Migrated from eDJGroupInc.com. Author: Greg Buckles. Published: 2014-11-29 19:00:00Format, images and links may no longer function correctly.
Kroll just released conclusions and limited data from a short eDiscovery survey covering trends such as Predictive Coding (PC/TAR) use, Social Media, BYOD, the ‘Internet of Things’ and Security. With over 550 law firm and corporate respondents, Kroll has certainly managed to get a statistically significant sampling, though they were the first to acknowledge that their questions were very basic data points. I just wish that they would publish the raw questions and response data so that readers could form their own interpretations with full confidence. I did find a couple of items interesting in light of my own recent research on analytic adoption. Without the full context/text of the original questions and response options we can only make limited observations. My interviews quickly pointed out problems with my own Analytic Adoption survey questions that I dissected in a blog.
On the PC/TAR trend, I agree with Kroll that the “use of predictive coding in on the rise”. What is fascinating to me is that over 61% of PC/TAR users only used it on 1 or 2 matters in 2014. With most Biglaw firms handling thousands of matters annually and typical U.S. corporations facing hundreds of lawsuits, the overall portion of matters using PC/TAR becomes vanishingly small. I estimated a PC/TAR usage rate of roughly 5-8% of litigation that actually went to production based on my interviews of twenty one peers on the cutting edge of eDiscovery. Kroll asserts “This indicates and increased willingness to rely on the technology, as an attempt to reduce costs or to increase efficiency in document review, or both.” My research confirms that PC/TAR adoption is increasing, but I believe that the whole ‘cost savings’ message is an over simplified red herring. Smart consumers can use a wide variety of aggressive analytics to filter out non-relevant ESI and re-organize the remainder for optimized review without facing tricky protocol negotiations or getting the bench’s approval. Moreover, using the analytics in the ‘processing’ stage avoids overproduction of non-relevant confidential documents and potential exposure of non-relevant training sets.
I am not sure that survey respondent’s or Kroll could differentiate between the use of PC/TAR in pre/post review when answering the “Why use Predictive Coding?” question. The fact that 72.6% of respondents selected “Discovery Productions” does not tell us if they used it to organize the collections but reviewed everything, used it to create exclusion filters prior to review, used it for QA/QC after review or many other non-review scenarios. I pretty much threw out my own results on a similar question once I started to talk to respondents. The good news is the relatively high numbers for Information Governance activities. I found similar rates of response in my detailed breakdown of analytics usage cases, but the numbers did not stand up to scrutiny on follow up interviews. Respondents who indicated analytics usage in defensible deletion projects turned out to use basic inventory tools to profile network shares or repurposing non-relevance categories from prior reviews of selective collections. Only Equivio had two case studies using purpose built PC/TAR workflow to tackle enterprise wide repositories. There are more success stories out there, but the vast majority of respondents either were still pilot testing technologies or were using them in very peripheral roles.
I was pleased to see the rising recognition of social media, BYOD and even the Internet of Things (ESI generated from automated device interaction). Again, the high proportion of respondents indicated that these data sources only come up in a very few cases per year. So calling out “58% of law firms and corporations reported that they experienced ediscovery matters involving data from personal devices.” may be an accurate statistic, but it has the potential to overstate the impact of these new ESI sources. eDJ’s take from active engagements and research is that most litigants still see these data sources as low probability exceptions to their normal discovery lifecycle. The 99% aggregate percentages for the five categories of the final “Next big trend in ediscovery” question makes me wonder if respondents had more options or even a free-form Other option. Broad survey questions on trends need to be open for real input from the market. Is there value in the Kroll survey? Absolutely. Should it be interpreted critically. Yes, just like every perspective piece, including my own.
Greg Buckles can be reached at Greg@eDJGroupInc.com for offline comment, questions or consulting. His active research topics include analytics, mobile device discovery, the discovery impact of the cloud, Microsoft’s 2013 eDiscovery Center and multi-matter discovery. Recent consulting engagements include managing preservation during enterprise migrations, legacy tape eliminations, retention enablement and many more. Blog market perspectives are personal opinions and should not be interpreted as a professional judgment.