Applying analytics to managed review is increasingly familiar territory to RVM’s clients. Sanjay Manocha and Shelley Brown describe the importance of their team’s relationships in demonstrating the value of analytics to clients who are technology novices, and what corporate law departments can do now to prepare for an efficient managed review down the road. Their remarks have been edited for length and style.
MCC: Please give our readers a brief overview of managed document review’s evolution over the past 10 to 15 years.
Brown: If you ask five people in the e-discovery industry what managed review is, you’ll get five different answers. For some, managed document review consists of staffing a review team with contract attorneys, while others define it as a much higher-level service between a client undergoing electronic discovery and a trusted e-discovery advisor. True managed review is the latter. It is the application of proven, defensible and repeatable best practices, including analytics and top-notch attorneys, to the management of document review. At RVM, we call this process structured review.
The evolution of managed review has followed the growth of big data. Where 15 years ago, smaller data volumes allowed for slower, methodical and even painstaking methodologies, like linear review, the sheer volume of data present in discovery today necessitates a more deliberate approach to uncovering the most important data more quickly. Reviewing everything before discovery closes is often untenable; therefore, managed review today, if it’s done well, includes time- and cost-saving strategies and advanced analytics.
Manocha: As Shelley was hitting on, one of the biggest changes is a more methodical approach that incorporates analytics within some of the earlier developed tactics, such as batching based on keyword searches or custodians, or even looking at date ranges. Approaches like email threading have come into play. In the past seven to eight years, we have seen analytics being methodically and more acutely applied.
MCC: From your perspective, how can corporate law departments best prepare
for a managed review, and what should they expect from the team with which they are collaborating?
Brown: The very best way that legal departments can prepare for managed review is by having their information governance house in order. Data that ends up in managed review is often dictated by what has been preserved and how well organized that effort is within the corporation. Therefore, an unmanageable volume of data to review is often a symptom of bad information governance.
Another very important step that corporate legal departments can take is to engage trusted advisors before any review begins and to collaborate with that team throughout the process, from collection through production and beyond. They should collaborate with counsel to make sure that the ultimate data set for review is meticulously culled and curated to eliminate any unnecessary expenditures of money or time while ensuring that all relevant data is reviewed.
MCC: Managed review requires the use of technology and highly trained experts who can design and execute a thorough and efficient review strategy. What are some of the tactics that you and your colleagues at RVM utilize to provide your clients with the best results?
Manocha: Before you drill into specific tactics, it’s useful to think about your overall philosophy for analytics-based e-discovery because of the advances in the technology. Analytics has changed nearly all stages of the e-discovery reference model. However, as Shelley indicated, it hasn’t really come of age yet in the information governance space. The earliest we’re seeing applications of analytics is to inform the scope of collection in early and targeted fact development, placing counsel in a position where they may potentially enter into a strategic disposition of the matter if they find what they’re looking for.
We find a tremendous value early on, to the very left of the e-discovery process, just after information governance and before the broader collection, in a stage that RVM calls Fact Discovery FirstTM. RVM applies analytics early on, before collection and as part of the identification process, and then carries that knowledge through processing and culling. We can apply further refined strategies in the attorney review, relating to tactics that enhance fact development, optimize review efficiencies and enhance the defensibility of the overall process. The overarching concept is that early analytics can set the stage for performance through the rest of the e-discovery process, all the way through to production and even into deposition preparation.
Given that philosophy, one of the tactics we may use is to focus on leveraging analytics in the review process; analytics are very effective at drawing patterns in data. By crystallizing those patterns and analyzing them, counsel can draw inferences from the data that would not otherwise be discernible to a human reviewer. This value-add differentiates RVM tremendously from the average linear review shop.
MCC: What are the most important skills or qualifications needed in the review process: project management, a legal background or a technology background?
Manocha: We look at the team composition for all matters. Some of RVM’s most robust and sophisticated structured review engagements – our technology analytics plus attorney review integration – employ several layers of expertise: typically a consultant, a review manager and an analytics manager. For the most part, all of these people are attorneys with deep experience in litigation and government investigations.
The consultant is a more senior person who demonstrates a strong capacity for exercising good judgment for relationship building, core communication, technical know-how and crisis management. Crisis management is important because the volume and complexity of engagements and the speed at which e-discovery moves often mimic a crisis situation in which we have to act well and clearly under tremendous pressure. The consultant’s role is to be the client’s trusted advisor, help the client define the project strategy and coordinate all the necessary resources within RVM’s domain to achieve the client’s objectives.
The review manager has experience in litigation and government investigations with a focus on all stages of the document review process. Sometimes, he or she will have substantive expertise in the particular subject matter of the litigation or investigation.
The analytics manager, similarly, has had a career path parallel to the review manager’s but with a focus on the application of analytics in the various stages of e-discovery. Both the analytics and the review managers work together under the advisement and guidance of the consultant and the client to ensure that the analytics and the review are properly integrated and that the client is getting the full value of those strategies.
