On Risk Webinar Replay: AI in Insurance (Consumer Dynamics)

0:01 Adam

Good morning, everyone. Thanks for joining our discussion on AI and the consumer.

I think it's a really timely topic. It's going to be timely for a long time, but obviously the early, early days. So we're hoping to get a great glimpse of, you know, the framework on what's happening today and, you know, potentially some long-term implications.

And with us, we have David Seider, who is Chief Commercial Officer of Zebra. Dave will talk a bit about Zebra, but they're one of the largest digital insurance auto shopping sites. But importantly, they're just not that. They're also one of the largest digital agencies, which is pretty interesting because that's, you know, I think in a ways a lot where the world is going and AI can accelerate that.

So, the typical format, I'm going to have David talk a bit about Zebra. The two of us will have a good discussion on what's going on around AI and insurance digital today, or insurance period, not just digital. And then we will have some time for questions in the chat box. So please put them in, and we will definitely try and get to them.

David, if you want to talk a bit about Zebra, that'd be great.

1:12 David

Yeah, absolutely. Thanks for having me, Adam. Always happy to join the William Blair events.

For those who don't know, as Adam mentioned, TheZebra is an online insurance comparison marketplace and agency. And so, we do predominantly auto and homeowners’ lines today, but we've got, you know, 130 people in our agency. We're getting traffic from all over the web from all different sources, but in an entirely digital format.

We are able to bind policies today either over the phone or, predominantly, what we do today do today, is over chat, so we can keep people entirely online. We combine auto and home policies, collect payment, do multi-policy attached, all in online flow.

And so we are especially interested in the topic that Adam's bringing up today, as we've spent a lot of time thinking about where traffic will come to us in the future.

2:08 Adam

Sticking with that, and you gave us a thumbnail, but maybe you could talk about, you know, just give us framework of pre-AI two months ago. How does the digital auto process generally work? Just give us a framework.

2:24 David

Sure. So, we get traffic from a lot of different sources, but I'd say a ton of it comes from SEO, search engine optimization, people searching for "I need insurance," and some of our online resources pop up and direct them to our site.

We also get a lot of traffic from general SEM, people searching on Google for "I need car insurance in the state of Florida." TheZebra is listed as a resource there.

Once they hit our site and we can answer their questions on our resource pages, we direct them to a funnel online where they answer questions about themselves, their vehicle, where they live. And all of that then connects us through to all of our insurance carrier partners, who return Rate Call 1 and Rate Call 2 Rates. So essentially estimates for consumers.

They can click on one they're interested in, and then a licensed advisor that is employed by TheZebra takes them through the last leg of the process, which is making sure we've got your VIN, all your information is correct. We do your billing and coverage presentations, and then we collect your information, we bind you on our book of business, and we go along our way.

So, it is really an end-to-end experience for the consumer today, entirely done online.

3:30 Adam
Bring us to the heart of the matter, the potential for AI to become a big part, and it will be a big part, no doubt.

But I guess, where is it going to be a big part?

And I think you hit on it towards the end that, to me, you know, thinking about it as a chicken-or-egg question, that, you know, the ability to do AI all the way through and get your policy bound.

You know that's, well, I'll let you describe it, but very dependent on the Rate Call issue.

And historically, insurance companies have not actually allowed that straight-through binding.

But could you talk about how that works today?

Rate Call 1, Rate Call 2, Rate Call 3, and, you know, where do you think that goes?

4:19 David

Yeah.

So, the way it works today is when somebody comes to our site, we take a limited amount of information, kind of like a cursory glance at a person's profile, and that hits our carriers through all of our integrations and brings back what we call a Rate Call 1 rate.

A Rate Call 1 rate is an estimate at its core.

The Rate Call 1 rate, though, is a good indicative flow for where the consumer should go, what they should click on, who they should go and enter more information with singularly to bring forward what we call a Rate Call 2 and then a Rate Call 3 rate, which are more or less just bindable rates, final rates.

