Integrating and Automating the Process

Data Integration & Automation

Whenever several sources of data are available for segmentation, the best strategy consists of combining them at once using an adequate machine learning algorithm. We have devised effective consumer-focused segmentation based on integration for clients so that they can penetrate different markets, effectively allocate resources to maximise ROI, and develop a competitive advantage.

Machine Learning for an Optimised Segmentation

During the last decade, machine learning has become the de facto standard for exploiting data in the most efficient way. The general principle consists of feeding algorithms with information (data sources) that has been prepared during a data engineering phase and fine-tuning them (and testing various options) in order to uncover complex hidden patterns. The fields of application are plentiful and customer segmentation is one of them where the approach can be used very successfully for a variety of tasks such as:
integration of several data sources
  • to assign HCPs to segments
  • to build an optimal list of target Drs in a specific context (therapy area, new product launch)
  • to understand what drives specific HCP characteristics
  • to create reliable indices that measure specific behaviour or importance of HCPs

Anatomy of a ML Engine Project

Identifying Data Sources

  • Existing data
  • New sources
  • Calibration Data

Data Engineering

  • Build data from your knowledge
  • Extract meaningful features from all sources
  • Clean/recode, etc.


  • Train the algorithms
  • Fine-tune them
  • Validate their knowledge/performance
  • Select the best approach
  • Prepare the selected approach for usage


  • Generate potential/influence scores
  • Investigate important patterns
  • Measure the current performance of your segmentation
  • Determine the value of each data source


  • Include updated data
  • Add new data sources
  • Re-train engine

Benefits of an Integrated Engine


Update input and generate new output as needed


Get the most accurate predictive model with your data


Any data source can be included - Engine can be easily upgraded


Know the value of your segmentation - Understand what drives segments