MCC: When dealing with corporate law departments, how would you describe the roles of technology and people? How do you sell analytics to a legal team that is not tech-savvy?
Manocha: The process varies depending on a lot of factors, the biggest one being the volume of the matter. Even the least savvy attorney when it comes to technology and analytics will recognize that if the volume is so compelling, there will be no choice but to use analytics. In those situations, education is significant. We do a lot of panel discussions. We stay involved in the industry, in ensuring that the industry, on a broader level, is aware of how these things can be used.
In situations where the volume of data is not as large, such that analytics is not a necessity, counsel who are not tech-savvy will invariably, unfortunately, choose the more familiar, less risky, linear-based approach. Given the speed at which e-discovery moves, this counsel often believes that it’s too late to become familiar with advanced approaches. This is where a relationship, and trust in that relationship, is important, because the well-traveled path is, quite frankly, analytics at this stage. Does the client trust our consultant on this engagement? Do we have a history with them? Can they place confidence in our personal experiences with the technology, even though they don’t have their own personal experiences with it? That’s a key component.
For a client who doesn’t really have an analytics history with RVM, we allow them to wade into the water. We introduce them to the idea of analytics by working with them to implement low-complexity, high-reward strategies, such as email threading and near-deduplication.
Brown: We’re finding fewer and fewer clients who are unaware of the benefits of analytics. Most have some experience, if not directly, then at least anecdotally. When we leverage analytics to cull a data set in such a way that the client ultimately saves close to $3 million on a review, people talk about it. They talk about it with their colleagues and with others who will one day need similar services. We are finding that our successes are leading us to a place where analytics all but sells itself.
MCC: What specific analytics tools are available, and what is their best use? How do analytics help with document management and review? Are you flipping a coin?
Manocha: Fundamentally, analytics technologies are statistical approaches – statistical data with science-based algorithms that, at their core, work in the same ways. We typically like to describe these technologies along the spectrum.
On the very left side of the spectrum are technologies that are objective in their structure in that they’re not meaning-based technologies. They’re used to organize content based on shared words or letters, the very structure of the data or the record. These are approaches like near duplicates, detection or email threading. Those technologies are good at eliminating redundancy within a data set without regard to what is relevant or not relevant to your matter.
On the opposite side, to the right, you’re talking about more substantive, meaning-based analytics that are designed to separate document populations based on their content but are less dependent on whether they share the same words, as long as the words are conceptually or topically related. This category includes predictive coding, concept clustering, categorization technology and conceptual search. You can get into even more robust meaning-based technologies to investigate story lines within a data set, which is even more knowledge extraction. Those technologies might use natural language processing or graph databases. They are really good at identifying hidden patterns of communication within a data set.
Another really great technology is social network analysis, which, along with clustering, can tell a client very quickly who’s talking to whom about what subject matters.
These tools are available in various software packages. RVM keeps a broad suite under its umbrella because we realize that different analytic tools are best leveraged in combination with one another. Having those at our disposal allows us to tackle almost any problem we’re faced with in the e-discovery context.
MCC: Managed review has gone well beyond emails and contracts. Can you tell our readers about some of the new messaging and communication platforms you are reviewing. And what’s your advice to companies using these alternative platforms?
Brown: In this scenario, the tail is just not going to wag the dog. E-discovery is not going to drive the business choices regarding messaging, collaboration and communication technologies, such as Slack, Yammer, Symphony or even Bloomberg. Our advice is that your trusted advisor understands the nuances of dealing with these types of data across the spectrum, from information governance to preservation to processing, analytics and review.
From the analytics and review perspective, the biggest change is the brevity or terse language characteristics of these messaging platforms. This communication is very different from email, so the same analytics and review approaches are not applicable. As a client, you want to work with a provider who can demonstrate that they have thought about these differences and that they have a plan to deal with them.
MCC: What advice would you give to corporate law departments as they develop and update their information governance policies?
Brown: Discovery has long been the most costly phase of litigation. What ends up in discovery can be directly related to a company’s information governance policies. Therefore, these policies should be created with discovery in mind. They should not be so inclusive as to create an environment where there is a torrent of unnecessary or duplicate data that may ultimately be collected and reviewed.
MCC: What, if any, changes are you seeing as a result of the changes to the federal rules of civil procedure?
Brown: The changes that we’re seeing are more on the information governance side. This affects analytics and review indirectly because, again, what ends up in review is often a symptom of bad information governance. Many clients are struggling to grasp the intricacies of the rules generally but of 37(e) in particular, which in my opinion is the sexiest of the amendments. This amendment deals with the topics of spoliation, sanctions and adverse inferences. As a result of the misinterpretation of Rule 37(e), we are seeing some trends toward overpreserving, which can lead to overcollecting and overreviewing, particularly where analytics is not leveraged or is not leveraged properly.
Published May 3, 2016.