And the difference between a Rate Call 1 and a Rate Call 2 is, I mean, majorly for our carrier partners, is expense.

It is usually a Rate Call 2 where you start running reports.

Your MVRs, the things that cost real money to build out the insurance profile of the consumer.

And so when you think about AI entering the conversation here, I think there is a lot of apprehension from some of our carrier partners, and rightly so, for a couple of different reasons.

But I would say the quote cost is definitely, perhaps, chief among them right now.

One of the two things I would say is chief among them, simply because, you know, Muse or any other agentic part-

5:33 Adam

Sorry, David. So, what sort of cost differential are we talking about once you go to that Rate Call 2, Rate Call 3?

5:38 David

So, I mean, MVRs and all the different reports they run, it really depends on the carrier.

It can cost some of our carrier partners a couple of bucks.

For some, it can cost $20.

And so if you have a volume problem, which is to say a high volume of unproductive quotes, what that can lead to is what amounts to, like, a very high marketing cost.

You just have a lot of dollars leaving the door that don't turn into premium, and that can get upside down on you pretty quick.

6:04 Adam

OK, sorry to cut you off, keep going.

6:07 David

Yeah, but that is the point.

I think, Adam, what you're bringing up there is the right question.

It's like, how much of a cost exposure is there, and for what partners, and at what point?

The way that Muse or any other agent seems to work today is they deploy a browser agent to go and knock on the front door of some of these carriers.

You know, the big carriers like Progressive and Allstate, they have a place where you can enter information and march through the funnel, and Muse will do that kind of with reckless abandon.

It has no regard for, you know, what kind of quote costs are being incurred on the other end, which is obviously keeping some of our carrier partners on edge.

I think, perhaps more importantly, and I'm sure we'll get into this is, is it going through that process compliantly and legally with the way that state regulations are set up today?

And so I think that there is a general fear, sure, about cost exposure from our carrier partners.

And there's also a big concern right now about a compliance exposure and, ultimately, a loss and liability exposure for the carrier partners getting shopped by these agentic tools.

7:09 Adam

So today, in pre-AI, if I understand, most of the large, branded carriers don't allow straight access to Rate Call 2, Rate Call 3. They don't allow you to bind straight through, if I understand correctly. The, you know, the State Farm, Farmers, Geico, whatever. One, is that the case?

And two, do the tier twos allow them still, going pre-AI?

7:35 David

Yeah.

So not really, no.

I mean, I'd say the vast majority of the carrier partners that work with us, we are going to have a licensed advisor in the mix to help facilitate those purchases.

That doesn't mean it doesn't exist anywhere.

There are absolutely carrier partners that have built the rails to return a Rate Call 3 with no licensed advisor in the mix.

But that is, I would say today, a highly curated path.

It is one where a lot of care and attention has gone into how they moderate the costs, at what point the reports get pulled, and, importantly, how the billing and coverage presentations and all of the consumer consents are built into the process.

And so, largely, that is handled by advisors today.

Even if it's over in a chat window, it still gets handled by us.

So, it's not to say it doesn't happen, but I would say it is the exception and not the rule.

8:25 Adam

Is that straight-through binding, is that just roughly, is that 5%, 10%, 15%, just a rough, rough estimate?

8:35 David

So, if you're talking about stuff that's done on TheZebra's website, speaking for myself specifically, the stuff that goes through our agency, we've cut all of it out because we found it to be less effective.

It doesn't mean that there are still some carrier partners where you can flip over to their site outside of our ecosystem and you can buy it on their site.

And so that's still done. That is available out there.

But I would say what goes through TheZebra's agency, our advisors are on every single one of those today.

9:03 Adam

OK.

And so with AI, does that, again, staying with the Rate Call 2, Rate Call 3, that the ability to go straight binding? Do you think that changes the landscape at all?

Does that make it, for the carriers, more attractive, easier, maybe more or less costly to allow Rate Call 2, Rate Call 3?

