e-Discovery software solutions: Lessons learned
This post was written by Joe Raczynski, technology manager at Thomson Reuters
e-Discovery software solutions are invaluable in the litigation preparation process. However, choosing a tool and implementing it poses challenges to law firms. In the last session of ILTA 2013 in Las Vegas last week, a panel with William Kellermann of Wilson Sonsini, Chad Ergun of Gibson Dunn, and David Hasman of Bricker & Eckler discussed the implementation of their discovery solutions including the models and workflows.
Software selection process
Wilson Sonsini chose iPro e-Discovery suite based on some fascinating independent analysis. They used a white paper written at George Mason University’s engineering program. The premise is based on SysML. The paper states, “SysML is a general-purpose graphical modeling language for specifying, analyzing, designing, and verifying complex systems that may include hardware, software, information, personnel, procedures, and facilities.” They applied the model to the software decision making process. The selection process allowed for seamless workflow by establishing waterfalls for decisions and made the application defensible.
The other two firms, Gibson Dunn and Bricker Eckler, both chose Relativity. They based their decisions on the high level of support that Relativity offers; predictive coding; and that analytics were ready out of box. Lastly, both firms felt that the application could handle a range of document sizes from small to large.
Contract negotiations
The Gibson Dunn negotiations lasted two months. While every member of the panel eschewed the subscription model currently en vogue, they accepted it as the current state of the e-Discovery business. In Gibson’s case, it was cumbersome figuring out how many seats of Relativity they needed. Their previous system, Concordance, did not track users. Ultimately, they started with 100-150 users to test acceptance and now they are at 600 users.
Bricker was deeply concerned about the subscription model as well and therefore limited usage for fear of breaching their number of user’s allotment. If they did surpass their quota this small firm would incur a $50,000 charge.
To underscore the general distain for the subscription model, Wilson Sonsini weighed in as well. They are troubled that subscriptions will become a consumption model. The flaw in the consumption model is the inability to pass along the per gigabit pricing to their client.
Installations
Gibson found the hardware necessary for Relativity difficult to comprehend as the numbers of variables, people, and usage can fluctuate. Their answer was to virtualize their entire architecture. They also used a SAN for storage creating elasticity with memory and CPU for small and large cases.
Wilson Sonsini had a unique perspective. For operational use they established a local server which was faster. Staging data for them works in the cloud while the heavy processing of data is on an in-house solid state drive for speed. Ultimately, Wilson Sonsini cautioned about how long it can take to set up a server, suggesting that this can take weeks to build.
Actual software install
All three firms echoed the same advice for the actual software installation:
-Make sure you plan it out
-Have a dedicated project manager at the firm and with the vendor
-Get access to a knowledge base for the product you are installing and create one for yourself inside the firm
-Confer with your peers on their experiences for installation
-Test, test, test.
Lessons learned
The main theme during this conversation of orchestrating a software installation dealt with planning. Map everything and create templates on how users will interact with the application, their data, and the overall workflow. They also reiterated that doing as much pretesting as possible, including load testing. i.e. 200 people hitting the service at the same time, is paramount. Lastly, using tools like a project management Gantt Chart will help determine dependencies and produce sequential alignment.
Some clarifications – SySML was used by George Mason graduate students to assist an international electronic discovery vendor in an analytics software selection process. The white-paper established a road map I used to architect our internal system and aid evaluation. As for storage, we do not use the cloud for primary storage. We can write aged staging data to secure cloud storage after it is initially stored on an appliance comprising several solid state drives.