Develop a template recommendation system to minimize the decision time of a user

Project by Doria

Abstract

Powtoon is an online video creation platform that makes it possible for people without any technical or design skills to create professional-looking videos. The user can decide whether to create a video from scratch or use a premade template. The project’s goal is to develop a template recommendation system to minimize the decision time and the time until the engagement of the user. 

Challenges

  • Selecting the calculation metric 
  • Inability to test the model before deployment
  • Normalizing the influence of the UI on the data when generating the evaluation metric

Achievements (according to KPIs)

  • Establishing an evaluation metric
  • Choosing a recommendation system model
  • Building a benchmark model
  • Wrapping the model with an API (application integration-ready)

Further Development 

  • Integrating the model into the website
  • Testing the model 
  • Adding new data features to the model

Supervisor Feedback

I’ll start with the fact that in 4.5 weeks – this project was completed from end2end (Not deployed yet, but production-ready), a true sign for Doria’s capabilities to complete tasks by a structured plan and business requirements. The project was handled in a professional way with care for the small details and tools to check data quality and impact on results.

I’m very glad for the primary results and sure that the model will be stable as we took the necessary decision to make it like that.

Maor Nativ – Head of Data & Analytics @Powtoon

Share this post

Share on facebook
Share on twitter
Share on linkedin
Share on email