Slashdot Mirror


Insurance Startup Uses Behavioral Science To Keep Customers Honest (fastcompany.com)

tedlistens quotes a report from Fast Company: Insurance startup Lemonade won itself headlines in January with the boast that it had successfully approved a claim in just three seconds. In that time, Lemonade's software had run 18 anti-fraud algorithms and sent a payment to the lucky customer's bank account -- a process that would have taken a traditional property and casualty insurer days, if not weeks. But it's what happened before Lemonade's artificial intelligence kicked into gear that makes the renegade insurer so potentially disruptive to this trillion-dollar industry, for which premiums alone comprise 7% of U.S. GDP. The customer, Brooklyn educator Brandon Pham, opened Lemonade's mobile app, signed an "honesty pledge" to attest to the truth of his claim, and then recorded a short video explaining that his Canada Goose parka, worth nearly $1,000, had been stolen. That deceptively simple claims process is the byproduct of academic research on psychology and behavioral economics conducted by Dan Arielyblog, one of the field's most prominent voices and Lemonade's chief behavioral officer. "There's a lot of science about when people behave and misbehave that has not been put to use," says Lemonade cofounder and CEO Daniel Schreiber. Lemonade is even applying behavioral science to itself, publishing unusually transparent blog posts that include data on customer growth, bank account balances, and more.

52 comments

  1. A more important question by Anonymous Coward · · Score: 0

    What algorithms will keep *them* honest, keep them from quietly tweaking things so that statistically slightly more (but increasing every few years) are rejected as fraud attempts with no recourse than actually are attempting such a feat?

    "Oh well he's already down on his luck *and* his car got stolen? Yeah no, we have a fraud case on our hands"

    1. Re:A more important question by jonwil · · Score: 2

      The founder of this company has set things up such that they get a fixed percentage of every premium people pay and everything else that doesn't get paid out in claims gets donated to a worthy cause.
      So there is no way to increase their revenue/profit by denying claims.

    2. Re:A more important question by Anonymous Coward · · Score: 0

      But there is a way to screw people.

    3. Re:A more important question by stephanruby · · Score: 2

      So there is no way to increase their revenue/profit by denying claims.

      Yes, there is.

      Let's say that the sweet spot for funding this insurance requires a premium of $100 a month, but that your market research indicates that if you sell premiums for $70 a month, that you'll get 10 times as many customers. This will incentivize you to deny claims.

      Also, another thing you could do is to simply overinflate any claims made by your family or your friends. So your incentive would be to deny claims, so that there would be money left over for your friends/family's overinflated/fake claims.

    4. Re:A more important question by YoungManKlaus · · Score: 1

      > gets donated to a worthy cause

      There's your weak spot right there. Because "worthy cause" can be anything, including "charities" that have various people on their board as paid members.

  2. $1000 for a parka?! by Anonymous Coward · · Score: 1

    $1000 for a parka?! What the fuck?! What the hell kind of parka is this?! I know some fashion can be fucking insanely expensive, and that this is probably pretty cheap compared to a lot of crap out there, but seriously. Even specialty coats meant for use in the Arctic, Siberia or the Antarctic can be purchased for a fraction of that.

    1. Re:$1000 for a parka?! by Anonymous Coward · · Score: 0

      He saved all kinds of money by buying insurance from a company that kept its costs low by paying out claims based on a song and dance.

      Also given that this was announced by press release I'm confident it wasn't staged in any way. The fourth paragraph in which the customer gives his testimonial absolutely sounds genuine and not written by a marketing intern.

      -Legal.Troll

    2. Re:$1000 for a parka?! by Anonymous Coward · · Score: 0

      Judging from the company's website, it looks like the kind of stuff people with too much money and no actual athletic ambition buy to make other people think they ski or climb everest or trek to the north pole or whatever the fuck. Heavy coats all seem to be in the $1000 price range give or take, lighter jackets marginally cheaper, beanies and baseball caps $50-$100, scarves for $125, those furry hunting hats with the earflaps around $300.

    3. Re:$1000 for a parka?! by jodido · · Score: 1

      I have one. It's insanely warm--the warmest coat I've ever owned. And I paid a lot less for it.

