Our client is a specialist private equity firm focused on 5 core sectors – software, e-commerce, financial technology, healthcare, and digital infrastructure. The client wanted to leverage data science on traditional and alternate B2B data to identify the good fit companies and help prioritize the leads.
The client wanted to
- Build models to identify good-fit companies from the incoming leads
- Score companies to rank and prioritize for due diligence
- Lack of historical internal data on prior deal evaluations
- No set rules for evaluation of a company due to subjective nature and different viewpoints of investment teams
- Sparsity and low reliability of data for smaller private companies limited the ability to go granular
- Created the right definition of dependent variables and chose the right set of independent variables from traditional and alternate data available.
- Derived multiple interaction attributes combining data from different sources
- Analyzed data and patterns at different levels of granularity (Sector, sub-sectors, size, etc.) and created logical groups for final model build.
- Chose a combination of data science algorithms to build explainable models for business users
- Two models were built-
- To be used at initial stages to identify good-fit companies;
- To be used pre-due diligence for prioritization
- Integrated models with Visualization layer to provide scores and insights in near real-time.
- Set up model performance monitoring framework to track model effectiveness.
- Models achieved more than 70% accuracy in predicting good fit companies; Top 30% selection captured more than 80% of good fits
- Models provided quick evaluation scores of companies with key insights, which helped prioritize incoming leads thus providing First mover advantage
- Augmented existing financial models by providing additional insights based on alternative data