Best Managing Practices for a TAR Review

Author

Co-written by Robert Schechinger and Traci Gray.

What is TAR?

Technology Assisted Review (TAR) essentially predicts the relevancy of a given document based upon coding decisions and preferences made by reviewers on documents within the same document set. As more and more documents are completed, the TAR software better understands the patterns, features, and characteristics of the review to predict which of the unreviewed documents are most likely to be relevant to the review and which are not. When a project includes hundreds of thousands, or millions, of documents, the predictions that are able to be generated by this software can allow a review team to better focus their energies, provide a more comprehensive final product, and save money.

The software is not omniscient, however. It is only capable of working with the information it is given. This means that the software needs to know what is important to your search (i.e., it needs to be trained). By utilizing relevant keywords, identifying and coding the most helpful documents, and constantly evaluating the decisions made by the document reviewers and the software, the TAR software will eventually be trained to identify and distinguish relevant and non-relevant documents with a high degree of success.

Best Practices: Defensibility

In developing the best practices for a TAR review, the managing attorneys should ensure that the working practices are defensible. Defensibility is a product of the mindset that is adopted when developing a process. This mindset is built on the use of innovation and best practices. Most managing attorneys are not software experts, so it is important that the communication with the hosting vendor is clear and explicit. The vendor can explain how the software analysis will affect the data results, and counsel should be clear on the data that it is expected to produce. 

The TAR Process

In the beginning, the review team will code the first few sets documents (“the seed set”) for relevancy, allow the analytics to score those documents for further review, and then compare the highest scored documents with the lowest scored documents. Note, however, that TAR software works on the “garbage in, garbage out” principle–if you are not providing the software with high-quality information, you will not receive high-quality results. The managing attorney needs to ensure that the review team coding the documents fully understands the protocol they are to follow when making coding decisions. Decision logs providing details on the protocol are updated routinely, and quality control (QC) needs to be a priority from day one.

The document results should then be shared with counsel on a regular basis so that the firm has a detailed view of the kinds of documents which fall into each category. Keeping abreast of the statistics is important, so that adjustments can be made in real time. Managing attorneys must analyze and measure the results to make sure nothing falls through the cracks. Counsel must then provide feedback to the review team on whether the results received are in line with their expectations. The most important feature in a TAR review, as with any review, is to be transparent in the methodology and create an audit trail. The entire team needs to be able to replicate the processes and steps used to achieve the results so as to ensure a successful review.

To learn more about TAR reviews and the services we offer, please contact your Special Counsel sales representative.

Attract & Retain Top Talent

With a rapidly changing industry, it's vital to offer the right compensation and set the right expectation. With our Salary Guide, get detailed job descriptions, industry insights and local salary data to equip your managers with hiring confidence and expertise.

Get your copy »

Get email updates about more content like this.

Comments

| Next articles in The Column blog |

Get the foundation you need to hire the best legal talent.

Request your copy of our 2019 Salary Guide »
GO NOW
LOAD MORE
LOAD MORE
LOAD MORE
LOAD MORE
LOAD MORE
LOAD MORE