Note: This article is based on an interview with PainWorth’s Mike Zouhri and Aaron Budnick on “The Insurance Podcast.” You can listen to the full podcast here.
Some edits have been made for clarity and brevity.
Pete Tessier: Mike, tell us, what is the opportunity for the disruption that PainWorth is embarking on? You created a company that is at the apex point between insurance claims and legal representation. What was the result and how did that come to be?
Mike Zouhri (CEO of PainWorth): The biggest obstacle for the personal injury claims process is a lack of knowledge. It seems and looks like it's going to be an intimidating thing. And the reality is, it is fairly simple cut and dry.
As soon as I learned the rules, and the procedures that had to be followed, I recognized that this was one of the simplest things that a machine can automate.
So I just designed a system that could automate the process! In truth, there's often really only 5 general categories of claims loss for personal injury claims situation. You're looking at things like:
Most of those categories of damages are simple arithmetic: you add up a bunch of numbers, and you get a result. And that's the value of your personal injury claim.
But the one category that's a little bit more complicated is pain and suffering. And the way that is supposed to be calculated is:
But lawyers don't often do that. Instead, they might look find 2 or 3 kind-of-related examples of case law, and then just use that to sort of “guesstimate” the amount.
And—what I was even more surprised to learn is—that, oftentimes, the adjuster on the other end effectively does the same thing. So you’ve got two sides that are effectively just guessing numbers.
And, if there's nobody out there to validate your assumptions, or say that you're doing it wrong or doing it right, you effectively work in an echo chamber, and you fill yourself with your own biases over time. I could talk to two different personal injury lawyers—in fact, I have done this experiment—and they'll come up with two completely different numbers for the exact same claim.
And the same is true on the other side of the table. Adjusters will come up with different numbers based off of where they worked previously, the people they worked around, and the number of years of experience that they have.
When humans come up with the numbers, it is really extremely variable. The magic of a machine is that a machine can actually go through 1000s of records of case law, aggregate them all, and then provide a statistical summary of all of the cases. So you can really hone in on a more precise value, and remove the guesswork.
That's effectively the most magical thing about what I did, in creating PainWorth. When we started, I was effectively a small operation, just me and a couple friends who were helping me out. And we started showing it off and talking about it. And then that's when that's when the opportunity sort of made itself known.
During the process, access-to-justice groups actually started reaching out. And one of them phoned me up and said: “Hey, listen, Mike, you don't realize what you've done. But what you’ve done is that you've built a tool that can make a huge dent in access-to-justice issues.”
And then they started sharing some really, really interesting figures with me, something like:
“Every four years, 15 million Canadians are completely locked out of the justice system.”
It was an astronomical number. I couldn't believe that it was that high. And access to justice groups started urging me to make PainWorth publicly available.
I still remember the day. It was July 4, 2019, we just put up a simple form on the internet. We said: “Listen, public, if you also have this problem, and you'd be interested in this tool, sign up and let us know.”
We launched our ugly baby, super lean MVP—minimum viable product—on October 29, 2019.
The day that we launched, we had over $5 million in personal injury claims signed up to use the platform, which actually made PainWorth larger by case volume than most of the little personal injury firms and professionals in our geographic area.
Pete Tessier: You talked about bias with adjusters and the difference between how adjusters and lawyers look at things.
You know, one of the things that I'm not, I'm no statistician. But I do know what the term recency bias means. It’s when you're only looking at what's just recently happened, what you've recently experienced.
And if you don't have a paralegal team that goes in and looks the details of the case you’re working on, you don't know what you could be accessing, because it's not in your mind. Right?
Aaron Budnick: Confirmation bias is another major bias. The human brain, unfortunately, is subject to a lot of biases that allow us to use even basic statistical information in a way that's really not very effective or not what it's intended to be used for, or to create statistics where statistics should not be created.
And then people use that information as a decision-making basis, which I think we've seen a lot with the lawyers and with adjusters.
And there's justification. For instance, if you make an offer to an insured, and they accept that offer, then that feels like that means it was a good offer, when that may or may not actually be the case. It could have been too high or it could have been too low. And without some system to verify that there's just no way to really know.
The only two ways to really effectively determine what the value of a personal injury claim is:
That doesn't mean just one time but for it to happen over and over again. It needs to play out over time, so we can establish reliable averages.
Unfortunately, an adjuster will only deal with a finite number of personal injury claims over the entire course of their career.
Certainly, that makes it difficult to come up with a really good set of data that you can use in an effective way as an individual or even within most insurance companies or testing firms.
Pete Tessier: So Mike, you've developed a machine learning model here. And the idea is that you are pulling information out of case files related to personal injury claims, obviously. So you draw your data from the legal side.
Is there any information that you can draw from past settlements that haven't gone to court? Or is there out-of-court settlement documents that you can use? Where are you drawing out all the information to help make this settlement matrix happen?
Mike Zouhri: We have to provide people with the best possible service. If you don't hold yourself to the highest possible benchmark, you're not really giving people a compelling enough reason not to use the existing systems and infrastructure.
The very same things that we're talking about are broken. So you have to provide the absolute best.
And the gold standard for everything in this entire space is always court.
It's always what a judge has said what under the eyes of the law is a fair and accurate assessment of a thing.