9:28 David

I would think that the advent of agentic traffic should make it more expensive, I think, to distribute a policy.

And my thinking on this is twofold.

The first one is today we think of insurance shopping as largely episodic.

It is something that you do infrequently as a consumer.

As much as I love insurance, most people do not.

And so it's not something that they want to be doing on a frequent basis.

I think that an agentic partner to a consumer will figure out pretty quickly that the best way to save money is not by shopping once, but by shopping continuously.

And so I think something we will see is an agentic, it's like a Muse, will tell the consumer, yes, I shopped for you. You should let me do this in the background at a regular interval.

And Muse has no real regard for going to Progressive's direct site, or Allstate's direct site, and running through that process, and pulling quotes in the background, and surfacing them to the consumer when it makes sense.

And so whereas the consumer may have shopped once every two or three years, now that consumer may, whether or not they know it, be shopped once every two or three months.

And so what I would keep an eye out for in the ecosystem is: how do you put reins on that?

How do you bridle that expense cost?

And then I would also say I wonder about costs from liability on the back end.

This is a bigger question.

I'm neither a lawyer nor a regulator, but are we seeing the proper consents and presentations when policies are being bound with a tool like Muse?

Does the ultimate consumer ever see what they need to see?

And if you were to go back and argue over a claim, paid or not paid, and you went through that purchase experience, did everything get presented to the end consumer in the way that it should?

And is there a liability and loss exposure there for our carrier partners?

And so, yes, distribution costs are absolutely on everybody's mind.

The longer-term thing here, and I bet you the thing the carriers, the big carriers, are even more worried about, is what does it look like when a purchase is made by Muse or another agent on behalf of the consumer?

And is the stuff that these carriers built into their online purchase flows, is it being presented with high fidelity and faithfully in a way that would keep them compliant in all the ways they want to be?

11:46 Adam

Right, right.

OK.

So it sounds like what I'm hearing is that, from the shopping side, I mean, it's obvious the, you know, the agents' AI process will make it much easier and almost prompt more frequent shopping than maybe in the past.

You know, is there one or two other ways you think, you know, AI will impact or alter the shopping process?

12:10 David

Yeah.

And so we, I'll plug us real quick, but we think of agile education as something that you can get from an advisor live on the phone, which is not necessarily something everybody wants to do.

But I think that there is such a wealth of insurance knowledge, especially from distributors that spend a lot of time shopping so many different profiles and demographics and vehicles and geographic locations.

I think that there is still a big education element that is missing sometimes from insurance purchasing.

I think the consumers are confused as to why a higher or lower deductible, what that means, or what a state minimum is.

And I think that there's an incredible opportunity for AI to step in and answer a lot of those questions.

I do think, once again, not to keep on ringing the same bell here, but we spent a lot of time training our AI tools that we have that are consumer-facing to know the rules of the road and the insurance regulation, what it can say, what it cannot say, when it is veering into a territory that really should be occupied by a licensed advisor and not a non-licensed entity.

And so I think there's a lot of potential and possibility as it relates to how you can educate the consumer, how quickly, how tailored those answers can be for each person at any hour of the day without having to have a licensed advisor on the phone to do non-licensed work.

But I think, critically, you've got to know when to flip it over into a licensed and compliant path.

But I think that we will have better-educated consumers over the next couple of years than we ever have in years prior.

13:42 Adam

And this was crystal ball, but just wondering. The way they, you know, auto, staying with auto, you know, shopping is that there's, from what I've always understood, maybe 25% to 30% of the market is a heavy shopping market because they’ve got to pay a lot of money, right?

And then, you know, 65% to 70% doesn't shop until they have to, unless they've been in an accident, whether their rates have been really moved up.

So, I mean, potentially, do you think it changes that paradigm or, you know, is that more of a social culture that if I don't have to shop my insurance, I'm not going to shop my insurance?

14:18 David

It's a good question.

And I think, strangely, the answer to that question does not lie in people's feelings about insurance.

