Analytics &
Managed Review

CHALLENGE

Our client was responding to a broad DOJ Civil Investigation Demand (CID) request with drastically overly inclusive search terms, resulting in a potential review universe of 3.2 million documents with a very low observed richness. For example, the client company name was part of 150+ of the nearly 40,000 search terms.

SOLUTION

We implemented a multi-tiered process to identify the most categorically nonrelevant material possible while simultaneously reducing the need for counsel’s involvement. This process started with automatically flagging false hits on search terms and measuring the accuracy of the remaining hits for further negotiations. We then flagged common non-relevant categories (such as newsletters, blast emails, personal emails, embedded items, recurring calendar invites, etc.) for potential suppression.

EXECUTION

We first identified false hits on the client name in signature blocks, physical addresses, websites, email addresses, and external sender warning banners. We then identified redundant search terms among those proposed and created a custom workflow to eliminate them. After addressing these items, the document volume dropped by about 600,000 documents or 19%. Next, we provided reports to counsel reflecting typical non-relevant document categories remaining in the data set, including newsletters, automated notifications, and embedded items. Pulling these reduced the universe by another roughly 650,000 documents or 25%. As the final step before starting the review, we ran an email thread analysis and reduced this volume to inclusive emails only, dropping another 630,000 documents or 32%.

RESULTS
In less than a day of work, the ID team and counsel shrunk the review universe from 3.2 million to 1.3 million – a reduction of about 60% or 1.9 million documents before executing any review or predictive coding, significantly reducing the burden of responding to the CID.

CHALLENGE

The client, a multi-million dollar investment firm, needed help preparing their response to a third-party subpoena for documents relevant to an ongoing lawsuit. They wanted to be confident that their response would not include unrelated sensitive information. The entire review universe would need to be reviewed to ensure all documents were appropriately withheld or redacted.

SOLUTION

As the review universe for this matter was relatively small (under 4,000 documents) and each document required review, ID deployed specific analytic tools to organize the data.

Email threading allowed for a chronological review of an entire email conversation at once, allowing a reviewer to identify relevant and non-relevant information throughout a thread, ensuring privileged or confidential information is not included in the client’s final response.

Clustering increased the consistency during review by conceptually grouping similar documents so they are reviewed by a single reviewer rather than piece-meal by multiple reviewers.

RESULTS

By utilizing email threading and clustering as organizational methods in this review universe, ID was able to complete this review in a shorter amount of time without sacrificing the quality of the results,

ID’s application of analytics and use of legal reviewers saved the client over $17,000 when compared to a linear review performed by outside counsel.

CHALLENGE

The client received an opposing party production consisting of approximately 100,000 documents that needed to be reviewed in anticipation of upcoming depositions and with a 14-day deadline. We consulted with the client to create a workflow integrating analytic tools to increase the efficiency of the limited resources available – specifically, the subject matter experts (SMEs) handling the litigation.

SOLUTION

We organized utilizing the analytic clustering tool. This organizational method allowed the SMEs to quickly identify groups of documents that required further review and groups that were of less importance and prioritize their review accordingly. As the SMEs reviewed and made decisions based on the relevance and content of those documents, we utilized those decisions to categorize the remaining unreviewed documents. This method allowed us to identify additional relevant documents for review that may not have been caught by applying search terms.

RESULTS

By applying clustering and categorization to this review universe, we were able to significantly decrease the number of documents requiring review, and as a result, the total amount of time needed for review was significantly decreased. Our integration of analytics into the workflow allowed the SMEs to prepare for upcoming depositions more thoroughly and prevented unanticipated documents from being brought forth – and ultimately saved the client over $85,000.