E-Discovery Tools: Computer Assisted Review in Action

Humans working in tandem with technology to discover key issues in litigation

Posted by Rebecca Shwayri in on April 11th, 2013

With recently issued court decisions endorsing or even ordering the use of predictive coding technology in litigation cases, the use of computer assisted review has become the new rage in e-discovery.

Often, computer assisted review is mistakenly believed to mean that e-discovery in a case can be put on auto-pilot. The reality is that the technology cannot simply discover the relevant electronic evidence without human input. Furthermore, human input is needed to ensure quality control and proper culling of electronic evidence. With the advent of technology assisted review, there is a greater need for educated e-discovery professionals who can determine whether the technology should be used in the first place.

This article will discuss: (1) how to determine whether computer assisted review is appropriate for a case; (2) how to apply analytics to a case involving electronically stored information (ESI); and (3) when to apply predictive coding.

1. Determining Whether Computer Assisted Review is Appropriate

Computer assisted review is not appropriate for all cases. To determine whether it is appropriate, it is important to analyze the amount of ESI at issue. This requires talking to the key players involved the litigation and learning where their ESI is stored. ESI could be stored on an employee’s desktop computer at work, personal laptop, PDA, voice mail, e-mail servers, and backup tapes, among other places. Understanding the volume of ESI at issue also involves having discussions with the company’s Information Technology team. These discussions may include analysis of:

  • the company’s IT infrastructure
  • the company’s document retention policy
  • how long the company keeps certain types of data on active servers
  • the company’s backup and legacy media systems.

Once you have an idea of the volume of data at issue, a decision whether to use computer assisted review can be made. If a case involves a small amount of data—such as a few gigabytes—computer assisted review may not be appropriate. Computer assisted review can be beneficial for cases more than 5 GB. Generally, the larger the data set, the higher the likelihood that the case will benefit from computer assisted review.

In addition to examining the volume of ESI at issue, it is important to analyze the type of case and your familiarity with the issues in the case. Certain types of cases like fraud disputes lend themselves to computer assisted review. After all, it can be difficult to conduct a keyword search for fraud. The more complex and nuanced the issues involved, the greater the likelihood that computer assisted review can make a difference in the litigation.

2. Applying Analytics to the ESI

FREE Investigation Report Template

Prepare thorough, consistent investigation reports with our free report template.

Download Template
After the ESI has been gathered and it has been determined that the case would benefit from computer assisted review, analytics, which provide a broad-based overview of what the case is about, can be applied to the ESI to gain an understanding of the issues in the case. Analytics usually involve:

  • surveying the data
  • analyzing how the key players discuss the issues
  • analyzing who communicates with the key players
  • determining if there are any temporal gaps in the data set.

3. Applying Predictive Coding to the ESI

After analytics have been applied to the case, you should have an overview of the issues in the litigation, the key players, and how the key players discuss the salient issues in the case. Analytics may have allowed you to make intelligent, educated decisions regarding what ESI is relevant to the litigation. At this point, you may be in a position to reduce the volume of ESI in the case. On the other hand, the application of analytics may have lead to the discovery of temporal gaps. The volume of relevant ESI may have increased.

Once you have an overview of the issues in the case, the types of ESI involved in the litigation, and the overall volume of ESI, it is important to analyze whether predictive coding should be applied to the case. Predictive coding can be helpful in cases involving 10 GB or more of ESI. Predictive coding will allow a subject matter expert to code and categorize documents based on relevance. The technology will learn from the expert and will apply math and statistics to determine the relevancy of documents in the rest of the data set. Predictive coding is different from the analytics applied at the beginning of the case. While analytics will give an attorney an overview of the issues in the litigation and will provide for an exploratory analysis of the data, predictive coding actually involves categorizing the ESI and provides an opportunity for the software to code the ESI in an expedient manner.

Technology is Your Friend

The keys to utilizing computer assisted review are to understand the volume of ESI at issue, apply analytics to get an overview of the ESI, and apply predictive coding to categorize the ESI where the volume dictates that it would be useful. Utilizing computer assisted review is not the case of technology replacing humans. Rather, educated professionals will work in tandem with the technology to discover the key issues to the litigation.

Rebecca Shwayri
Rebecca Shwayri


Rebecca Shwayri is a business litigator and information technology lawyer.

Want to conduct better investigations?

Sign up for i-Sight’s newsletter and get new articles, templates, CE eligible webinars and more delivered to your inbox every week.