We live our lives on social media these days and often may be unaware of just how much information can be garnered from online posts. Insurers can use these sites to verify workers' comp injuries, new purchases or auto accidents, giving them a more accurate picture of risks they're underwriting or claims they're paying. Social media can also be used to prove or disprove fraudulent claims and more. Shane Riedman, president of anti-fraud analytics at Verisk will help carriers and brokers unpack what kinds of data can be found on social sites, how it can help them accurately identify and price risks, and sometimes uncover bad actors who think they're above the law.
Transcription:
Transcripts are generated using a combination of speech recognition software and human transcribers, and may contain errors. Please check the corresponding audio for the authoritative record.
Patti Harman (00:10):
Good morning and welcome to our Leaders session, how insurers use social media for more than just likes. I'm Patti Harman, editor-in-chief of Digital Insurance, and I'll be your host today. We live our lives on social media these days and often may be unaware of just how much information can be gathered from online posts. Insurers can use these sites to verify workers' comp injuries, new purchases or auto accidents, giving them a more accurate picture of the risks they're underwriting or the claims that they're paying. Social media can also be used to prove or disprove fraudulent claims and more. Shane Reman, president of anti-fraud Analytics at Ferris is going to help us unpack what kinds of data can be found on social sites, how it aids insurers and brokers to accurately identify and price risks, and sometimes even uncover bad actors who think that maybe they're above the law. So thank you so much for joining us today, Shane.
Shane Riedman (01:13):
Yeah, it's my pleasure, Patti. Nice to be here.
Patti Harman (01:16):
So insurers collect a lot of information from different sources. Can you share maybe what some of them are and then how do they go about verifying that the data is accurate?
Shane Riedman (01:29):
Yeah, so first I think it's probably important to set a little bit of foundation. Social media is just really one source of information hosted online that can be and is used in the PNC insurance process. And when we think of social media as an information source, it's continuously evolving, but it's important to take a step back and remember that the problems that we're trying to solve with this data, they're not really novel problems. So just a couple of examples of problems that we might solve with social media. In underwriting, we might use this data, this media data to assess a risk, as an example, maybe to improve precision and accuracy by providing a more holistic objective and factual view that describe the risk in question. In claims, it's essentially the same equation. More objective and factual information leads to improved claim outcome accuracy. So there's definite applications of social media data across the life of a policy from the point of sale all the way through to the end of a claim. But where we face challenges regardless of the problem that we're trying to solve, regardless if it's an underwriting problem or a claims problem, the challenges with social media really in my mind are categorized into two areas. The first is in data acquisition, so this is how do we get this information and where do we get it from? And then the second is in data application. So once we have it, what do we do with it and how are we effectively applying it to solve our problems?
(03:25):
The first step acquisition generally, I think signals intelligence is generally pretty easy here. It's most things on the internet are indexed, including social media data, and just about anyone can do a search and find social media data. So finding the information is actually pretty easy, but it's the application that's where things become a little bit more hard. So when we think about application at Veris, it really boils down to two questions. Number one is, do we have the rights to use the data? Or maybe is there a prohibition against using the data? And tied into that question is ethics, is it ethical to use the data? So that's the first question we're always going to ask. Do we have the rights? Is there something that says that we shouldn't be using this data and is our intended use ethical? But once we have it, and it's really how do we make sense of it algorithmically and apply the data to solve our most urgent problems?
(04:29):
So there really is aren't easy questions. It's not a really easy question to answer in social media. It's a complicated, continuously changing landscape for a national insurer, for example, you may have to be, when you're asking that first question, can I use this, is there a prohibition? Is it ethical? We may be having to answer that question through the lens of more than 50 jurisdictions if you write insurance nationally in the United States. And then we have the terms of service for these aggregators and social media services and other data suppliers that are continuously changing. So we have to make sure that even if we find information online that we're not in breach of some sort of a terms of service agreement with the social media company. So the cool thing is there are smart people out there and great companies that are competing to solve these questions and to solve these questions. Many of them partner with us at Verisk where we can help really connect, I think these application service providers to the PNC industry despite or regardless of the size of the PNC company. So it's a little bit, hopefully that foundation sort of helps with our conversation that there's a lot of information out there and really you really have to use care when you're thinking about how to collect it. But the biggest challenge is really how to the data.
Patti Harman (05:58):
That was a great overview. Thank you. And I was really happy to hear you talking about the ethics of both acquisition and how you're using it because those are all important things. And it kind of leads into my next question, which is how is access to this broader range of data changing different aspects of the insurance process like underwriting or risk management and fraud detection? Because I would think that it would be helpful in some instances, but maybe it creates some other questions as well too.
