Next Best Engagement Recommendation System

Home / Intelligence / Case Studies / Next Best Engagement Recommendation System

Client Situation

A large biopharma customer wanted to optimize their digital channel strategy and outreach to most effectively engage with HCPs

Trinity’s Role

  • Trinity developed an AI solution that predicts the next best digital channel, content and cadence to engage with individual HCPs
  • Although originally created for one brand, the technology and AI model was crafted so that it could be scaled and easily configured to add different criteria and datasets
  • After success with the initial brand, Trinity expanded the solution to direct efforts for 5 brands and 4 additional channels, integrating with multiple activation vendors
    and agencies

Project Outcomes

The Next Best Engagement (NBE) model implementation for digital platforms resulted in a ~22% lift in engagement (weighted) across digital platforms. This solution included:

Feature
Engineering
  • Scaling
  • Normalization
  • Missing value imputation
  • One-hot encoding
  • Embedding Vectorization
Business Rule Implementation
  • No repetition of touchpoints in consecutive weeks
    per prescriber
  • Skip list
  • Prescriber should have only one touchpoint per vendor per week
Model Training & Deployment
  • Model training
  • Model experimentation and registry
  • Hyperparameter tuning
  • Model deployment via data pipeline
  • Process logging
Model Scoring & Monitoring
  • Model weekly runs
  • Incremental training
  • Model drift detection
  • Re-training

Related Intelligence