    4. Re:$1000 for a parka?! by thesupraman · · Score: 2

      Actually, Canada Goose make gear for arctic (and antarctic, the ones I have seen) use, and they ARE the real deal.
      However, yes, many of their items look like fashion accessories for people with more money than sense.

      The larger issue here is going to be when they REFUSE a claim, based on some fuzzy AI probabilistic interpretation
      of how someone acted during the claim. THAT aint going to hold up quite so well. Accepted claims are the EASY part.

      I suspect they are actually relying on people making less false claimed because they are afraid of the 'all knowing AI'

      Good luck with that.

    5. Re:$1000 for a parka?! by russotto · · Score: 2

      I have one. It's insanely warm--the warmest coat I've ever owned. And I paid a lot less for it.

      Naturally you paid a lot less for it, it used to be owned by Mr. Pham.

      Seems like everyone in NYC has one of these. IT'S NOT THAT COLD!

    6. Re:$1000 for a parka?! by Anonymous Coward · · Score: 1

      I suspect they are actually relying on people making less false claimed because they are afraid of the 'all knowing AI'

      False claims, and probably some true claims too. "I'm pretty sure I'm telling the truth, but what if the 'all knowing AI' says I'm not? Will I be punished somehow? Without a chance to explain myself?".

    7. Re:$1000 for a parka?! by Aighearach · · Score: 1

      And if the traditional algorithms actually only take 5 minutes to run, not days or weeks, it is all within the margin of error explained in the fine print!

    8. Re:$1000 for a parka?! by tlhIngan · · Score: 1

      Actually, Canada Goose make gear for arctic (and antarctic, the ones I have seen) use, and they ARE the real deal.
      However, yes, many of their items look like fashion accessories for people with more money than sense.

      Yes, they are pretty much the #1 brand of gear if you want to stay warm while in the arctic or antarctic (the "Canada" part should give it away as to where they're located AND why they're popular).

      In fact, once you head into the arctic region of Canada, pretty much everyone is either wearing native parkas or Canada Goose, and anyone who isn't is freezing. It's so popular it's a VERY counterfeited brand especially in less-cold areas like the United States where you don't need such a warm parka (so they do make a lighter line of clothes). Counterfeiting is not such a big problem in North America, but it is in China where Canada Goose has become somewhat fashionable because it's high end (and local distrust of Chinese brands have made North American brands more popular due to better perceived quality control).

      They used to only make parkas for the region (and any Canadian who traveled to the region had one) but they've long diversified. Being where I am, I started hearing a lot more about Canada Goose maybe 10 years ago. Before that, everyone sorta knew, but unless you traveled there, you didn't really need one (though enough people have for work that there's one in the closet gathering moths).

  3. Done in the 90's? by thinkwaitfast · · Score: 1

    I attended a presentations in the mid 90's sometime by Dr. Hecht-Neilson who had a company that evaluated people for their credit worthiness using neural networks.

    1. Re:Done in the 90's? by Anonymous Coward · · Score: 0

      Well great, then any patents that they filed back in the 90s will have expired.

    2. Re:Done in the 90's? by lucm · · Score: 4, Insightful

      I attended a presentations in the mid 90's sometime by Dr. Hecht-Neilson who had a company that evaluated people for their credit worthiness using neural networks.

      That's like saying that Amazon is making money by providing customers with an online shopping cart.

      Algorithms are no longer a barrier to entry in analytics; you can get them for free from various Apache projects (Spark, Mahout, etc). The challenge is in acquiring the right data sets and finding features that deliver the kind of indicator you need by constantly evaluating samples and tuning your model. Everyone and their neighbor is using neural nets these days; most fail at achieving something meaningful with them.

      --
      lucm, indeed.
    3. Re:Done in the 90's? by speedplane · · Score: 1

      I attended a presentations in the mid 90's sometime by Dr. Hecht-Neilson who had a company that evaluated people for their credit worthiness using neural networks.

      There's these biomass machines with neural networks that do this a lot. It takes a bit of processing power (several billion neurons), and on the latest hardware, it still can take a few hours or days, but it has surprisingly good accuracy.