I think the answer to that question lies in how people feel about sharing their complete information and life with one of these agents.

And so, you know, just using ChatGPT on a Tuesday isn't going to, like, drastically change how you shop for insurance on an ongoing basis.

What will really change that is if you use an AI assistant like Instinct, or Muse, and give it access to your email, your texts, essentially your full profile, where your policy documents will live, your life changes will flow through.

And if it has that constant stream of information, then that enables that shopping to happen.

Then it's as easy as, you know, snapping your fingers and turning it on.

And then you have become a person who shops their insurance frequently because you personally are not doing it.

The friction tax of our industry has disappeared, essentially.

And so I think it has less to do with people's feelings about insurance, in particular.

I think it has everything to do with their feelings on, do I want to share everything, all of my data and personal information, with Muse, with Meta?

And if the answer to that becomes increasingly yes, I think it is easy enough for an AI assistant to say, just suggest you should turn me on for these things when they read your insurance policy docs for the first time.

15:46 Adam

Which has a lot of long-term implications.

Because then if, you know, you're, as an insurance company, as an insurance carrier, if you're good at, you know, customer acquisition, good in a digital environment, you know, could be more of a separation because you always, you know, I think from an insurance company standpoint, you've always had the, you know, the pillar to rest that most of my clients aren't going to shop and aren't going to move.

So again, that is actually really interesting.

So, let's switch now to what's actually going on with AI today.

One first question is, to the extent you know or can see, how much traffic is actually going through the AI interfaces?

16:26 David

Yeah, it's a good question.

I think the fun answer is the real answer, which is lots of people don't know.

We spent a lot of time, even over the last four weeks, working with Cloudflare and other vendors trying to say, how do you figure out what is Muse traffic or Instinct traffic that comes to your site? Because they are launching these browser sessions from behind a proxy wall, it's almost impossible to tell.

We went out of our way to build a Muse connector literally as fast as we could because we believe we need to answer the question from our carrier partners, how much traffic is coming in and how are you going to handle it?

Like, how will you divert this to an experience that we feel is safe and compliant?

But the real answer, Adam, is like nobody actually knows right now.

And the industry is scrambling to answer that question with a higher amount of confidence.

And so, if you look at, you know, on a relative basis, I think that yesterday it was published that Muse has 1.8 million daily active users.

I think that Google does, like, just shy of 14 billion queries daily.

That's probably on a worldwide basis.

And so, on a relative basis, this entire industry is still running largely the way it did before.

What I will say is the difference between those queries on Google and anything that comes out of Muse is, as we talked about before, Muse has a propensity to hit your site multiple times for multiple reasons, over and over.

And so there is an exponential and compounding dynamic in play for each one of those daily active users, should they choose to be interested in insurance.

So today, you know, on October 8th, things really aren't that different.

The question will be, how quickly does that 1.8 million daily active users continue to march up?

And of those daily active users, how many people have exposed their entire profile to Muse to make it a really dangerous tool?

18:14 Adam

One thing that's interesting, just, and you said a couple times, but a more active shopping.

That's not a bad thing for Zebra or some of the other folks in Zebra's, you know, space. And that's actually a good thing.

Is that right?

18:29 David

You know, I think it all depends.

I think that everybody has to kind of prove why they have value in this ecosystem.

I think that we are obviously self-interested but would argue with a straight face that you need somebody who can think about all the different stakeholders in the equation here.

You know, if you were going to march through and only do things in the name of the consumer, as we talked about before, you're going to run carrier rates through the roof when marketing costs go up and liability and loss exposure goes up, and that doesn't work.

But if you only act in the interest of the carriers and you wall off the ecosystem, then the consumers can't shop.

And so, I do think that there is a very logical argument for having somebody at the center of this that is a licensed entity, that is regulated, but has the best interest of all the parties in mind.

I think, importantly too, Adam, Muse is going to have to, or any other of these agents, is going to have to come to a point where they figure out how they want to make money on this transaction.

I think that is one of the more interesting and untouched portions of this that I've seen.

