eDiscovery Search: Focus on Negotiating Search Process Not Keywords

While I hate to show my age, this blog topic reminds me of a statement I heard at one of the first e-discovery seminars I attended in San Francisco in the 1990s. In those early days, someone from the audience commented that e-discovery will require the parties to cooperate in litigation. A panelist who was then General Counsel of one of the largest tech companies in the world responded with tongue in cheek: “Yes and the litigants will all join arms, sit around a campfire and sing Kumbaya.”

I think it is safe to report that counsel in litigation are still not joining arms with their opponents and singing Kumbaya together. Many lawyers still remind us that e-discovery is part of an adversarial process and clients pay their lawyers to win, not to make friends with the other side.

However, the ever-increasing cost of e-discovery and the growing volume of data that lawyers are facing is resulting in some movement towards a more cooperative process. There is no doubt that we are seeing more reported cases where judges are requiring parties to cooperate on search protocols. And, many courts are referring specifically to the Sedona Conference Cooperation Proclamation in their opinions.

There are a number of practical reasons why counsel should seek input from their opponents in the search culling phase of e-discovery. Providing some transparency to the search process and engaging the opponent in negotiation will head off many of today’s e-discovery disputes. A party who has agreed to, and been involved in, a search process is less likely to complain about the scope of the document production. Even if the opponent refuses to participate in the search phase (which still is common), the fact that a party attempted to engage the opponent in the process should have some influence in the event of an e-discovery dispute.

Engaging the opponent, however, can increase problems if counsel focus solely on an agreement regarding keywords. The requesting party is primarily interested in finding the relevant documents and has less of a concern with the burden of review imposed on the responding party. This often results in the requesting party proposing very general search terms which will further drive down the responsive rate and substantially increase the cost of review.

It is still common today for us to receive a list of search terms with the collected data and a request that all files returned by the search be loaded into the review system. It is not uncommon for us to also hear that the parties have agreed to the keywords so “just run them.”

Developing search terms without access to search the data and evaluate the effectiveness of the search is probably the most costly decision counsel can make in the e-discovery process. And, entering into an agreement with the opponent to include more general terms without testing the search is likely to make matters even worse.

The reality is that the keyword agreement is often not in the interest of either party. The studies (TREC Legal etc.) have consistently found that a list of keywords produce a low recall rate meaning that the search is not finding a large percentage of the responsive documents that the requesting party is is trying to discover. At the same time, the list of keywords often produces a very low precision rate which means that the producing party is burdened with the cost of reviewing a large percentage of non-responsive documents.

Instead of agreeing on a single list of keywords, counsel should look to obtain agreement on the process; after all search is a learning process not an event. The touchstone should be transparency of the process.

Counsel should not expect that they will obtain agreement to implement “black-box” technology. Advanced technology to supplement (not completely replace) keyword searching is available. Active learning technology can identify more relevant documents than simple keywords and can also slash the amount of irrelevant documents that must be reviewed. Technology, such as Servient’s Predictive Review, can be combined with input from the opponent to craft an iterative search and analysis process that is the interest of both parties. Add statistical validation to the mix and you truly have an improved process that will change the economics of e-discovery.

The first step, however, is for counsel to begin to negotiate the parameters of the process. No amount of rule changing will move us beyond the current issues we face in e-discovery; the answer lies in the adoption of reasonable technology solutions.