      --
      Fast Federal Court and I.T.C. updates
    4. Re:Done in the 90's? by Anonymous Coward · · Score: 0

      Neural networks were popular in the 1990's with credit agencies because they afford a "legal" means to discriminate. We couldn't make models that explicitly included things like race because it was illegal. But we could include a lot of proxy indicators and mix them all up in a neural network. If challenged, we had some deniability. See we don't explicitly include race. There's no explicit coefficient that weights the racial mix of your neighborhood. It was science, the future, AI in action.

      It was racism, pure and simple.

  4. Even more important by no-body · · Score: 1

    have this applied to the goons on top - foremost the compulsive liars creating all those smoke screens for doing it apparently right but in reality cheat the world until blood drips out!

  5. Free money for psychopaths? by Anonymous Coward · · Score: 0

    Given the current state of AI generally I can't imagine it being remotely difficult for any half-decent liar to figure out a way to exploit the fuck out of this.

    1. Re:Free money for psychopaths? by lucm · · Score: 2

      It is very difficult to "exploit the fuck out" of insurance companies because they pool hard fraud indicators and they have access to more computing power and advanced software than some guy with a $25/month AWS instance.

      The weakness in this industry is soft fraud: people claiming that their Xbox One Special Edition With 1TB SSD was stolen, while they actually had the Xbox One Cheapskate Edition ("sorry I lost the receipt but I have a blurry picture of this device which unfortunately you can't inspect since it's gone!"). There's room for abuse but it's not really a huge business opportunity unless you control a big network of people who consistently make inflated claims.

      --
      lucm, indeed.
    2. Re: Free money for psychopaths? by Anonymous Coward · · Score: 0

      How does this work with a pyschopath? Seems like an easy way for them to get, say, a new phone every other year.

    3. Re: Free money for psychopaths? by Anonymous Coward · · Score: 0

      Ugh I shouldn't post on /. when drowsy. I suppose the rates would go up right away so the phone example is dumb, and my question was already there in your original post lol. What about a bigger score though, like getting brand new electronics gear because your apartment was "broken" into?

    4. Re: Free money for psychopaths? by Aighearach · · Score: 1

      If somebody is engaging in that sort of behavior because of their mental illness they will most likely escalate the behavior and get caught. It is actually quite rare for somebody to be willing to take the risk of theft or fraud for some small gain and be also OK with just that small gain. The person risk-averse enough to maintain their boundaries and not escalate will probably not take the risk in the first place!

      It may even be some high percentage of the people who get caught for insurance fraud would be in the group you describe.

  6. misleading numbers by lucm · · Score: 5, Interesting

    potentially disruptive to this trillion-dollar industry, for which premiums alone comprise 7% of U.S. GDP

    Those insurance startups don't disrupt the trillion-dollar insurance industry any more than hotels.com disrupt the business of Hilton or Starwood. This is not at all a situation similar to Airbnb or Uber. For the most part these startups are simply an additional revenue stream for the big companies, allowing them to reach out to the low-end market without having to foot the bill for all the automation and streamlining required to turn a profit on policies with a razor-thin margin.

    Traditional insurance companies don't make the bulk of their profit by charging more in premiums than they pay in claims; they make a profit by investing those premiums until the moment where the money goes back to policy holders as claims are submitted. Insurance companies will never be made obsolete by the small peddlers; they will however go out of business if the financial system keeps making low-risk investments worth less than inflation. The disruption here comes from the retards at the Federal Reserve who keep handing taxpayers money for free to Wall Street banks; Lemonade is just an iPhone version of a traditional insurance broker.

    --
    lucm, indeed.
    1. Re:misleading numbers by Anonymous Coward · · Score: 1

      The disruption here comes from the retards at the Federal Reserve who keep handing taxpayers money for free to Wall Street banks

      It's a good thing for the American people that simpletons like you aren't in charge of our monetary policy. We'd appreciate it if you'd just shut up and say thank you, but I suspect that even that is too much to ask from an ingrate such as yourself.

