Dispute analytics is the practice of utilising data from past cases to better inform lawyers on their litigation strategy, and even on whether to litigate at all. The analysis of data has long been praised as a tool for addressing some of the legal industry’s problems (speeding up document review and optimising legal operations for example), and it certainly seems that reaching its newest frontier - dispute analytics - may prove more valuable than even the data analysts themselves could have predicted.

The problem

The pursuit of litigation often involves highly complex financial decisions, with outcomes having vast commercial implications for a business and its employees. Traditionally however, it seems that litigation is to the legal market what the legal market is to the rest of the commercial realm: the slowest to adopt innovative practices. Indeed, it comes to many as a surprise to witness just how strangely most litigation decisions are made, as they often rely on gut instinct or experience of a lead partner, rather than operating within a tactical framework as to how similar decisions have fared in the past. This is precisely what is at odds with how other forms of complex corporate decisions are made.

A solution

Whilst an undeniably tragic public health crisis, the Covid-19 pandemic possesses significant second-order effects on the economy, such as increased restructuring and insolvency as a result of mounting debt and crippled profits. As a result, one can only expect increased disputes in certain areas ranging from bankruptcy to breach of contract. Now more than ever before, clients will need to harness the cash they have available - if any - to save their businesses from falling on the wrong side of a court decision. This is where dispute analytics fits in, offering the ever attractive ‘faster, better, cheaper’ solution to its users.

At its core, the litigation process entails a careful balance between managing client expectations, and predicting possible outcomes. What’s more, all predictions, as Philip Tetlock points out, involve a ‘base-rate’, which is the rate of success of past claims. According to Tetlock, it is astonishingly easy to have a bad anchor, a statistical baseline that reflects more the beliefs and biases of legal advisors rather than any concrete data set. Therefore, when making decisions pertinent to one’s litigation strategy, it is fundamental to start with a healthy, data driven base-rate, and then adjust our prediction according to the specific facts of a case. In essence, better analysis of data allows better decisions, which in turns allows lawyers to better advise, and manage the expectations of their clients.

Consider this compelling analogy, provided by Tetlock himself:

“The Renzettis live in a small house. Frank Renzetti is forty-four and works as a bookkeeper for a moving company. Mary Renzetti is thirty-five and works part-time at a day care. They have one child, Tommy, who is five. My question: How likely is it that the Renzettis have a pet?“

To answer that, most people would zero in on the family’s details. “Renzetti is an Italian name,” someone might think. That may mean Frank grew up with lots of brothers and sisters, but he’s only got one child. He probably wants to have a big family but he can’t afford it. So it would make sense that he compensated a little by getting a pet.”[1]

Yet, as Tetlock points out, this compelling storytelling cannot outweigh the value of knowing what percentage of American households own a pet, which is in fact 67%. As such, one should start with a 67% base rate, and only then should we use the specifics of the Renzetti family to adjust that number up or down.

This same thinking can be applied to litigation decisions, where lawyers can better answer their clients’ questions of “when should I expect a decision?” and “how much will I be paid?” based on past outcomes. One can answer these questions by using the vast data that sits within precedent to investigate questions such as:

  • How has the opposing counsel conducted similar negotiations in the past (eg how many claims have they settled or fought, and how long has this taken?)
  • How substantial is the typical remedy?
  • How has the decision maker approached cases like this in the past (eg which textbooks they tend to rely on, how frequently they treat particular arguments as decisive, how often they tend to disagree with precedent)? It is not a far stretch to assume that factors like sex, nationality or political stance of the counsel may also influence certain unconscious biases in the decision maker.  
  • What fact patterns typically result in success? Should this number inform our decision to settle - perhaps using Alternative Dispute Resolution (ADR) - or to litigate?

