Earlier this week, I reported here that the litigation analytics company Gavelytics, which shut down operations in June, has been acquired by a relative newcomer to the legal analytics space, Pre/Dicta, which launched its product in July after two years of development.
Yesterday, I met via Zoom with the CEOs of the two companies, Dan Rabinowitz of Pre/Dicta and Rick Merrill of Gavelytics, to learn more about what led to the deal and what it might mean for the field of litigation analytics.
What follows is a transcript of that conversation, which I have condensed and edited for style and continuity.
BOB: Congratulations to both of you. Let me start by asking each of you why you wanted to do this deal and why you think it makes sense.
DAN: When I decided to launch Pre/Dicta, I wanted to make sure that whatever we did was comprehensive. So we elected to go the federal route and have total coverage of all the federal judges, all the federal jurisdictions, and to cover all the types of suits, rather than limiting it to a particular case type. But at the same time, we also recognized that, at the end of the day, for our continued success, we would ultimately have to have state coverage. The vast majority of cases are in state court – just in terms of potential clients, that obviously opens it up to a substantial number more than just those that practice in federal court. So we had already been contemplating how we might attack that.
As Rick can speak to, it’s a fairly complex problem to get state court data. Every jurisdiction has its own system of record. When it comes to the federal system, whatever limitations PACER has, there’s still a singular system that you can work with. But when it comes to state court data, we were going to have to make a strategic decision about which jurisdictions to begin with. We understood it would be a limited rollout, jurisdiction by jurisdiction. And then there also would be a lot of work up front in terms of cleaning the data classifications, categorizations and the like.
So we understood it was going to be a longer-term project. At the same time, it was critical to our success. So when Rick reached out to me and we started talking, we saw the opportunity for us to leapfrog over all of that. And once we get set up – which shouldn’t take that long, because we’re essentially applying the same approach that we did at the federal level, focusing on judges in order to create our predictions – this would be exponential for our growth. Rick had 25-state coverage and some 200-plus local jurisdictions.
There were two aspects of this that were really appealing. Number one, Rick and Gavelytics had done such a great job of pulling that data together in a meaningful way – not just high level, but really getting more granular. When our team took a look at it, they were thrilled – the state of the data was wonderful. Then the second piece was the IP, that we would also gain the technology to not only work with the data that Gavelytics had already gathered, but that would also allow us to continue on the path, continue to be able to scrape data, continue to be able to classify and process that data.
So it wasn’t simply getting the data set and leaving us to figure out how we are going to keep this updated. It gave us both components that would enable us to move very quickly into state court predictions.
BOB: Rick, what about you?
RICK: We were all very saddened by what happened to Gavelytics, of course, and frankly, I think we just got a little unlucky. Upon the closing of the business, we received a great outpouring of comments, most of it very kind, from clients, from other companies, from partners, from lots of different folks around the industry. And when it became clear that we had closed, we received 15 or more different offers to buy our assets. The offer that was by far the most interesting came from Dan and his team. And the reason for that was the vision about predictions.
As a reminder, Bob, in the analytics space, there are generally deemed to be three different types of analytics. You’ve got descriptive analytics that are merely describing things. Think of baseball statistics, or what Gavelytics or other companies in this space all do. Everyone is describing the past. The concept with descriptive analytics is that you can make an inference about what may happen in the future on the basis of what happened in the past. The next type of analytics would be prescriptive analytics. Those would be analytics that might make a recommendation of some kind – because X happened in the past, you might consider taking action Y.
But the crown jewel of all of this, the Holy Grail, are predictive analytics. Weather reports are predictive analytics. They tell you it’s going to rain on Tuesday, whereas descriptive analytics would be much more like an almanac, to say, in January it tends to rain. You can see how one is fantastically more valuable than the other. If descriptive is all you can do, that’s great, there is genuine value there. And I’m proud of what we did at Gavelytics in descriptive analytics. But we didn’t even try to predict things, nor did we ever claim that we could.
So when I heard from Dan and heard that they wanted to make high-quality, scientifically sound, statistically sound predictions, I was intrigued. And, I have to admit, I was skeptical at first. But as I learned more about it and saw the product, understood some of the behind-the-scenes technology, I was really impressed. I think that the combination of industry-leading federal motion practice predictions coupled with state court coverage that can do much of the same thing is an unmatched product offering in the industry. Nobody’s doing that anywhere that I’m aware of. For that reason, choosing to go with Dan and his team was a very easy decision.
