A multinational company is a defendant in a putative class action lawsuit. Outside counsel negotiated the plaintiffs’ document requests as much as possible, but the parties still expect that the defendant will produce millions of documents in discovery. The defendant’s general counsel would like information regarding ways to manage the costs associated with the large volume of data, including considering various data analytical tools to manage the costs and burdens of an extensive electronic discovery efforts.

Increased Acceptance of Technology-Assisted Review

A key goal for companies responding to discovery demands is to identify responsive documents in a cost-effective and defensible manner. Using Technology-Assisted Review (TAR) allows companies to reduce the massive volume of data involved to a more manageable level by ranking, prioritizing and classifying electronic data. Studies have consistently found that TAR processes yield more accurate and cost-effective results than traditional search methods.

The most common form of TAR is predictive coding—a process by which computer software is trained to identify responsive documents for production and then conducts the review process instead of humans. As an initial step, a senior lawyer reviews a small subset of the overall universe of potentially responsive documents to develop a seed set of documents determined to be responsive or non-responsive. This seed set is used to train the e-discovery vendor’s proprietary computer software, which will then rank documents based on their likelihood of responsiveness. The predictive coding process is iterative, and the senior lawyer will review subsequent subsets of the computer-classified documents to verify and improve TAR’s accuracy. When the program demonstrates the necessary level of accuracy after several such iterations, the rest of the documents will be reviewed using the program without human intervention beyond that necessary for quality control processes.

Over the past few years, federal courts have increasingly accepted and endorsed the use of TAR. The Southern District of New York, for instance, held that it is now “black letter law” that a party may use TAR for document review if so desired. Internationally, courts have also begun to endorse the use of TAR. For example, two published decisions by English courts earlier this year granted applications to use predictive coding based on the accuracy, consistency and cost benefits of such technology.

Strategies and Best Practices

TAR is still relatively new and not yet fully understood by the legal community. Counsel should implement best practices to ensure the process is defensible if challenged by the opposing party.

  • Choose a knowledgeable e-discovery partner. When contemplating the use of TAR for addressing large data volumes, hiring the appropriate e-discovery vendor is an important first step. Counsel should consider the nature of the review tools used by the potential e-discovery vendors and their level of experience using those tools in similar situations.
  • Determine whether the case is suited for TAR. Some cases are better suited for the use of predictive coding than others. Factors to consider when determining whether TAR might save time and money in a case include the number and kinds of documents to be reviewed and the number of custodians. TAR is best suited for cases with a significant volume of text-based documents from numerous custodians. TAR is less likely to be appropriate for cases involving a high number of non-text based documents, such as photographs, images or audio files.
  • Decide whether all documents identified by TAR as responsive will be reviewed prior to production. TAR may be implemented to either prioritize documents for human review, while simultaneously removing non-responsive documents from the human review pool, or as a method of determining the production population, subject to appropriate quality control procedures. There are benefits to reviewing all responsive documents prior to production that should not be overlooked before deciding to rely on TAR. However, in appropriate cases, a company may choose to produce all documents that TAR identifies as responsive to save time and money.
  • Document decisions. It is important to carefully document the choices made during the process. TAR is still imperfectly understood by some in the legal community and can appear to be a “black box.” Counsel is encouraged to work closely with their e-discovery provider to create supporting documentation that describes the process and the method for identifying the seed sets. This documentation can be used to replicate the process in future litigations and to explain and defend the process if a challenge occurs.
  • Implement appropriate privilege screens and clawback provisions. The use of TAR without a robust and appropriate privilege screen carries the risk of inadvertently producing privileged documents. Thus, it is vital that counsel develop a comprehensive set of privilege search terms to identify potentially privileged documents within the responsive set identified by TAR. Even if other responsive documents are not reviewed by humans prior to production, documents identified as potentially privileged should be manually reviewed to identify ones to withhold from production as privileged.
  • Implement quality control processes. Using a control set to test an implementation of TAR’s recall rate—which is the percentage of responsive documents identified by the technology—should be considered. The control set is a sample of documents manually coded for responsiveness. By comparing the manual coding determinations to TAR’s responsiveness determination and addressing any issues that are discovered, counsel can increase its confidence that the TAR implementation has been appropriately trained to locate responsive documents.


In the appropriate case, TAR can reduce costs while producing results that are as good as or better than those of a traditional review process. Before starting discovery, companies should consider whether TAR is appropriate for a particular case. When you proceed with TAR, implement best practices to support this decision if it’s challenged by opposing counsel.