Defensibly Reducing Review Volume by 75% using Statistical Search Term Optimization
Frequent, high-profile litigation and regulatory matters involving large volumes of extremely sensitive data that must be reviewed against tight deadlines and high expectations of quality.
Fortune 20 global financial services company was seeking to reduce costs by defensibly minimizing the volume of documents for review, while ensuring a high level of responsiveness and security. Specific challenges included:
- A large and unnecessary volume of irrelevant documents being reviewed due to inefficient culling, driving high review costs.
- Massive amounts of data were collected with low richness/prevalence of likely relevant documents, exceeding 5 million documents.
- Sensitive documents were at risk of inadvertent disclosure and security failures.
Our Technology and Consulting Services teams leveraged analytics services to enable: • Statistical analysis of keyword search terms, encompassing sampling, assessment, and optimization of terms.
- A standardized, defensible process to create statistically defensible samples and infer characteristics of a large data set, streamlining and reducing manual review of 5 million documents.
- A secure review environment, including biometric access control and other state-of-the-art physical and virtual security measures.
- Returned $660,000 in savings from an $18,000 investment in optimization and validation.
- Reduced volume of documents requiring human review by 75%.
- Improved precision by more than 50% and maintained a recall rate of ~93%.
- Eliminated risks and potential exposure via physical and logical security of proprietary and sensitive data.