A global construction and engineering company faced many uncertainties pertaining to cost concerns and scope of effort in their response to a thirdparty subpoena for documents. With the potential for expanding data volumes (70+ custodians) and mounting production requests, the cost and timeline of manual review--even with an in-house review team – presented a challenge to contain costs and meet deadlines due to high volume. The matter was a prime candidate for the application of technology-assisted review (TAR) – however the client needed assurance that TAR would work with their documents.
The solution would need the ability to scale with expanded data volumes in the event that the subpoena identified additional custodians. Due to the scale of the case and the range of services needed, the client required an eDiscovery partner that could accommodate processing, hosting, TAR, and analytics consulting, and potentially managed review. The client also needed a consultative partner. Because the client anticipated the potential for additional incoming data, the solution had to accommodate rolling data loads.
The Epiq TAR consultant team engaged the client from the outset of the case. From citing case law supporting predictive coding, to providing demos of the tools, Epiq outlined the workflow and illustrated the nuances of best practice. Addressing the concerns of scale and TAR compatibility with the client’s data, Epiq executed a client-driven proof-of-concept exercise to train the predictive model on a small sample set of client documents. An in-house subject matter expert successfully trained the predictive coding tool and it was agreed upon to go ahead with the larger data set, further reducing review time and costs.
After a modest cumulative training effort using 3,500 documents, the predictive coding tool scored the population for presumptive review, incorporating both email threading and privilege search terms to reduce the review set further. Ultimately, Epiq helped the client reduce the eyes-on review population from 2.1 million documents to 152,000 documents. Resulting benefits included massive reduction in eyes-on review required and acquiring substantially more data after initial setup with little additional effort.