ACT Litigation Services

Skip to main content

The E-discovery Avalanche

On a typical case today, 90% or more of the discoverable documents are coming from electronic collections. And the volume is growing. In 2006, the Federal Rules of Civil Procedure (FRCP) were amended to address this new reality. The FRCP has even sponsored a new acronym: ESI (Electronically Stored Information).

ACT has conducted benchmarking studies that show approximately 75% to 90% of the total cost of discovery comes from attorneys and other legal professionals reviewing documents for privilege, responsiveness and relevancy. In separate benchmarking studies, ACT has observed that responsive/relevancy rates of reviewed populations typically range from 10% to 40%. This means that the bulk of a client’s discovery investment is going into reviewing non-relevant electronic documents.

The processes used for performing review on large electronic discovery matters have not significantly evolved from the processes used to review the hardcopy discovery matters of the last century. While reviewers nowadays are usually looking at computer screens, they are just using images of the same documents they used to review on paper. The processes governing quality and throughput have not materially changed. As a result, the costs for document review have increased at the same exponential rate as electronic discovery.

From our work with The Sedona Conference, ACT has observed a keen awareness to this problem and a growing openness to non-conventional solutions. Corporations polled by ACT are consistently reporting their primary discovery priority is cost reduction. Corporations are inviting ACT to sponsor process alternatives that are targeted at substantial cost savings.

But is it Defensible?

The number of cases reporting discovery miscues seems to be growing daily. These miscues have resulted in severe sanctions in a number of high profile matters (Zubulake, Morgan Stanley, Philip Morris, etc.). Corporations and their outside counsel are understandably concerned about changes that invite challenges to their discovery practices.

ACT is a collaborative participant in an effort to create a standard method of verifying discovery process quality. Additionally, ACT has developed third party relationships with experts who can testify as to the methods used to generate quality metrics.

ACT strongly advocates the inclusion of quality metrics with any new processes.

Intelligent FilteringSM

Following our successful formula for Intelligent TriageSM, ACT recognizes the first step in cost reduction is to reduce the number of documents reviewed. Even with the employment of processes such as Tier 1 and de-duplication, the remaining volume of documents can number in the millions or tens of millions. ACT recognizes the value of applying filtering techniques to further reduce the review set.

The challenge with filtering is that it’s an imperfect science. While some filters (such as date filters) are usually reliable, keyword filters are notoriously inaccurate. Keyword filters are both over-inclusive and under-inclusive in that they pull-in non-relevant documents and leave behind relevant documents. In a study evaluating keyword search accuracy, the study’s authors show a weighted average for keyword searching accuracy of less than 20%. In more recent ACT internal studies, ACT found keyword filtering yielding accuracy levels as low as 6% to 8%.

Intelligent FilteringSM is ACT’s alternative to keyword filtering. ACT uses advanced search engines to extend the search capability of standard Boolean (keyword) algorithms. Using an iterative process of sampling and feedback, ACT works to raise the performance of the search engines to a level more in line with expectations.

Intelligent Issue CodingSM

Once document populations have been reduced to the extent possible, the remaining discovery populations should be organized and categorized into useful subsets in anticipation of deposition and trial. Both client and other party documents need this organization.

Using techniques similar to those described with Intelligent FilteringSM above, ACT can electronically categorize on issue code document populations. While computer processes cannot as yet match the inductive reasoning abilities of the human mind for identifying important documents, computers can be useful for electronically organizing them.

ACT’s Intelligent Issue CodingSM process creates a “score” for a particular document as to how it relates to a particular issue. Human reviewers can then review the documents focusing on those that are deemed most likely to be on point to the issue.

Summary

Review Acceleration is a series of processes, using advance search technology and statistical sampling and verification, to reduce and optimize the human review of discovery populations. At ACT, we have already used these processes to save our clients’ months of cycle time and millions of dollars and by–prioritizing the human review–and allowing it to focus on documents that are more likely to be responsive and relevant to the matter at hand.