I read a growing number of articles about data analytics and AI in the legal market. Few, however, provide as much detail of how systems work as I would like. Around the time I was ruing that fact, I exchanged tweets with Intapp about data analytics. That led to some private messages and, subsequently, a couple of phone calls to learn more about Intapp Pricing and Engagement DNA.

I report here on what I learned in those calls because I think the specifics illustrate more broadly the details of AI and data analytics. As with most vendor- and product-specific posts I write, I am neither reviewing nor comparing products. Rather, I take advantage of opportunities to have interesting conversations.

What I Learned about Intapp Pricing and Engagement DNA

I spoke with Jennifer Roberts, Manager, Strategic Research at Intapp. Her training includes a Master of Public Policy, Econometrics. Econometrics is statistical economics and economic modeling.

She manages a research team that studies the application of data science and artificial intelligence (AI) to problems that current and future Intapp products can solve. Examples of Intapp R&D underway include working with specific firms’ data to predict client retention, identify untapped opportunities for clients, predict conflicts, and support pricing.

The pricing R&D is making its way into the recently available product (officially released this month – June 2019), Intapp Pricing, which includes proprietary AI called “Engagement DNA.” It goes through millions of law firm records and leverages task sets from SALI and the ABA to cluster similar matters. By clustering, the software helps firms better price future matters. Beyond pricing, the output of Engagement DNA helps predict the nature of the work: phases and tasks, resources required, and the estimated matter duration.

Engagement DNA also draws on several analytic approaches such as natural language processing (NLP) applied to time narratives and supervised machine learning to predict phases and tasks. Intapp works closely with law firm pricing teams, hires its own lawyers, and taps industry standards such as SALI to develop training sets to supervise the learning.

The algorithms work by building profiles for each matter and then clustering all matters based on these profiles. Matter profiles draw together 30-40 different attributes (industry, client type, responsible lawyers, duration, etc.) and create an outline that describes the matter and the nature of the work being performed. The clusters determine matter similarity—the ones with closely related profiles in a mathematical multi-dimensional space define the clusters.   (See chart below)

I asked if phase and task codes are really that helpful given lawyer timekeeping hygiene around these is typically weak. Jennifer explained that including them helps show similarities and differences among matters (or clusters in multidimensional space). However, even without a robust set of phase and task codes, by combining data like hours spent by role, practice group and leverage alongside phase and task codes, the algorithms derive a composite variable that enhances clustering and adds more nuance to aid in understanding the nature of the work.

For now, Engagment DNA clusters matters using only financial management systems data. It collects between 40 and 60 attributes per matter from those data. Attributes include matter type, the responsible lawyer, practice area, client characterization (industry, size, and geography), and duration. Over time, Intapp hopes to add attributes by including documents (from firms’ document management systems) as well as time capture (billable and not) and conflicts data. However, these sources add to complexity and raise some difficult questions about permissions, access, and sheer volume. Intapp also plans eventually to add attributes based on external data to further refine the clustering. For example, they plan to add D&B industry information for deeper and more accurate client attributes.

Intapp has found that there really are no “special snowflake” matters, meaning all matters within a firm share similar characteristics with at least some others, which is the basis for the name Engagement DNA: DNA makes an individual both similar to those related but also unique. In my view, even with data to prove otherwise, some partners likely will continue to think their matters are indeed special. Change management and adoption are challenges for any new approach!

Though there are no special snowflakes within a firm, there can be variation across firms when it comes to practices and industries so Intapp designed the algorithms to deal with all these nuances rather than to fit to one firm. One could expand the genetics analogy to say that firms are slightly different species. (Some readers may remember KPCOFGS from high school biology: the last S stands for species.)

Intapp assumes in all its AI work that data are dirty. Engagement DNA therefore includes a methodology to clean and normalize data. All applications of AI require this data wrangling (collecting, cleaning, and normalizing) As I have pointed out in other blog posts and tweets, many articles and some vendors fail to mention the big lift required for wrangling. Some of Intapp’s clean-up relies on the data analytics itself. For example, few firms officially close matters. For pricing analysis, however, it is helpful to distinguish between open and closed matters. The analytics can “virtually close” matters for purpose of pricing.

Engagement DNA today helps firms meet budgets by tracking and monitoring progress, price appropriately, and redistribute scarce resources (i.e. delegating to associates with similar expertise at a lower rate) to scope work properly, increase client value, and protect matter profitability. Additionally, this AI, in combination with other methodology, supports other use cases such as conflicts and extracting key terms from outside counsel guidelines (OCG). For conflicts, the software predicts what is a conflict and what is not.  For OCG, the software applies natural language processing to efficiently search, identify, extract, and code key provisions of outside counsel guidelines so that firms can meet billing and other engagement standards. Over time, Intapp plans to have analytics and AI infused in all elements of the matter life cycle, from identifying the cross-sell opportunity, pitching the work and, ultimately, delivering the engagement.

Concluding Remarks

Engagement DNA represents one type of data analytics. For law firms, accurately budgeting and pricing matters – and knowing how to resource them – have become critical profit drivers. It is no surprise firms invest in products like Intapp Pricing. More generally, the discussion here shows how data analytics likely will permeate law firm practice and business management more and more.

We have to wonder what the equivalent is for in-house counsel. In my view, it should be data analytics to control costs. And I don’t mean outside counsel fees, though that is important and, as noted above, Intapp Pricing already helps law firms add value. What I mean here is that clients have an opportunity to avoid or eliminate costs by practicing preventive law – an element of what I call  #DoLessLaw. To date, there are few signs of demand for such a product. I hope that changes.