These answerable questions all point to the undeniable advantages that dispute analytics bring to the table:


Past claim features - in aggregate - are extremely powerful in predicting outcomes. For example, startups like Solomonic are already demonstrating the unparalleled insight into - and uncanny prediction accuracy of - complex disputes cases, from evaluating prospects to customising arguments.[2]


Junior lawyers no longer need to wade through masses of case law to find corresponding fact patterns, or worse, previously successful or unsuccessful arguments. Law firms like HSF understand this, citing AI as being capable of taking search functionality several steps further than the traditional approach, allowing programmes to “understand the context of words used; analyse and connect relevant information; and cluster by theme (or cause of action) producing faster and more tailored results.”[3]


Finally, firms that are more hard pressed for cash than ever before can put their money into more useful avenues, potentially even deciding that settlement would save more than a potentially long-winded dispute process.[4]

Challenges to Dispute Analytics

Nonetheless, this opportunity will unfortunately not be captured universally, as the arguably most intriguing element of dispute analytics, the analysis of judge behaviour and decision making, is banned in France, who seemingly prefer the old way of doing things. It is shocking to see this course taken by the French, at least from a UK perspective, since legal predictability is considered fundamental to the rule of law[5], and litigation is usually considered one of the most inefficient legal mechanisms to pursue.

Furthermore, a more troubling question is whether we can effectively rely on analytics based on previous decisions in a now fundamentally different socio-political and economic era? We have already seen how AI models have been messed with as a result of our changing behaviour[6], and it is unclear as to how judges will approach the coming wave of litigation as a result of this global economic disaster, particularly if the court system is overburdened with this surge in new cases. Indeed, using data on the whole is a tricky business: if it is mismanaged or inadequately protected, the vast swathes of data generated could “create real headaches”, according to Morgan and Reed.[7]

Moreover, unfortunately no litigation outcome guarantees a saved business, if, as a result of the pandemic, liabilities have exceeded assets for too long. Even if success is likely, some businesses may not even have the cash available to follow through. Thus, this could also be an opportune moment for litigation funders, who could capitalise on providing financial backing to the most promising causes, earning a substantial sum in return for their services.


Overall, whilst outcomes are not purely based on decision quality, and will almost always feature an element of unpredictability and luck, it certainly seems that now, more than ever before, we must steer clear of the approach adopted by our french colleagues across the channel. To act as if litigation were to remain an art, and not develop into a science, is to be stuck in the ways of the past: times are changing, and we must in turn hasten our course into the depths of data analytics - the lives and livelihoods of so many could depend on it.

Author: Bernie Rivard - https://www.linkedin.com/in/bernierivard/

Edited: Panteleimon Athanasiou

Further Reading:



[1]Philip Tetlock, ‘Superforecasting: The Art and Science of Prediction’ (first published 2015, Crown Publishers)

[2]https://www.solomonic.co.uk/videos, accessed June 26 2020

[3]Morgan and Reed, Dispute Resolution in the era of big data and AI’ (Herbert Smith Freehills, Legal briefings, 18 September 2019) <https://www.herbertsmithfreehills.com/latest-thinking/dispute-resolution-in-the-era-of-big-data-and-ai> accessed June 26 2020

[4]Masoud Gerami, ‘The future of litigation analytics’ (The Barrister, 21 November 2017) <http://www.barristermagazine.com/the-future-of-litigation-analytics/> accessed June 26 2020

[5]Tom Bingham, The Rule of Law (first published in 2010, Penguin Books) p69

[6]Will Douglas Heaven, ‘Our weird behaviour during the pandemic is messing with AI models’ (MIT Technology Reviews, 11 May 2020) <https://www.technologyreview.com/2020/05/11/1001563/covid-pandemic-broken-ai-machine-learning-amazon-retail-fraud-humans-in-the-loop/> accessed June 26 2020

[7]Morgan and Reed, Dispute Resolution in the era of big data and AI’ (Herbert Smith Freehills, Legal briefings, 18 September 2019) <https://www.herbertsmithfreehills.com/latest-thinking/dispute-resolution-in-the-era-of-big-data-and-ai> accessed June 26 2020