The good news is companies are becoming increasingly familiar with predictive coding, but the more unsettling news is they are facing up to $20 million in annual e-discovery expenses. According to the third annual study of streamlining and reducing the cost of e-discovery on inside counsel at mostly Fortune 1000 companies, 81% of respondents are familiar with predictive coding to determine whether a document is appropriate to include in a case, states legal industry analyst Ari Kaplan, who conducted the study in tandem with e-discovery provider FTI Technology. Predictive coding is an algorithm that provides the ability to review documents combining artificial intelligence with a lawyer's input.
"Now there is a movement to get more and more technologies [in-house] to make a determination at least at the most basic level, and 55% of respondents said they would consider using it,'' he says. Kaplan surveyed 31 inside counsels late last year for the study, "Advice from Counsel: An Inside Look at Streamlining E-Discovery Programs."
"This is a real shift in comfort level ... more than half of these people, all of whom are in-house counsel with decision-making capabilities and involved personally in this process, say this has become so sensitive they're willing to use even the newest technologies to see how it works. And as the technologies develop, you'll see adoption of this in some form in terms of how they treat e-discovery."
The survey found that in 2011, 45% of respondents handled more than 500 legal matters annually, up from only 7% in 2009, and that while the annual costs of e-discovery are spiking as high as $20 million, expenditures of between $3 million and $5 million are more common. However, the ability to calculate cost savings has not significantly improved between 2009 and 2011, even though 94% of respondents regarded cost as important, the study found. The most expensive phase of e-discovery is legal review, according to 90% of respondents, and 20% said focusing on review costs is a key area for cost control.
"Review is so expensive, so if you can reduce the total number of records, and then if you can do it faster, you've now come up with a combination to address a problem that everyone's facing, regardless of the size of their organization," says Kaplan. He adds that implementing new solutions like data classification tools, which allow a set of documents related to the same issue to be grouped together, gives an attorney the ability to drill down based on a concept the documents share. This will make the process of finding relevant information more efficient.
"There is a lot of duplicate and redundant and useless data that can ultimately be collected, and sometimes it makes its way into review," says Mike Kinnaman, senior managing director, FTI Technology. "Companies are trying hard to put in retention policies around the data onsite to get rid of data that isn't necessary and to classify data that is useful and better serves the organization. The reality is it's very early for companies in this area."
Kinnaman says there is a "pretty healthy level of skepticism" about using predictive coding and getting to a point where everyone is comfortable with the results. "The point is, the computer doesn't do everything; it's based on humans, and they need to know what they're doing and how it will stand up in court," he says. Kaplan concurs, saying that the technology is now in the experimentation phase and that, initially, it will be for internal purposes only.
"There are examples where the technology is being used, but, like anything, it comes down to the expertise being used with it,'' says Kinnaman. "It's classifying documents based on human input, so there are a lot of tricks of the trade [to make sure it is being used properly]. We think that's a trend that will continue."
Kaplan says another interesting finding was that more legal counsel are interested in sharing their experiences on an anonymous basis for the purposes of benchmarking. More than half of respondents (52%) would consider some cloud product for e-discovery, rather than housing an application behind the corporate firewall, which is up from 47% in 2010. "We're entering a sharing culture,'' he says.
Learn more about Strategy: Stop Illicit Data Dumps by subscribing to Network Computing Pro Reports (free, registration required).