    2. Re:misleading numbers by sydbarrett74 · · Score: 1

      Thank you, Colonel Jessup, sir!

      --
      'He who has to break a thing to find out what it is, has left the path of wisdom.' -- Gandalf to Saruman
    3. Re:misleading numbers by Anonymous Coward · · Score: 0

      TOTALMETE DE ACUERDO
      LOS CURSOS A DISTANCIA y cursos online ahorran tiempo y combustile. cursos a distancia y online.

      CURSOS A DISTANCIA
      http://www.es-cursosadistancia.es/

  7. Re: by Anonymous Coward · · Score: 1

    I bet they didn't donate part of the profits. That is the millennial spin on this one.

  8. ...just like a lead zeppelin... by Anonymous Coward · · Score: 0

    I just paid the renewal premium on my policy. This is an insult. It's the insurance companies that are far to trust.

  9. Unfair by Anonymous Coward · · Score: 0

    The algorithms are forcing us to conform to the opinion of what a small group of people consider to be honest behavior.

  10. Obey the Law by Anonymous Coward · · Score: 0

    Goodhart's law: "When a measure becomes a target, it ceases to be a good measure."

    1. Re: Obey the Law by Anonymous Coward · · Score: 0

      It's not a good measure to begin with if it's unfair. Let humans evolve.

  11. conditional probability? by Anonymous Coward · · Score: 1

    I'd worry that the statistical confidence they depend on is based on a random sample of insurance customers, not a sample of the self-selected customers that their policies will attract. (Recall the Sears Diehard lifetime-battery fiasco? People who bought those batteries tended to keep their car 15 years, unlike the average battery customer.)

  12. industry where algorithms are always wrong by Anonymous Coward · · Score: 0

    I've always had to pay higher rates for insurance, I've always been quoted stupid reasons why (too young, too old, not married, no kids, etc.). Meanwhile, my peers who fit their algorithms were constantly in 'accidents', and I've never been in a collision of any kind, and avoided dozens of tragedys by acting fast when other drivers on the road screw up. I even had higher rates than a married friend with 3 DUIs. Why? Cause they have really retarded algorithms. I seriously doubt these will be any 'smarter'. Especially since most 'AI' that comes out is pretty damn dumb.

    1. Re:industry where algorithms are always wrong by Anonymous Coward · · Score: 0

      It's due to your micropenis.

    2. Re: industry where algorithms are always wrong by Anonymous Coward · · Score: 0

      Maybe you are just a cÃrner case, bit statically it is working fine

  13. Yes but, by Anonymous Coward · · Score: 1

    can i save 15% or more?

  14. No, $1000 for a Canada Goose... by Anonymous Coward · · Score: 0

    named "parka"

  15. Why? by Anonymous Coward · · Score: 0

    ... to attest to the truth of his claim ...

    Does he employ behaviourists to study the video of a claimant? Psychopaths and sociopaths have no qualms about lying to get what they want. The fact he uses "18 anti-fraud algorithms" suggests these emotional speed-bumps don't slow criminals. At best, he is making it more difficult for opportunists.

  16. Onions by speedplane · · Score: 1

    The press release is from lemonade, but it sounds like an onion article.

    --
    Fast Federal Court and I.T.C. updates
  17. So easy a caveman can insure a lemon by Anonymous Coward · · Score: 0

    Those who are gullible enough to insure with Lemonade are already malleable enough to be talked out of their claims by the chatbot lawyers.

  18. three seconds? by Anonymous Coward · · Score: 0

    I wish direct deposits traveled that fast. So far the article should read "three seconds and three days" or something.

  19. Bad idea. by Anonymous Coward · · Score: 0

    You'll end up recompensating the kind if people who think they are entitled to recompensation. Entitlement is a progressive trait and so you may never get around to actually drawing the line somewhere. And if you do, it will be completely different than the lines you draw with other customers. Because your experts are then the customers themselves, and they are not particularly versed in insurance law.

    It's another "reward the assholes and greedy" mechanism of which the U.S. has no shortage.