BOB: Dan, when I spoke to you back when you launched, you made the point that part of the reason you’re able to do this kind of predictive analytics is that you’re drawing on a greater body of data than just the docket data. So as you look to state courts, you get the docket data here from Gavelytics, but how do you get the rest of that picture, the rest of the data you’ve been using to distinguish your analytics?
DAN: First of all, Gavelytics has more than simply the high-level docket data. So a lot of that is critical to our algorithmic models. Even when it comes to federal data, none of this, unfortunately, resides in a single repository. For the federal data, we had to go to a number of different sources in order to build our database, in order to build our judicial information. Some of them are very easy to get and some of them, frankly, are very difficult to get. At the state court level, there are similar challenges, and for each jurisdiction, there are challenges specific to that jurisdiction.
We are very interested in biographical information as well as docket information, as well as past history of decisions. As you would imagine, in a state that elects judges versus a state that appoints judges, there’s going to be a certain amount of information that’s going to be in the public space, and in other states it is not going to be as obvious. That certainly is a challenge for us, but we’re well on the way to addressing that. But that is inherent in creating a system, creating predictive modeling that’s predicated on information and data that no one else has thought to look at. Inevitably, because no one’s thought to look at it, no one’s gathered it up into a single repository. That was something that was put on us because of our particular needs. But our particular needs are, frankly, what gets us to those predictive models, what goes well beyond the status quo and really enables you to arrive at those predictions.
BOB: What kind of a timeline you’re looking to in terms of beginning to roll out some of these state analytics?
DAN: We’re hopeful that we can get this out in the next three to four months because, again, there are two aspects to this. One is obtaining the data, and what Gavelytics had is still in such a great condition that it allows us to compress the timeline. And then the other piece, in building out the federal system, we had to do a lot of analysis as to which criteria, which data points, are the most relevant, are the most impactful. We’ve done a lot of that work as it relates to the federal system, and now it’s simply transferring that. While there is some nuance around that, a lot of the heavy lift has already been done. It’s simply now applying it to the states.
BOB: Will that be on a rolling basis, state by state?
DAN: We’ll launch with as many as we can. Our team is already hard at work at this. While you can imagine there are certain states that have priority, we’re not interested in a slow rollout. Much like on the federal side, we recognize that there are many smaller firms that are regional, and a lot of them litigate cases in many different jurisdictions. For us to provide true value for our clients, we at rollout would like to hit as many of the 25 as possible.
BOB: Rick, you’re joining as a strategic advisor. What does that mean? What’s your role going to be going forward?
RICK: I am going to help in the adaptation of the Gavelytics assets into the predictive schema. There will be certain technology help that I will provide in a senior-guidance type role. I will help with general strategy, sales and marketing. I would like to think of myself as the, I don’t know, the guy who’s been there and done that a little bit. We launched our product in September of 2017, so we’ve done a lot of this stuff and we know a lot of people and firms and corporate legal departments and insurance and all that. So I would humbly suggest that I’m an experienced voice in this slice of our industry.
Certainly, I can help the team avoid some of the many mistakes that I made and that we made at Gavelytics, but then also try to focus on some of the things that we did very well and try to do that. I’m sure I will wear many hats and I’ll probably be on a plane a whole lot, since I’m here in sunny, rainy California and Dan’s all the way out east.
Bob, I’ve known you for a number of years now. I’m very excited to say that I’m proud to be a small part of this team. Dan’s really built something great. I think they’re onto something here that’s not true with lots of other legal tech companies that are out there. This one, I think, is a real gem and has a very bright future in front of it.
BOB: Dan, have you taken in any investment or are you bootstrapping or how are you funding all of this?
DAN: I have a partner, just one partner. There’s no need for me to do any sort of capital raise. We’re well capitalized. That’s why you haven’t seen any news announcements that we’ve needed to raise millions of dollars. We’re already well on the way to that. We don’t have any outside money. It’s just the two of us.
BOB: What else might we expect to see from you over the next six months to a year or so? What else are you thinking about or working on?
DAN: Our approach, we believe, is one that can be applied to a number of different docket events that we’re also going to be exploring. While we kicked off with motions to dismiss and now we’re expanding it to the state level, there are any number of different types of motions or outcomes of cases that are driven ultimately by the judge. Knowing the judge is really what both differentiates us from everyone else, but then also really allows for predictions.
The way I like to think about it is like Google. Their capability to predict our buying patterns or the future of what we’re going to buy is because they have deep knowledge about who we are. They understand who we are, where we live, who our friends are, what our relationships are, how much money we make, where went to school, et cetera. Once you have that information, you really can apply it anywhere. That’s what big data means. Having a large data set and effectively using and mining those data allows you to apply predictions in any number of different spaces.
