It's been less than a year since the first rulings were issued validating the use of predictive technologies in e-discovery. While review remains the predominant focal point in the predictive discussion, those who know the technology are well aware that it can be used for far more extensive purposes.
During Exterro's recent webcast, Practical Predictive Intelligence for Proactive E-Discovery, moderator Barry Murphy, co-founder and principal analyst for eDJ Group, led a “moving beyond the hype" discussion focused around the real-world use cases for predictive technologies. The panel comprised Bennett Borden, a partner at Williams Mullen, Linda Luperchio, director of information lifecycle governance for The Hanover Insurance Group and Scott Giordano, corporate technology counsel for Exterro.
Surveys Show Continued Market Maturity but Plenty of Room for Growth
The panel began the webcast by analyzing a number of eDJ Group polls. The survey data suggests that predictive technologies are gaining wider acceptance. According to Murphy, the issues that are still being raised about predictive technologies are far from “showstoppers." They are surmountable concerns that highlight fairly widespread fear of the unknown in the legal industry. The same survey data also reveals that most people still aren't recognizing the tremendous value predictive technologies have outside the review bubble. Murphy observed that while we tend to think of the e-discovery workflow in the context of the EDRM, from data identification on the left to review and production on the right, the application of predictive technologies has progressed in the opposite right-to-left fashion. Organizations are now just beginning to recognize the tremendous cost savings that can be achieved by moving the technologies to earlier discovery phases and into the realm of information governance for things like automatic document classification and defensible deletion.
User Comfort Levels with Technology Still a Barrier
Though fears regarding defensibility and transparency remain, the panel suggested that user comfort levels may be the most significant barrier remaining to universal predictive adoption. Luperchio, who manages e-discovery activities at The Hanover Insurance Group, described an evolving attitude towards predictive technologies at her company, where members of the legal department are increasingly looking towards predictive methods to replace more antiquated approaches. “Our people come to me now and say 'here's the problem, what do you suggest?' Now that I'm more comfortable with the technology, I can say have you considered using predictive for this or considered that we may be able to get the data this way. So I think it's a maturity of the users, as well as the maturity of the courts and the environment," she said.
Moving Predictive Technologies Upstream
Document review accounts for the largest share of e-discovery expenses, and it's easy to see why predictive technologies have been applied to this stage first. However, Borden made the point that when predictive technologies are applied upstream, huge categories or “buckets" of documents can be eliminated right away, making the review process much more manageable and cost effective. “For companies that face lots of litigation or regulatory investigations, they're starting to understand that they're doing most of this over and over and over again, and they're paying for it over and over and over again…The more mature companies are investing in the upstream side and that has a multiplicative effect in savings on the downstream side," Borden said.
Borden also pointed out that while you frequently see the review process challenged by an opposing party or judge, there is almost no legal way to challenge how an organization classifies and manages its electronically stored information (ESI) or what sort of defensible deletion strategies are employed. In that sense, moving predictive technologies upstream allows organizations to consistently leverage predictive capabilities without having to deal with any potential legal quarrels that might surround its use in the review stage.
Predictive Technologies in the Real World
Both Borden and Luperchio shared examples of how their organizations are currently applying predictive technologies in proactive ways. Luperchio described working on a case involving more than 5 terabytes (5,000 gigabytes) of potentially relevant ESI. Under a very tight deadline, the legal team didn't have time to do a large scale collection and assess the documents manually. Instead, they used an in-place early case assessment tool to immediately cull more than 90% of the documents in a matter of minutes, before training a predictive model to search the remaining set for relevance. A discovery project that would have taken weeks to complete using traditional methods, Luperchio and her team were able to produce a set of relevant documents in just a handful of days.
Borden described a different sort of use case that is nevertheless a powerful example of the versatility of predictive technologies. His firm has worked with clients that frequently deal with similar regulatory investigations. One client, in particular, was dealing with repeated harassment violations from the Department of Labor. The client wanted a way to proactively leverage predictive technologies to quickly identify when and where problems were occurring as well as prevent investigations from happening in the first place. Through predictive technologies, Borden and his team were able to develop a list of keywords and a seed set of problematic documents gathered from prior investigations. The company ran a weekly search of its entire email corpus against that seed set and quickly detected developing issues around harassment. As a result of proactively identifying potential problems and dealing with them immediately, the company saw a 70% reduction in harassment claims in just one year.
To hear more examples of how organizations are leveraging predictive technologies, watch Exterro's on-demand webcast Practical Predictive Intelligence for Proactive E-Discovery.