This law firm is one of the largest in the U.S. with nearly 1,000 attorneys. It has more than a dozen offices across the country and a wide range of practice areas. The firm's attorneys represent some of the most notable corporations in technology, healthcare, hospitality, insurance and more.
A law firm's corporate client pursued a merger that required government approval and expected two phases of discovery—with the second phase significantly larger than the first. To assist with an accelerated two-month schedule and a massive set of source data, the law firm sought a litigation support partner to assist with targeted collections of more than two terabytes of ESI, scanning, logical unitization, review, redaction, and production. Additionally, a large percentage of the data needed to be redacted for personally identifiable information, which is typically a time-consuming, manual process.
Given the size of the data set and the tight deadline, the law firm sought a partner that could assist with all aspects of the discovery process, provide seasoned project management, enlist expert teams, deliver innovative solutions, and create an integrated approach to the entire discovery process.
The Ricoh team, consisting of a Senior Project Manager, Forensic Lab and Managed Review Services, helped the law firm meet its very tight deadlines. The team dramatically accelerated the collecting, processing, reviewing, redacting and producing of documents for government review.
In the first “quick peek” phase of the review, 200GB of data was reduced to 110k files for review—using term, date, and file type filtering and email threading. Next, the Managed Review Services team reviewed 40k files using Relativity’s advanced Active Learning analytics application to complete the review. Using Blackout’s automated redaction application, Ricoh applied more than 67k auto-redactions within an hour and more than 8k native Excel redactions manually over the course of the short review.
This set the stage for the second, much larger request. Ricoh collected 1.6TB of targeted data and reduced the collection to 1.5 million files using term and date culling and email threading. Ricoh then applied the responsive rankings of the Active Learning project from the first phase to the newly collected data. This resulted in a review set of 372k files ranked non-responsive for a supplemental review in response to a new document request. The big win—attorneys only had to review approximately 40k files in the second phase to complete the responsiveness review.
The two Active Learning reviews reduced the 1.6 million files down to approximately 80k files that attorneys had to review. The attorney teams then performed strategically targeted privileged searches on the 808k responsive files to prepare a Privilege Log, redact for privilege and PII, and produce approximately 500k documents.
The eDiscovery process from collections through production was managed with Ricoh’s Intelligent eDiscovery solutions. For both phases of discovery, the Remlox™ Remote ESI Collection devices enabled the law firm to target and collect the relevant files from workstations and network servers remotely in a user-friendly and forensically defensible way. Collected data was then reduced by culling by date, file type, and search terms before globally deduping and processing. Once the processed data was loaded into Relativity, email threading was performed and multiple search term reports were crafted to identify the most potentially responsive files for review.
For the government's initial "quick peek" phase review, a Relativity Active Learning project was created. It was comprised of only the most inclusive email and electronic file search hits and allowed the review of data to begin immediately. In addition to reviewing for responsiveness, the data was reviewed for privilege and any documents that required redaction of personally identifiable information (PII) were tagged. Responsive privileged families were then batched out for the law firm attorneys to redact and generate a privileged log.
A separate Ricoh Managed Review Services team reviewed non-privileged, responsive documents to redact for PII and provide samples before production. By conducting these three reviews simultaneously and manually reviewing a small percentage using the Active Learning technology, production was able to be rolled out within a few days from the start of the review and completed quickly.
Most documents in both reviews contained PII for at least one individual, if not thousands. A Milyli’s Blackout Application’s auto-redaction feature was used to redact all social security numbers, as well as potential birth dates. This automated process of applying redactions saved hundreds of reviewer hours and greatly reduced the possibility of human oversight. Blackout also saved the team from having to redact Excel file images, allowing reviewers to redact native Excel files in the viewer and perform external redaction on Excel files too large to view in the Relativity’s viewer. Native Excel redaction enabled reviewers to redact entire columns, rows, and worksheets with one click. PII redaction consumed the greatest number of reviewer hours. Therefore, using Blackout's features greatly expedited the process and helped the team identify extremely sensitive PII with far greater certainty.
In the Second Request, data was collected from a much larger number of custodians and an expanded discovery request was received, requiring the review of non-responsive documents from the first phase. The Remlox™ devices were dispatched directly to custodians to collect data from workstations. The devices were also sent to the client firm's IT department to collect from the network server. Then, the data was processed and loaded on a rolling basis as it was received. From there, the data continued to be culled through deduping, email threading and applying search, date, and file type filters. The data was further reduced through a series of text and metadata searches.
Once the new 1.6TB data set was received, processed, and culled, it was loaded into the existing Active Learning project—which immediately ranked all new documents for responsiveness. After calculating recall and precision from samples to determine the rank accuracy, three concurrent reviews were conducted. Law firm attorneys conducted a privilege review of the new documents ranked responsive while the Managed Review Services team redacted responsive documents for PII to expedite rolling productions. A second Active Learning project was also created to load all documents ranked non-responsive to review for the new document request. After reviewing 10% of the data set, the review was concluded and responsive documents from the second project were flowed into the privilege review.
All in all, the complex and sensitive multi-phase eDiscovery process was streamlined and expedited by making the best use of a wide variety of service offerings and tools. Combined with an innovative, customized and dynamic approach, multiple concurrent reviews were able to be conducted amid tight timelines. By integrating conceptual analytics and automated processes with strategic manual review, the law firm saved its attorneys hundreds of hours of document review and redaction—while providing a much higher level of certainty than a full manual review.