If you attended LegalTech in April, you certainly noticed the great number of vendors offering “predictive coding,” typically as part of a “technology assisted review” process intended to reduce the time and cost required for document review.
What is predictive coding? In simplified terms, predictive coding is a technology-based workflow that allows a computer application to “read” and sort or organize electronic documents. In a typical workflow, human reviewers code a limited number of random documents, and the technology applies those decisions to a much larger document set based upon patterns of relevance. Human reviewers then spot check the coded documents, creating a feedback loop that improves results.
Predictive coding can be extremely useful and powerful in a discovery context by dramatically reducing the size of review sets. But the tools can be used in many other ways, and law departments would be wise to start considering them. It's easy to imagine predictive coding engines being used to organize leases, identify key elements of contracts, monitor communications for regulatory compliance and much more.
In fact, some are starting to use predictive coding technology to analyze legal bills to reduce spend and support better decision making about legal strategy, selection of counsel and alternative fee arrangements. This makes a lot of sense.
In the mid-1990s, the Association of Corporate Counsel, the American Bar Association and PricewaterhouseCoopers created a unified electronic billing standard. The idea was to create a series of standard codes that law firms would apply to their bills in order to facilitate analysis of legal spend by corporate clients. These code sets became known as the Uniform Task-Based Management System (UTBMS).
In the mid-2000s, I co-founded and chaired a committee charged with updating the UTBMS. By then the limitations were obvious. Most law departments viewed it as a good starting point but wanted to track slightly different metrics; this caused them to create “custom codes,” which increased the complexity for law firms. Firms must support different rules for each client and also likely have different perceptions about where certain work fits into the code set. The result is not the standard data patterns that we all hoped for.
This is where predictive coding can come in. Rather than counting on busy lawyers to properly apply codes to their billing entries, if they are descriptive about the work performed, then predictive coding tools can classify it after the fact.
“One obvious benefit to utilizing predictive coding to categorize legal activities is that the cost to get better information can be reduced through an automated, versus manual, categorization process,” explains Jack Walker, principal, Deloitte Transactions and Business Analytics LLP. “Also, the normal flow of work by the attorney is not disrupted, helping to reduce hard or soft dollar costs associated with change management. With better information, of course, comes the ability to make better strategic decisions, as well as insight to make cost-effective adjustments to improve information quality going forward.”
Such a process can provide law departments with better information to set better strategies and make better decisions.
“Law departments are better able to manage a holistic sourcing strategy by determining what tasks and matter types are better performed by internal versus outside resources,” said David Cambria, director of global operations of law, compliance and government affairs at Archer Daniels Midland Company. “They can also use the information to create meaningful alternative fee arrangements and ultimately to driving better outcomes and value to the corporations we work for.”
The bottom line is that the type of consistency delivered by predictive coding tools is imperative for law departments to achieve optimal performance.