There are so many free tiers, and they have contemplated, and Meta has said this publicly, we're going to make money on a per-transaction-fee basis.

As you probably know, in all 50 states, it is illegal to pay a per-transaction fee to a non-licensed entity.

And so just that construct right now is incredibly complicated, and working out all these individual deals with individual carriers also seems complicated.

And so I think there's a great argument for people that can sit as a nexus in the middle and weigh out all the concerns of all the parties involved.

But maybe more importantly, and simply for economic purposes, have a singular commercial agreement that facilitates it that's legal, candidly.

And so I think we need to find, you know, kind of the rules of the road, quickly.

And we'd argue to put ourselves and others in the middle of that conversation as licensed distributors before things kind of get out of control, candidly.

20:28 Adam

OK.

And as you guys talk strategically, because it obviously impacts your business, do you think a significant ramp-up in flow, is that 2027, is that a 2028, another way?

I mean, in other words, are we going to see a real ramp-up in the next 6 to 12 months? Or just the way the wheels turn, it's going to ramp up no matter what, but does it become a real factor within 12 months or so?

20:55 David

That is the question.

And so it's funny, if you ask my chief AI officer, he'd tell you it's going to happen in three weeks.

And if you ask some of our insurance carrier partners, they say, "What's Muse?"

And so, you know, there's a very wide gap in the middle between that.

What I would say is we build as if it's going to happen soon because if there's any amount of traffic that comes through, you have to be ready for it or else you're going to do something that, as we talked about before, is going to disadvantage one of the stakeholders and the whole system breaks down.

So, we're going to build like it's going to be here now.

We are going to lean into it quickly.

We're going to be ready to catch it when it comes.

It is a good constant reminder, when you think about the bleeding edge of adoption on things like this, that insurance is purchased by everyday Americans.

I spend a lot of time with everyday Americans, and they'd say they're not that interested in Muse right now.

It's not to say they won't be in the future.

And, to keep coming back to this topic, I think a lot of it has to do with data privacy and how comfortable the average American shopper is with sharing a lot of their information on an ongoing and open basis with somebody like Meta.

And so, I think as you see that breakdown, or as there is a higher-trust agent in the market that people have a greater affinity or trust in, I think you see adoption accelerate at that point.

And so, I think it may have more to do with who is offering it in the near future than it does the tool itself, because I think the tools are probably pretty interchangeable.

I think it's just about who can get Americans to trust that they're not going to sell out their information.

22:41 Adam

Today, of the top seven, eight, nine or so who dominate the market, how would you quadrant them as far as moving aggressively to be set up in AI, to be able to take traffic in an AI fashion?

22:49 David

I would say we are very much in an observe-and-report posture right now.

We have discussed our approach and had largely positive responses for all the reasons we've talked about today, which is we spend a lot of time thinking about what their cost load, what their liability looks like.

I would say, in terms of their personal movements, though, they're waiting to see how a lot of things play out.

You know, there's nobody that is sprinting to the goal line right now to make an accommodation or to build something that we know about.

And they may not be sharing it with us, which would also be completely reasonable.

But what I would say is, I think you probably underestimate somebody like Progressive or Allstate at your own peril.

I think we've seen them do this through one iteration of, kind of, digital commerce before, which is the one we're still sitting in right now.

And maybe they were slow at the launch, but they dominated it over time.

And so, while we may not see them sprint out of the gate, I've got no doubt that they are talking about it a lot.

And I would imagine that they will try to, you know, measure three times and cut once in how they implement this.

23:57 Adam

Interesting.

And then I guess this is twofold.

You have a couple of, you know, decent-sized, more tech-oriented, Root to Lemonade, you know, a couple that are, you know, really built with a new chassis.

You know, it would seem like they'd be better set up in this environment for probably two reasons.

One, you know, again, their tech's a little more modern.

The analytics are probably more modern than most of them.

Two, you know, they don't have the big brand to lose, that some of them may lose.

