Play Sports Network delivers personalised content to its users

play-sports-network-vector-logo-600x333 (1)-1

Play Sports Network is the world’s largest cycling media company and community, reaching and engaging with more than 40 million cycling fans and riders around the world.

About the project


Play Sports Network (PSN) wanted to increase user engagement of their Global Cycling Network (GCN) social media app, allowing like-minded cycling enthusiasts to upload, share, comment on and consume content.


Ancoris was approached by PSN with a complex machine-learning problem, to develop a collaborative based recommendation engine to deliver personalised content feeds to users of the GCN app. Working closely with PSN’s data science team, Ancoris data and analytics practice delivered a leading-edge matrix factorisation model, trained using BigQuery ML. Ancoris also delivered reusable, high-value data assets, including a Single Customer View (SCV) and an engagement scoring algorithm that PSN will continue to leverage in future optimization.

GCN Lone cyclist-1


  • Users of the GCN app were all being shown the same content, with only limited personalisation based on user characteristics such as the user’s language.
  • Personalisation was further complicated because users didn’t have a unique user ID across PSN channels or even within the GCN app. Their activity may be recorded under different user IDs, according to whether they’re logged in or out of their account on the app, or if they switch device or network.
  • PSN’s in-house data science team had begun building a content-based recommendation engine that would filter content based on selections made by each user. They quickly discovered the high-complexity nature of the problem* and turned to external support to help accelerate their progress and fully leverage the capabilities of the Google Cloud Platform.
Play Sports Network Cycling team-1


  • PSN has a leading-edge recommendation engine to serve highly engaging content. The solution includes:
    • The very latest machine learning technology via BigQuery ML
    • Hyper-scalable storage via BigQuery; anticipated to scale to over a trillion rows in the matrix output table alone
    • dbt (data build tool) for easy productionisation of the ML pipeline using SQL
    • A blueprint for a modern data analytics platform that they can take forward for future use-cases
  • A Single Customer View (SCV) that consolidates different user IDs belonging to the same individual across all channels. The recommendations generated by both the content-based and the collaborative engines are more accurate. Reporting at marketing department and board level has also improved in accuracy.
  • An engagement scoring algorithm that uses a stream of activity data from users’ phones to provide an accurate assessment of how engaging users found each piece of content they interacted with.
  • Up to one year of work saved over the course of the three-month project. The combination of the technical infrastructure and the coaching and mentoring delivered throughout the project has set the PSN team up to now carry the work forward independently for the next two years.

Want the full story?

Read blog by David Taylor, Technical Lead at Play Sports Network for all the details on their collaborative-based recommendation engine.

Sign up for our newsletter to get updates on projects, technologies and more.