Shane Riedman (06:34):
Yeah, so access to this data, I'm not sure that it's changing the process of insurance. In my mind, it's enhancing the process of insurance by again, providing objective, more holistic information about the subject matter in front of you, whether you're assessing the risk and underwriting, or whether you're trying to determine the validity and the veracity of a claim. The information the individuals and businesses share on social media, it grows exponentially every year. We know this, and the trend has been on that path for many, many years. And with government modernization, courts becoming more tied into the internet, believe it or not. Not all courts are online, but they're getting there. But with government information adding to it, the information sources continue to grow again, but they're not necessarily solving new problems, they're just creating that precision and accuracy in solving our old problems, the problems that we all think about in our business every day.
Patti Harman (07:49):
Okay. So one of the areas that I wanted to talk about is fraud and perspectives on fraud seem to have changed as evidenced by several studies that have shown that respondents thought that there were times when it was like okay to commit fraud in some instances, or maybe they saw that they didn't see it as an issue at all to maybe fudge the amount on their claim or what they were claiming or whatever. Are you seeing some of this, or what are you seeing from your perspective?
Shane Riedman (08:23):
Yeah, it wasn't a surprise, but I'd love to share the results of a study that we sponsored with the Coalition against Insurance fraud in 2023. So this is current information, this is just recent information, and I'll get right to the big reveal. The study that we conducted with the coalition in 2023 confirmed that younger generations have a very different attitude about financial structures, and we saw that reflected in their responses to how they feel and how they see insurance fraud. So two categories. The first individuals in the United States between the ages of 18 and 44 of those that we surveyed, only 73% of them viewed insurance fraud as a crime. So I mean, if you flip that around, and that means 27% more than one out of four Americans between the ages of 18 and 44 do not see insurance fraud as a crime. A crime one out of four, between 18 and 44 do not see insurance fraud as a crime.
(09:38):
It's pretty remarkable, especially when you compare that to the second category. And that's Americans that are over the age of 45. When we pulled Americans over the age of 45, we found that 93% consider insurance fraud crime. So it's less than one out of 10 that see it the opposite way compared to one out of four for the younger generation. So this is a really interesting observation, and it is not really surprising in the sense that we know that younger people are more likely to commit a crime, but what it is surprising is in the sense that these younger generations are more likely to just not see insurance fraud as a crime to begin with. Not at all. And what I think that means is that there's an opportunity for us to do a better job educating the younger generations about the importance of financial structures and financial systems, explaining and educating the younger generations about why insurance is a necessary component to society, how it helps a startup enter a new market, how it helps the restaurant down the street serve great burgers and how it helps my parents and the small businesses that they own take risks that they might not have been able to otherwise take.
(11:02):
And all of this is really good for society. So any erosion of that trust in the insurance market can have damaging ripple effects, in my opinion, across the entirety of society. So this really, it really is something we need to be thinking about and we need to be investing in, is how do we do a better job of educating the younger generations on the importance and the value of the insurance system and financial systems in general?
Patti Harman (11:30):
Yeah, you're so right. And I'm wondering if the ability to purchase insurance online makes it easier, but I think it also gives us a greater opportunity to educate them there because that's where they're used to getting their information from. So if there's some way to work an education component into the purchase process, that might be a great way to do that. But I remember seeing that study and I was just astounded at the difference between the way the different generations view insurance fraud, and it's almost like having a short-term view versus a long-term view. I don't realize that what I do today is going to have ramifications maybe five or 10 years from now, or it's going to make a difference and it will affect people outside of myself or maybe my immediate realm of individuals. So cool. I mentioned in my opening that we live much of our lives on social media, and it can be things as simple as what we eat to where we go, the activities we participate in, and a lot of times we do this without giving any thought to what we're sharing. How has the popularity of social media changed maybe how carriers and other organizations gather and verify information?
Shane Riedman (12:57):
It really has become a standard practice, especially in claim handling. So in our opinion, a carrier, an insurance company that is not in some way leveraging social media data, especially in their claims handling process, they're really putting themselves at a competitive disadvantage. You need to incorporate, it's an imperative to incorporate this really important data source in your process. And I think there's a really interesting example that just comes to mind of where incorporating a very simple and popular example where incorporating some sort of a social media algorithm into a claim handling process could have prevented a pretty large fraud. And the example that comes to mind is this case out of California. You may remember it was during the pandemic, there was a Memphis wrapper named Nuke Beil, great name stole more than $700,000 in unemployment compensation benefits using a fraud scheme against the agency in California that administers that insurance program.
