Psychographic User Segmentation Based Model to Overcome One-Size-Fits-All

16.09. | 09:30 - 10:00 | data.stage

XING has a vast data pool of user activity of their 16 million registered users and their activity on the site, but as many other companies struggles to extract the desired value out of the large available dataset. In this project XING used a three step approach to increase relevance of its content across marketing and product development:

  1. Large quantitative survey to reveal psychographic motives of active users on site
  2. Creation of machine learning model with identified users as training set with defined motives and activity on site
  3. Application of learnings onto remaining user base and subsequent adaptation of marketing and product development to improve user activity and satisfaction.

The presentation will walk through the project development steps (what was the model solving for), how the model was set up (parameters, applied statistical model, etc.), what results came out of the machine learning model in terms of uplift and statistical output and finally discuss how the model results impacted downstream activities at XING in both our marketing and product departments.


Marc Roulet (XING GmbH & Co. KG)
Michael Horn (XING GmbH & Co. KG)