  20. Not so much... by argStyopa · · Score: 2

    ....because throwing $1000 away validating some dumbasses' claim on a parka (seriously, $1000 for a jacket?) is WORTH IT to be lauded across the world's media organizations for some hand-wavy claims of magically quick claim resolution?

    Behavioral algorithms, my ass. In my day we just called this what it is: a publicity stunt.

    --
    -Styopa
  21. Customers arent the problem by Anonymous Coward · · Score: 0

    What I want to know is if there's any way to keep the insurance -companies- honest.

  22. humans adapt by ooloorie · · Score: 1

    There's a lot of science about when people behave and misbehave that has not been put to use," says Lemonade cofounder and CEO Daniel Schreiber

    Behavioral science generally measures the behavior of people who don't know what the experiment is about, don't know the outcome the experimenter desires, and don't care about the actual outcome. Fraud involves people who have a strong interest in specific outcomes and can shape their behaviors accordingly.

    So, if you use behavioral science, humans will figure it out and adapt; they will "game the system". That's why the outcome of social policies is often unexpected: after you implement the policy, people change their behavior, and the statistical patterns that caused you to adopt the policy change.

    1. Re:humans adapt by Anonymous Coward · · Score: 0

      Exactly. Insurance fraud is a game. You're going to need not just behavioral "science" (often with non-reproducible results), but also game theory and lots and lots and lots of data. But, in the meantime, you can make money off startup investors.

  23. Hard to be less honest than that industry by damn_registrars · · Score: 1

    I see the insurance industry as being on par with lawyers and used car salesmen when it comes to honesty. It's rather telling just how much the public sees them as atrociously dishonest as we see when this one is promising to strive for honesty.

    --
    Damn_registrars has no butt-hole. Damn_registrars has no use for a butt-hole.
  24. This smells like B.S. by rsilvergun · · Score: 1

    claims this small would tend to get approved because it's not worth the time/effort to do major investigation. Most of the real work woulda been done when they customer signed up looking into his history to determine how likely he'd be to file fraudulent claims. The video is just there so they can show it to the courts if he keeps filing claims and they actually investigate them.

    --
    Hi! I make Firefox Plug-ins. Check 'em out @ https://addons.mozilla.org/en-US/firefox/addon/youtube-mp3-podcaster/
  25. What monetary policy by lucm · · Score: 2

    The disruption here comes from the retards at the Federal Reserve who keep handing taxpayers money for free to Wall Street banks

    It's a good thing for the American people that simpletons like you aren't in charge of our monetary policy.

    Okay I will explain it so even you can understand. Whenever the Fed lowers the interest rate, to a point where it gets to zero or near zero, they make it cheaper for the banks to get money from them than from you and me, so the banks have no incentive to pay you and me interest on money we put in their care. That's why a certificate of deposit at Bank of America currently pays a magnificent 0.05% annual return. Yes, this means that in order to make a $1 return you need to invest $2,000 for a year.

    What do you think is the impact of those interest rates? Your pension fund has to take more and more risk just to beat inflation; and that's not enough to meet their payments to current retirees so they have to take even more risk. They have to buy derivative products, they have to invest in startups, they have to gamble YOUR pension money on forex and commodities and whatnot because if they just leave it in the bank it won't be enough to pay you when you retire. And whenever they take a bath like with those CDOs in 2007-2008, they have to double down and take more risk to make up for it.

    But see, although they spin things differently (for instance by excluding inconvenient things from the way they calculate inflation), the geniuses at the Fed can tell that their model of pumping your cash in Wall Street banks for free doesn't work. So what's next? They want to be allowed by law to buy stocks. Yes. They think they're so smart that they should be able to influence the stock market in order to fix the economy they've been sabotaging for decades now.

    And why does the Fed geniuses do that? Either because their mathematical models tell them that the economy needs a strong Goldman Sachs to survive and they won't let reality get in the way ("Russia was not supposed to default on their bonds! People were not supposed to take mortages they can't afford! etc") or because they want to go work at Goldman Sachs after having spent their purgatory 2-3 years at the Fed.

    I'm not a simpleton. The simpletons are the people who gladly bend over and take it from those bastards who play roulette with the savings and pension money of other people.

    --
    lucm, indeed.