(14:17):
Nuke Bale bragged about this activity in multiple postings on Instagram with a rap video on YouTube, and I can speak to this as if it's fact because he ended up ultimately pleading guilty to this crime in 2022. What this teaches us is that just a simple program that just looks and checks to see if an individual has posted anything on social media, perhaps that mentions the word insurance, or perhaps that mentions the word claims. Just something as simple as that kind of algorithm or that kind of check could have prevented what ended up becoming more than a million dollars in loss to the government. So I think that, yeah, the popularity of social media, it's ubiquitous, but the message there to companies, to carriers is that it's an imperative to incorporate some kind of social media check in your processes, especially in the claim handling side.
Patti Harman (15:20):
And as a follow-up to that, how are insurers using online data that they gather from social media and other outlets as part of their fact gathering process? And you touched on this a little bit earlier, but can you expand on that some?
Shane Riedman (15:34):
Yeah, yeah, yeah. So in claims, I think I can speak to claims and underwriting here, but in claims, I think that the strongest application are probably the most obvious applications, maybe personal injury claims or casualty claims and workers' compensation claims. The idea is really simple there that if you have a claim in front of you presenting a claim for personal injuries, it may be helpful to know if their public activities contradict those claims. That's something that's a very simple example, but the underwriting example is also really interesting. This also comes to mind is I mentioned that my parents are small business owners, and I got a call from my dad the other day who owns a very small company, only two employees, but they have a rather large warehouse space, but the space is really just used for storage. There isn't warehouse crews, there's not a lot of activity happening there.
(16:41):
But his insurance company using publicly available media, including social media and other data sources, and an algorithm was able to identify that the size of the square footage of the warehouse, it doesn't really match up with the number of employees, and it triggered an audit. So of course, he called me and said, what should I do? And my advice is, cooperate, cooperate, cooperate. And the auditor quickly discovered that there was no nefarious activity. But it's a really interesting example that is close enough to me where a carrier is using publicly available media to identify potential risk accuracy opportunities and using that information to trigger audits. So there are quite a few ways to process the data, but that's one that comes to mind for me.
Patti Harman (17:35):
Are there certain types of information that they're searching for, like images or videos or written comments? I'm thinking about the rise of TikTok with all the videos that people are posting, but images on social media are a really big deal. So are there certain kinds of information maybe that they're searching for?
Shane Riedman (17:59):
So there isn't a certain kind. It's really problem specific. The way that you have to think about your data acquisition strategy is about starting with your problem and backing into that and asking yourselves what information, what objective information that's verifiable could help me solve a particular problem in my claim handling process. So if you are trying to solve a problem with fraud in casualty claims, then yeah, social media with videos is going to be a very valuable piece of information. If you're trying to solve a problem with accuracy in underwriting risk, then a document that is posted on a government's publicly available website may provide the answers to that question. So comments about a business written on a review site are also very powerful. We tend to learn quite a bit from that information. So Yelp comments about a business can be really, really insightful.
(19:07):
And an interesting example of that is perhaps we've underwritten a restaurant that says that they don't serve alcohol and that they are closed at 7:00 PM every night. But on the Yelp review, we have reviews from patrons that say, I had too much to drink there last night, and the mechanical bowl that I was writing at 11:00 PM was really fun until I fell off. Those sorts of publicly public reviews that are going to be on social media, those could be really powerful indicators to assess risk. Other pieces of information, aside from what you mentioned, images and video and written comments, even things like news records or what was the published weather for the particular date. It really all comes into play depending on the problem that we're trying to solve.
Patti Harman (20:05):
Wow. I never would've thought of the importance of Yelp reviews and that sort of thing. And I write a lot of reviews for food places, so I'll be really careful about what I getting them in trouble. No, just kidding. How are insurers using data analytics and social media then to flag either unusual or irregular claims, maybe to identify fraud or other issues?
Shane Riedman (20:32):
There's really two approaches to this, to how you can apply the data that you've now learned and found to acquire. The first is an ad hoc investigation. So it could be just an adjuster or an investigator searching the internet and looking at Facebook, Instagram, x, blue sky, and these other sources of readily available information about individuals and businesses. So it's pretty manual and it requires quite a bit of individual judgment and training of an individual of how to conduct these searches, how to find and capture store authenticate the information that they receive. But another method that I think is a little bit more powerful is maybe a more programmatic method, so maybe something you might think of as on mass monitoring and flagging. So this would be using data from social media across your entire casualty claims book as an example, and perhaps only digging deeper when there's some sort of a flag that occurs.
