Predictive analytics for lawyers: a five step process

Developing a successful strategy for predictive analytics lies in knowing a balance between big data and people

01 Know your strategy

Creating and implementing any successful strategy demands a good grasp of various scenarios and a probable universe of outcomes. Predictive analytics (PA) informs users about what is likely to happen, using algorithms and machine-learning to interpret data to give the fullest picture of a situation and predict logical outcomes. This certainly makes it a useful tool for lawyers, though more useful in some areas than others. In litigation, for instance, PA is a powerful tool in case law research and ediscovery. Intraspexion uses deep-learning and PA to predict and prevent potential litigation based on their patented software. An early-warning system, it analyses company emails to identify risk factors so, for example, it can flag emails containing text patterns indicating sexual harassment or discrimination.

DISCO is another provider using predictive analytics to support ediscovery, with more than 200 AmLaw firms currently utilising their tool to automate tasks and conduct large-volume document review with the help of artificial intelligence (AI) software. PA can help provide robustness in a legal strategy. It can help lawyers assess the merits of a client’s case and provide analysis for offering sound legal advice. However, even with PA, elements of potential bias remain, and a good strategy needs to excise such bias to avoid incompetence and error being built into the strategy. There is a danger in overstating big data’s renowned objectivity due to these concerns, but also because human beings do not behave optimally. The data we choose or exclude, and how we interpret the data, influences outcomes.

02 Know your data

Data-mining in law firms requires interrogating big data sets: docket data, legislation, case law, client contracts, property titles, to name a few. Bad data risks decisions and policy being based on simple correlations and averages thrown up by the data, hence lawyers need to interpret the meaning of such correlations and  see underlying causal relationships in context. Facts, and thus data, may be stubborn things, but they can still remain elusive and undefined. Predictive analytics can discover patterns, abnormalities and correlations in large amounts of data to help build a case, generate a legal strategy, understand opposing counsel strategies, assess suspects and predict case outcomes. Machine-learning tools can analyse data, while regression analytics can estimate impacts in relationships between variables, establishing causal relationships in different aspects of a case.

Brainspace uses machine-learning and intuitive semantic technology to automate workflows and tackle discovery documents, and is integrated with other ediscovery platforms and management software. Docket Alarm has long helped lawyers by replacing manual checks for updates on a case’s docket. Their PA tool linked to a database of court filings enables users to track and analyse full court records, large volumes of cases and provide judicial profiles. Time-series modelling can collate data points in chronological intervals, enabling lawyers to predict future cycles or patterns. Graph analytics, or network analysis, aids identification of criminal patterns, fraud detection and other such research. In document handling, NoSQL search technology can help find the relevant document in a timely fashion.

03 Know your people

Predictive analytics can help law firms decide optimal composition of teams and ensure all needs are covered by relevant expertise, including deciding on what outside counsel, consulting, strategic partnership or individuals best fits the client’s needs. Using PA in the hiring process helps match the right candidates to the firm as well as selecting individuals both for ad hoc projects and long-term relationships. The emergence of gig lawyers will expedite new ways of building teams. Who can best represent a client is as important as detecting trends and highlighting patterns in the data. Unique and proprietary data can be leveraged, such as case notes, records, legal strategies, models, resources and expert profiles to create an effective team with the right data collated to tackle the workload.

PA, like other legal technology, can take over many legal tasks and reduce the mundane but essential work involved in a case, which usually comes  at a cost. For instance, Everlaw provides control over the end-product in discovery and sifts documents in readiness for lawyer review as a second step. Everlaw’s Conduent Analytics Hub gives insight and real-time visibility across the range of legal and compliance matters, identifying risk that may become a future liability. Commoditisation of such areas of law opens the way to emphasise the people involved, bringing together the right legal talent to do the high-level work supported by predictive analytics to tackle the complexity and sheer volume of data that needs to be managed on behalf of a client.

04 Know your research

Research is a thinking process, not simply data analysis, which puts technology and people into a dynamic relationship. Clean and enriched data underpins the technology and the people. Predictive analytics supports quantitative research, making sense of big data and unwieldy data sets, which support qualitative data. PA uses statistics, algorithms and heuristics to predict outcomes. But the courts will take into account more than the rational as they also include human behaviour and emotion. In case management, lawyers often seek to second-guess the best time to file a case, which jurisdiction to use or what judge might be more sympathetic, but also to understand the human task of legal reasoning by judges.

Ravel Law, part of LexisNexis, seeks to help lawyers to be data scientists through user-friendly data-mining that can uncover patterns in a range of outcomes. Ravel’s Judge Analytics enables lawyers to find out whether a judge is sympathetic towards an argument by analysing past decisions. Lex Machina also mines and analyses data from past lawsuits, revealing connections and making predictions about outcomes, assisting legal departments to select and manage outside counsel. One of its most popular products is ContraxSuite, which uses AI and human input to augment and integrate the user’s experience in document research. However, this is to suggest a sense of optimality and artificial inevitability to legal reasoning that may not be present. Lawyers are well advised to see predictive analytics the way they use their satnav; it offers powerful support, but does not replace them.

05 Know how to make it pay

Whatever we call it, we are seeing a commodisation, increased offshoring or “Amazon-ification” of the law business. Tools like AI are reducing costs by replacing mundane tasks, but not necessarily lawyers. Proofpoint tackles the data growth in the legal field and the bottlenecks experienced by firms, seeking to reduce the financial and time-costs associated with ediscovery, while keeping processes in the control of lawyers instead of third parties. Predictive analytics is part of a shift away from the billable-hour model and traditional “pass-through” pricing, as firms seek to maintain profits and reduce costs. Fast access to strategic insights into law-firm operations can generate significant time-savings for senior partners as they shift gears from non-billable administrative work to billable client work, thus increasing competitive advantage.

What may be happening is an evolution of lawyers, raising them to new levels of service with the expensive labour-intensive tasks taken out of the equation. Certainly we will see more cases like JPMorgan Chase, which  reduced some 360,000 billable hours at an average of $200 an hour resulting in $72 million of legal fees evaporating. Equally, there are opportunities for firms to use predictive analytics in identifying new business, being more effective in bringing new clients on board, leveraging data use within the firm, cross-selling and upselling existing clients, and reducing their own operational costs. Thus creating better metrics, enhancing partner profitability, matching lawyers to client need, using gig lawyers or other consultants and more creative fee structures can all be advanced by effective use of PA.