German publisher Axel Springer is using machine learning to fuse disparate data sources and create three-dimensional “maps” that cluster users around likely interest groups.
Carolin Bink, Axel Springer’s head of data products, addressed this topic at a recent event in London where she explained that the publisher had found itself hamstrung by a combination of lofty ambitions and inadequate data collection practices. […]
The publisher partnered with Swiss-based data management platform (DMP) 1PlusX to use machine learning to analyse up to 1.8 billion ‘events’ each month – the company receives visits from its users on average twice daily across two brands.
“We can see very clearly is [the user] searching for car insurance? Is he more interested in politics, or is he a soccer fan of a specific club?” said Bink.
One of the quickest wins of the 1PlusX partnership has been enabling the publisher to monetise those interests through keyword targeting.
Content focused on German football team Bayern Munich, for example, received 7.6 million views over a 90-day period, while Axel Springer titles have produced 4,380 articles related to Brexit, attracting 2.8 million readers over the previous 30 days.
The publisher has created more than 400 audience segments in the last two years, with particular granularity in the auto category.
By eliminating inappropriate articles through recency and user visit frequency search, Bink said Axel Springer can help brands to access highly relevant audiences.
World Advertising Research Center (2019): Axel Springer builds audience segments with machine learning. Online verfügbar unter https://www.warc.com/newsandopinion/news/axel_springer_builds_audience_segments_with_machine_learning/41966, zuletzt geprüft am 27.11.2020.