(21:45):
So when the algorithm, for example, detects that there's been a post or there's been information found on social media that may be subject matter related to the claim or may even be contrary or even supportive of the claim. So there's two approaches. There's the ad hoc one off approach, which is inefficient, inexpensive, and time consuming. And then there's the more of a monitoring or an on mass approach, which is I think a little bit more efficient and provides more of an algorithmic or programmatic way of using the information to help with the claims or underwriting processes.
Patti Harman (22:19):
Are there certain social media sites that maybe investigators are more likely to check, say TikTok or Facebook or YouTube as opposed to maybe LinkedIn or some of the other sites that you mentioned? I am not sure across that social media ecosystem if one maybe is more important or used more than maybe some other type of site.
Shane Riedman (22:45):
So we actually think about this in a slightly different way. It's very difficult to keep that running list of all of the sources available.
(22:58):
And I know that I've spoken with clients and their claims professionals who have sort of a checklist on their desk. They're going to check Facebook and then they're going to check Instagram, and that's great, but you really should be using a tool to perform the search. There are commercially available tools now where we've left it to experts who spend all of their time thinking about sources of social media information. And instead of you having to go out and do a search against every source, the tool can do that for you automatically. So we really like the tools. There are plenty of them that will simultaneously search and monitor the sources for information about your subject. So we think a tool is probably a better approach rather than trying to figure out if Facebook is better than YouTube, for example.
Patti Harman (23:54):
Yes. And I can understand that that makes a lot more sense to operate in that manner. Do you think that people have just become so comfortable with posting things online that they don't even think about the information that they're sharing? And I'll preface this by saying I am aware of how fast information can travel, so when I post things, they're usually sunrises, sunsets or food, because those are fairly innocuous. But I think other people are just, sometimes it seems to me like they will post something. I'm like, really? Why would you do that?
Shane Riedman (24:35):
Yeah, no, I that's think that's definitely a trend, a pattern that we're seeing in society, especially in the us especially with those younger generations. It's this sort of freeform post it all mentality. Everything goes on my feed or onto my page or everything goes onto Instagram. That is definitely there. But we also have to be thoughtful and to think about the opposite. So Patti, you post photos of food and sunsets and sunrises, three great things. Me personally, I committed Facebook suicide almost 10 years ago. That means I've just killed my account. And so I think you have to sort of remember that there's definitely a large portion of the population that's going to post everything. Then there's going to be another section of the population that posts zero, that might be me. And then you're going to have something in between. So definitely we're seeing more, at least I believe we're seeing an increase in the population that is just willing to post everything. So that does create even more information and data out there for us to use in our processes.
Patti Harman (25:59):
Yeah. You mentioned earlier the importance of using a tool to comb through some of this. Are there companies that specialize in helping insurers and brokers like comb through social media and other data platforms to gather information?
Shane Riedman (26:16):
Yes, absolutely. So Verisk is always a great starting point, especially for underwriting and claim risk assessment. But we work and partner with a lot of other great companies that are helping to solve these problems as well. Companies like Legent, social Discovery, Carpe Data Scope, now we really like. So there are some great companies that are out there that each are applying their own unique perspective and that have their own unique strategy for how they collect data and then how they apply it. Remember those two foundational questions that we talked about at the beginning. So what we suggest is don't invest all in just one solution provider. Think about having some choice and finding the right provider to solve your specific problem. And that may mean more than one. Maybe that means two or three of these companies that are doing great work to help solve depending on what the particular problem is that you're solving in underwriting or claims. So you have some great partners out there for us.
Patti Harman (27:26):
Great. It's nice to know that there are a number of different opportunities and companies to work with, and I guess in some respects it depends on what type of information you're looking for as well. We've covered a lot over the last few minutes. Is there anything that I haven't asked you about using social media data that you think our audience should know or something else related to that that you think they should be aware of?
Shane Riedman (27:52):
Yeah. I would say that despite the popular news stories and the story about Nuke from Memphis, we should not expect a panacea. So turning on a social media monitoring service or a service that's finding the right kind of intelligence and information from social media data and applying that to your underwriting or claims process is not going to solve every one of your problems. So don't expect a panacea, but I think thoughtfully problem focused application of social media data and similar sources is no longer a nice to have. I think it's an imperative, and I really do believe that you're putting yourself at a competitive disadvantage if you're not applying this really important data source in your claims, your insurance processes.
Patti Harman (28:55):
Well, thank you so much for joining us today, Shane, and for sharing your insights on how insurers can use social media for more than just likes. I also want to thank our audience for joining us today. Please join us for our next Leaders session when we pull out the crystal ball and ask our guests for their prognostications for 2025. I'm Patti Harman for Digital Insurance. Please enjoy the rest of your day.
