Comment: Is predictive coding becoming a game changer for the legal industry?

With the extension of costs budgeting to all multi-track claims with a value of less than £10 million (see Practice Direction 3E of the English Civil Procedure Rules 1998), and the May 2016 decision of the English High Court in Brown v BCA Trading Limited [2016] EWHC 1464 (Ch), in which it permitted a party to use predictive coding technology in a contested application, it goes without saying that knowing how and when to use predictive coding will provide both lawyers and their clients with a clear tactical advantage in the form of effective case management, as well as time and cost efficiencies. 

Although predictive coding technology has been around for some time, English lawyers, both in private practice and in-house, continue to be unsure as to how it actually works in practice and therefore prefer to stick to more traditional linear methods of review.  However, as the decisions in Pyrrho Investments Ltd and another v MWB Property Ltd and others [2016] EWHC 256 (Ch), in which the English High Court delivered its first reported decision expressly approving the use of predictive coding technology, and the case cited above demonstrate, predictive coding is here to stay.  It is therefore vital that lawyers move with the times, otherwise they risk managing cases inefficiently and being caught off guard when they are confronted by a more tech-savvy opponent.

What is predictive coding?

Predictive coding involves training the relevant predictive coding software or technology to categorise each document in the whole document set as either relevant or not.  This machine learning is achieved by a process in which a senior lawyer (who has mastered the issues in the case) reviews a sample set of documents for relevance.  This sample set is then used to train the software, and through a process of iterative refinement (using further sample sets), an appropriate confidence level and margin of error is reached. Once the technology has been trained in this way, it is applied across the entire document set to identify relevant documents.

What are the benefits of using predictive coding?

Used appropriately, predictive coding technology can result in a number of positive outcomes.  It can save a party significant time and cost when compared to other methods of document review.  For instance, in the BCA Trading case cited above, the judge noted that “the costs envisaged for predictive coding are in the region of £132,000, as opposed to costs for a key word search approach of at least £250,000 (and indeed I believe that figure may even reach £338,000 on a worst case scenario)”.  The issue of costs will be particularly relevant in cases where parties are required by the Civil Procedure Rules 1998 to prepare costs budgets.  As predictive coding technology may provide more certainty once the software has been trained, this in turn should enable a party to make more accurate assessments of its likely costs, thereby allowing for higher cost recoveries in the long-term.

Predictive coding can also be used as a form of early case assessment tool, thereby enabling a party to be in a position to assess the merits of its case at a very early stage, be it litigation, an investigation or any other matter which requires the review of a large volume of documents.  Furthermore, being in possession of the facts at an early stage is likely to provide a party with a clear tactical advantage as it should be able to plan ahead and anticipate problems before they arise.  It should also enable it to consider settlement opportunities in a more effective way.

Knowledge of how predictive coding works will also allow a party to challenge the other party’s proposed methodology, which is what happened in the BCA Trading case.  In this way, a party has an opportunity to control the costs of the litigation and to tactically put pressure on the other side to do something different to what they had originally intended.

What are the risks of ignoring predictive coding technology?

As noted above, predictive coding is not going to go away.  If anything, it is likely to be used more and more frequently as parties realise, particularly in the light of the recent English High Court decisions, that they must take it into account when considering which technology is most appropriate in any given case.  By not applying it in the right circumstances, a party will potentially miss out on all of the potential benefits that it may bring, including those outlined above.  Furthermore, by failing to become sufficiently knowledgeable as to how this technology works, a party runs the risk of being caught out when they are confronted by a scenario, as in the BCA Trading case, in which the other party is advocating its use.  It is therefore important that a party is familiar with the use of predictive coding so that it is in a position to challenge (or indeed concede to, if necessary) the other side’s arguments for its use.  Also, a failure to use early case assessment tools, such as predictive coding, when there is always a possibility that the other side may do so, may mean that the other party has a tactical advantage in that it may know more about the evidence and therefore be in a better position to run the case that is advantageous to its client.

The landscape is changing and parties can no longer afford to ignore new technologies such as predictive coding.

By Tess Blair, partner, and Afzalah Sarwar, associate, Morgan Lewis. Tess is the founder and leader of Morgan Lewis’s eData practice.