November 16, 2017

Mark Walls: The Myth of Predictive Analytics

By Mark Walls, Vice President Communications & Strategic Analysis, Safety National

Mark WallsThe use of predictive analytics on workers’ compensation claims has been a hot topic for several years. Companies tout that their predictive analytics can reduce claims costs in many different areas. I have even seen published studies talking about the costs savings achieved with claims-related predictive analytics. There is just one problem with all of these discussions.

They are all a myth.

Yes, you read that correctly. For all the hype, I have yet to see a single predictive analytics claims model that actually saves money.

Please allow me to explain. The purpose of predictive analytics on claims is to identify the potential for problems. Psychosocial screening tools look for things in the injured worker’s background that could impede their recovery. Data mining models look for key words and phrases in the medical records that indicate the claim has potential for adverse development. Some models monitor for prescription drug use and identify dangerous interactions or the potential for abuse. There are a wide variety of these models and they all have one thing in common. Their output is a red flag. A warning. They provide an indication that there is potential for something to go wrong.

Red flags do not produce better outcomes on claims, nor do they provide cost savings. They simply tell you something needs attention.

I’m not saying that predictive analytics on claims do not have value. They are a tool in the tool box. There is value in flagging claims with potential for problems. However, the potential for cost savings doesn’t come from the flag, but what you do in response to it. You need to take action and do something differently than you would have done without the flag. Common intervention strategies include reassigning the claim to a more-experienced adjuster, assigning a nurse case manager to the file and referring the injured worker to a new physician for an evaluation.

All these intervention strategies cost money. That is why the claims predictive models I have seen tend to increase costs in the first year. You have both the costs of the modeling, plus the costs of the intervention. To change the course of the claim, you have to spend money.

The hope is, by changing the course of action on the claims you will reduce the long-term costs on the claim. The money invested in the front end will result in savings on the back end. Because workers’ compensation claims have a long tail, it can take a few years to know whether or not the different course of action on the claims actually produced cost savings.

As much as people like to talk about the “new” technology of claims predictive analytics, the intervention techniques that may actually produce the cost savings have been around for many years. If injured workers are treated by high-quality physicians, their medical outcomes are likely to be better. Experienced and well-trained adjusters tend to get better results than those with a lesser skill set. The use of nurse case management in the right situation can improve communication with the injured worker, medical providers, claims handlers, and ultimately lead to a faster return to work and lower costs on the claim. Taking the time to speak with the injured worker, explain the process, listen to their concerns and answer all their questions can reduce litigation. All these intervention techniques have been shown to produce cost savings and you don’t need predictive analytics on your claims in order to use them.

In the end, good old-fashioned claims handling skills are still the best way to achieve superior outcomes on claims. Who could have predicted that?

Mark Walls
Mark Walls is the Vice President Communications & Strategic Analysis with Safety National where his focus is on industry education and monitoring workers’ compensation issues nationwide. The majority of Mark’s 25-year career was on the claim side, giving him perspective into the day-to-day challenges faced by employers, carriers and TPAs. Mark is also the founder of the Work Comp Analysis Group on LinkedIn. With over 25,000 members, it is the largest online discussion community dedicated exclusively to workers’ compensation issues. In addition, Mark serves as co-host of the Out Front Ideas educational series, is a frequent speaker at industry conferences, writes columns for a variety of sources and is quoted regularly in multiple media outlets.

Safety National
Safety National 75Safety National is a leading provider of alternative risk funding products such as excess workers’ compensation, deductible casualty, loss portfolio transfers and reinsurance. Safety National is a member of the Tokio Marine Group and is rated “A+” (Superior), FSC XIV by A.M. Best. Learn more at

Safety National is a WorkCompWire ad partner.
This is not a paid placement.

  • RSS
  • Twitter
  • LinkedIn