You have it in commerce and now we have it at the judge level. So anywhere else that we can apply predictions throughout the litigation lifecycle, those are areas that we’re exploring. The obvious ones would be summary judgment, expert motions, discovery motions, and the like. But each of those are really just docket events – events that a judge is critical to, that their decision is critical to. If we can understand who they are and their preferences, biases, et cetera, that enables us to make our verifiable predictions.
BOB: When you roll out state analytics, is that also going to remain focused on motions to dismiss to begin with, or will it be a different set of motions?
DAN: For us to roll this out as quickly as possible, yes, we are just going to focus on motions to dismiss – do what we do well right now. But, again, this is not a dead end. We’re looking to keep going through all of the litigation lifecycle.
BOB: When you launched, you spoke of an 86.7% accuracy rate. Is that still the number, or have you gotten better?
DAN: It really depends. The way that we arrived at that number was that we back-tested it through 50,000 motions. But we always get asked the questions, Have you continued to test it? What has been the effect of deploying it in the real world? I was recently on a panel with a federal judge. Before we got on the panel, I went ahead and applied our predictive models to all his cases. In that instance, we were at 88%. Admittedly it was a smaller data set, but it depends.
Hopefully, we will continue to improve our models as we get more information, both in terms of docket information as well as judge information. And, frankly, this is a learning process for us as well. We’re always interested in improving this and refining this. The hope is that, over time, we absolutely will keep pushing the envelope.
BOB: What’s been the reception from the legal market.
DAN: It’s been very positive. We’ve gotten terrific feedback, from defense firms, from plaintiffs’ firms, from insurance companies, from litigation funders – really, across the board, it’s been very positive. At the same time, the reality of the sales cycle is what it is. Especially with new technology, people always want to fully understand that and make sure that they have buy-in across the organization, which makes sense to me. So, yeah, overall, incredibly positive.
RICK: I think it will take a little bit of time. Gavelytics saw this early in our lifecycle when no one knew what analytics was. Think back five, six years, it was unthinkable that you could do what Gavelytics first did. I think it will take a little bit of time for people to understand how this can have a tremendous effect on litigation practice, both from the perspective of law firms, big and small, plaintiff and defense, to insurance companies, to corporate legal departments, and, in particular, to litigation finance. I think they’re going to eat this thing up once they understand it.
This is one of the few pieces of technology out there that could genuinely change how litigators do their jobs. I’m so excited for this year. Once the state court stuff is cooking and added to all the great federal things that the Pre/Dicta already doing, I think it could, over time, literally revolutionize how lawyers do their job.
BOB: You’re preaching to the choir on that point. But I continue to hear from lawyers, ‘I don’t care what the analytics say, my judge is going to hear my brilliance as a lawyer and read this brilliant brief I’ve written and it’s going to come out my way.’
RICK: Dan and I both were big firm litigators, and we felt the same way. We all think that we’re special snowflakes, and our case is the real special one. What I found with Gavelytics when we first started pitching law firms was that there was a generational divide. I hate to say it, but the younger partners, the senior associates, the folks that are a little more technologically comfortable, were the ones that got it much more immediately. The longer-time practitioners who say, ‘I know every judge in New York or wherever,’ you’re not necessarily going to be able to change their minds, and that’s okay.
But I think that the more forward-thinking people, the people that are already into this stuff and recognize its power, I think they’re the ones that are going to get it.
BOB: Anything else that you guys want to mention that I haven’t asked you about?
RICK: It’s going to be a good year. I’m not sure I fully believe that globally, but I think in this context, for predictive analytics, it’s going to be a brilliant year.
DAN: I want to echo that. I have to say that when we initially had our discussions with Rick about the data and we had our team look at it, it was incredibly exciting. The developers and the data scientists were like, ‘This is an amazing data set.’ They were thrilled to have access to it.
For me, that was exciting, and understanding that we can expand to this space was certainly exciting. But I got to spend a couple of days with Rick out in LA, just talking with him. Understanding his appreciation of our technology, with him having been in the space for the amount of time he was and the deep knowledge that he has, and still really respecting what we do, and the fact that he would join us, that it wouldn’t simply be that we would buy it and then he would go to the wayside, was perhaps one of the most exciting aspects of the transaction for me.
RICK: I must say, Dan shared none of this with me during the negotiations. I now feel that Dan underpaid.
Just kidding. But, yeah, we had a great time, and there’s a lot that we can do together, so it’s going to be a bright future.
BOB: Thanks a lot for speaking with me. Appreciate it.