So to answer your question, we benchmark ourselves completely around court records.
Interestingly, I do think that there are learnings to be had around settlement data too. There are certainly things that we’ve noticed and heard from our users regarding settlements.
For instance, certain times of year, certain types of people, in certain industries who might be facing economic stability or economic need, might feel to less pressured or more pressured to settle one way or another.
There are probably things like that, in that race, age, and gender bias and personal socio-economics may play a role in settlement figures. I think those are important things.
However, there is also some interesting data around the economic model of settling:
And then we can start saying to an insurance company: “Listen, it's going to take you on average—based off of your historical performance—6 months to 8 months to close this record. This record is probably going to cost your company an additional $10,000.” But the claimants only asked for an additional $5,000. So instead of fighting it, you might as well just pay the $5,000 because you're going to end up netting yourself $5,000 in potential savings that you would lose otherwise.
There's actually a really good anecdote. About a year ago, Aaron and I sat down with one insurance company and they shared a story about a frontline adjuster who was settling a claim and frontline adjusters decided the personal injury claim was worth $20,000. The claimant wanted something like $25,000 but the adjuster was adamant that it was only worth $20,000 and fought it for 18 months or 2 years—something like that—and ended up getting his or her way, finally settling the case at $20,000.
But the CEO of the company was not impressed with the situation at all. Because all that extra time spent arguing ended up costing the company an additional $15,000 in overhead. Unnecessarily so.
So from the accounting standpoint, they'd much rather would have just called it a day at the $25,000 that the personal injury claimant wanted, because it wasn't that big of a spread and it would have actually saved the insurance company money in the longterm. Those are the types of muddy situations that our algorithms can really just cut right through. Because humans have a really difficult time seeing that. Especially if you're online and you don't have organizational oversight.
Pete Tessier: So how does PainWorth help this whole process? It gives value to the consumer, to the adjusters AND to the insurance companies? You figured out what brings value to all three parties, and created a solution that adds that value and reduces an adversarial triangle?
Mike Zouhri: Yes, you zeroed in on it right there at the end. The current model we’re trying to replace is this adversarial system that is introduces all kinds of inefficiencies.
We aim to cut out those inefficiencies by having a non-adversarial approach, allowing the claimant and the insurance company or adjuster to collaborate and negotiate directly—cutting out all the middlemen effectively.
We have been told that involving the typical middlemen (personal injury lawyers, insurance lawyers, economists, expert witnesses, etc) instantly drives up the costs.
For example, one insurance company told me that on his books, he expects a $100,000 settlement—when all is said and done—to actually cost somewhere in the ballpark of $130,000 to $140,000. And, after that settlement has been paid out the $100,000, and the claimant’s lawyer has been paid, the claimant is realistically only taking home about $60,000.
So we're looking at a drop from $130,000 to settle a $100,000 claim where the claimant only takes home $60,000. That's $70,000 of wasted inefficiency, just sitting up in the middle.
And there are several reasons for it:
If everything that I saw for the professionals was abhorrent in terms of how many of them go about evaluating a claim, it's almost impossible then to expect a layperson to be able to come up with anything at all.
And that there is the root bottleneck of the problem: Claimants just don't know. And they're scared, and they don't want to be taken advantage of.
So they do the thing they don't want to do, they hire a personal injury lawyer to help them navigate what should be a simple process. They bring in their middlemen, and then the insurance company has to bring in theirs.
As soon as that happens, the cost instantly goes up. And the process instantly slows down for everybody involved—claimants and insurers. And it's really, really bad.
I didn't realize that the problem was so painful on the insurance side too. But that was something that we learned immediately after we launched because insurance companies started reaching out to PainWorth and sharing that information with us. The way that they see it is almost the same way that claimants see it.
We are a transparent third-party, neutral data science company that just shows all the data, all the evidence, all the steps. And with that gives confidence to both sides so they can directly negotiate. So they can directly communicate and directly settle everything.
One of the interesting things that we also do is we show the parties, the data and the cost of stalling and the cost of going down that legal route unnecessarily. We can really show when that legal route is not needed or can be avoided.
We already know that more than 99% of cases end up settling outside of court, meaning court is probably not necessary for them. But there are real world costs to that.
So we have a calculator that shows people: “Listen, you could do it this way, and you could fight for $100,000 and it will take you 5 years on average here in Canada.”
That is actually the average number of years from incident date to settlement date that we see in the court records.
It takes about 5 years and—for a $100,000 case—lawyers fees and disbursements are about 40% of a total personal injury claim, so $40,000. So you would go home, in 5 years, with $60,000.
Or, you could use PainWorth’s personal injury claim data… and—knowing that a fair assessment of your total claim is $100,000—you could negotiate directly with your insurance company. And, without a lawyer, and quickly call it a day for $80,000.
You settled out of court AND you're better off because you're $20,000 or 33% ahead now then you would have been with a lawyer. And if you do that, and you also save yourself 5 years of your life dealing with lawyers and insurance companies.
On the insurance side, it's a huge win-win too, because they just cut their costs from $130,000 down to $80,000—or maybe closer to $90,000 as they still have a bit of overhead to pay.
Mira Soullen - March 17th, 2021