Reduce costs and increase efficiency
Leveraging predictive coding to improve traditional document review
Through collaborative consultation between counsel and Deloitte, the client was able to achieve cost savings and improve process efficiency.
A global sports and entertainment company needed to comply with a Foreign Corrupt Practices Act investigation. Using initial search terms that were negotiated between counsel and the Department of Justice resulted in more than 235,000 documents for review. In addition, linear review on the search term hits initially achieved a responsiveness rate of just three percent. Faced with the prospect of spending significant client resources to review material mostly unresponsive to the needs of the matter, counsel needed a solution to achieve better results and improve efficiency.
The client was faced with an ongoing linear review, but became convinced that technology could help improve the review process. Upon deciding to use technology, the client sought an analytics process that would effectuate review efficiencies and achieve linear review cost savings, all without stopping or interrupting the ongoing linear review.
How we helped
Deloitte’s discovery team leveraged previously-reviewed documents to create a training set for its Dynamic Review process. From this training set, our proprietary software created a model for the remainder of the document corpus. To measure the efficacy of this model, we compared counsel’s coding of a sample set of documents, known as the “validation set,” to the Dynamic Review model, and identified the documents where the model disagreed with counsel’s coding. After ongoing collaboration with counsel through a series of iterations, agreement was reached on the ideal model. The refined model was then used to score the entire remaining review population. The Deloitte team used the latest scores to target the documents being reviewed, prioritizing the documents that the model indicated were most likely to be responsive.
At the height of its modeling effectiveness, Dynamic Review achieved a 52 percent responsiveness rate among the materials indicated as most likely to be responsive. This enabled counsel to focus on the responsive materials first, relegating an ever-growing group of documents to a batch that did not merit review. Upon exhausting the documents with a reasonable likelihood of being responsive, counsel and Deloitte collaborated to vet the remaining group, statistically validating a responsiveness rate of less than one tenth of one percent (0.1 percent). With a predicted linear review cost of over $1,000 per responsive document, counsel was armed to discuss setting this group aside with the regulatory authorities.
By not reviewing this material, which constituted 38 percent of the documents responsive to search terms, counsel saved its client approximately $100,000 of linear review cost.
Through collaborative consultation between counsel and Deloitte, the client was able to achieve approximately $100,000 in cost savings. In addition, there was no interruption of the existing linear review, but instead the incorporation of Dynamic Review was able to bolster the review efforts and improve the efficiency of the overall process.
For more information about how Deloitte used Dynamic Review, contact: Steve Shebest, director, Deloitte Risk and Financial Advisory Deloitte Transactions and Business Analytics LLP
Federal CFO Insights
Multistate tax alert | July 1, 2015