing algorithms to display related topics. This is where our team – Mohammed Elhakim, Huseyn Gadirov, Nurshesari Budiasriati,and Abdul Rehman Liaqat – found the inspiration. My colleague Mohammed then had the constructive idea of developing an interface that displays and makes accessible complementary topics in order to make information more easily available on websites. This laid the foundation for the Publisher Widget idea, our “Knowledge Tree.” When we heard about the Xaxis Xcellerate 2019 accelerator program, the timing was perfect to bring the idea to life. We are now in the second round of the program.

How exactly does the Knowledge Tree work?

At the core of our widget is the creation of autonomy for the user while at the same time making use of the information a user provides during his/her visit to the publisher’s website. This touches on two essential factors of native advertising: autonomy and knowledge. To enable the creation of autonomy and the acquisition of knowledge, we implement our decision tree, which shows topics related to the content of the article under articles on publisher websites. The user is enabled to inform him-/herself about topics that are not shown in regular widgets but are relevant for the user. This way, if you like, the user gets full digital content control and autonomy. The user is not presented with content that has the highest CTR or articles that have been read particularly often, but  rather with content that has the highest relevance for the user. The more information we collect on content relationships, the better the widget will work. In addition, there is, of course, an algorithm that can make predictions. All in all, the system is based on an open source model, which we have further developed and which searches in text sections for topics such as people or places. The goal is not to filter all topics out of the content, but only the relevant ones.

How exactly is this different from other widgets?

We go one step further into the thematic depth and don’t play similarity. Take, for example, an article about a football match. The reader may not be interested in more sports news or football content, but rather in a trip to a particular country or even the politics of that country. So, it’s not content of the same kind, but content that informs beyond that. If we only play out recommendations on the same topics, we miss out on interesting starting points. The target level changes here. This is the decisive difference to other systems. Although they also look at individual content that is played in an article, it cannot determine the main interests of the users. In addition, they are quite expensive for the publisher.

How do publishers benefit from the widget?

Monetisation on three levels. On the publisher side, the widget primarily helps to generate insights about which topics are of interest to the reader, which enables the user to better curate content and of course,  the shown ads. So, we see several benefits that pay for different revenue streams of publishers: 1 ads directly on the widget, 2 ads shown in other widgets, and 3 increased subscriptions. The first point targets the usage of the data insights. They help show exact ads related to the users’ current interest, which leads to a better performance. The aggregated feedback of the users can be used to identify the average interest related to the article. This information can, in regard to the second point, be used to improve the ads on the other widgets by utilizing it as insights for setting up more precise bids. The third point relates to the fact that the user benefits from a better experience and that the feeling of autonomy on the site creates a good feeling towards the publisher. Hence, the user is more likely to stay on the website, as it displays further internal content and establishes a trustworthy connection to the publisher – the first step to become a subscriber.

What are the benefits for advertisers?

Advertisers face the challenge of adapting their content to the interest of users without having much information about what they actually look like. This knowledge gap is closed using the Knowledge Tree. Ads that are placed in the environment created by the Knowledge Tree thus meets more interested users. The probability of conversion increases. Targeting becomes easier overall. This is particularly advantageous with regard to the programmatic playout of content. The advertiser can use the collected information from its AdTech provider to participate in suitable bids. Therefore, the right budget is used in the right place. In this publisher environment, it makes a lot of sense to place ads that go live on content. This includes, for example, the Native Advertorial. Within our capacity as consultants, we can also provide advertisers with better advice on which topics they should concentrate on in terms of content.

What about data protection?

More and more users want personalized content without tracking. We solve this problem. The clue of the widget is that it can work completely without cookies. Hence, there is no problem with data protection. This is in line with the current trend that users do not want to be tracked digitally during every step of the way. We don’t need cookies because we play ads and articles with the help of collected information on the interrelationships between topic interests. The tracking of users is therefore superfluous. At the core is the feedback of their interests. One can imagine this as a continuous market analysis, where the respondents remain anonymous. At the same time, we meet the user’s demand for personalized content. However, the content is not personalized for the user, but individualized for the interests associated with in an article. This also creates the advantage that new content is displayed to the user. Where the user has always come across the same content but now, a whole new world of information awaits the user, stimulating curiosity, which in turn encourages the user to more likely convert.