So they can afford to be more aggressive.

But I guess that's sort of a surface observation.

So I don't know. What do you think about, you know, the more tech-oriented players?

24:40 David

Yeah, it's a good question.

You know, I'll start with the second part first.

You mentioned brand, which I think is a really interesting question here.

If you are being presented a price via a text message from a personal agent, in some ways you'd think, well, brand goes out the window.

It's like price will win, right?

But at the end of the day, I think that brand still carries a lot of weight.

I think that the average consumer is concerned about who will take care of them, who will pay their claims.

It's not to say it’s going to win every single time, because it certainly does not.

Price is really the most important factor in insurance shopping.

But I think that if there are two results that are similar, brand will still win.

And so, in some ways, I think that brand can help people follow through and convert more quickly if they're familiar with who they get served up and they say, "I know what that is. I know who that is. Seen them on TV a lot. I'm happy to convert now."

I think what is interesting about this is, less about the technology they have to quote in our current ecosystem and whether or not their front door is open on the web at all.

And what I mean by that is Muse, or Instinct, and others, are deploying browser agents to literally go crawl the websites and just do the human actions.

And I think what we're seeing here is speed matters less, probably, than it used to, because the speed that we think about is how many minutes of the consumer's time is spent interacting with the agent, and the agent goes and does the task.

If it takes that agent 90 seconds or three minutes, the consumer's not working either way.

They're going to go do something else and wait for it to come back.

And so I think maybe the bigger deciding factor is, are you part of an ecosystem that has access to it?

If you are an independent-agency-only carrier, you have no front door for a browser agent to go and crawl through and fill out the forms.

You're on TheZebra? Sure, you have access because we present you out to the market.

And so, in some ways, I'd say it is an equalizer of sorts as it relates to these agents going out and doing a lot of the grunt work for you.

I mean, there are examples of these agents calling local municipalities and, like, paying parking tickets and stuff.

And so they're not afraid to do the gritty, dirty work on your behalf.

And so part of it is, do you have an infrastructure that's easy for them to crawl in the first place?

Is it easy to get rates back?

Is it straightforward and to the point?

And so, as long as you've got that, you should be in pretty good shape.

27:11 Adam

OK.

And then let's switch to what's going on today in the, you know, in the auto insurance market, particularly the digital market.

Is demand from auto insurers still increasing in the third quarter on a year-over-year basis and sequential basis? It's been ramping nicely throughout the year.

But what's going on in the third quarter?

27:33 David

It's been more of the same from earlier in the year. I think demand is steady.

If everyone is reading Adam's reports like they're supposed to, you see that earnings have been better than expected, loss ratios have been better than expected.

That is translating into enthusiasm around growth that's permeating the third quarter and into the fourth quarter.

We haven't seen any signs right now that would indicate folks have any reason to pull back or slow down.

And so enthusiasm is about the same as it has been, and we would anticipate growth to continue at that same pace.

28:04 Adam

OK.

Are, in the last three, four, five months, are you seeing more auto insurers jump into the digital market, increase budget, Tier One versus Tier Two?

Any thoughts there?

28:15 David

Yeah, so I think so. And not just auto. I'd say home has seen a lot of interest in a way that we did not see in years prior.

You know, in 2024 and 2025, the auto insurance market recovered more quickly than the home market, and there was kind of a rush back in.

We're seeing a lot more competition in the home market right now as it relates to folks that want access to our agency to bind policies and folks that want access to our ad-click network too.

And so we're seeing it on both sides of the house.

But I'd also say something interesting we've seen is carriers are getting, I think, more creative and forward-thinking in how they are approaching, especially, the ad-click market.

They've got sophisticated marketing apparatuses.

They always have.

I think that the machine learning and data science teams that are being created in the background, that are helping to drive some of the big decisions, continue to get more and more sophisticated.

And with that level of sophistication comes a higher level of confidence, I think, in the dollars that they spend.

And so that has been something that's been really interesting. It's not just the marketing dollars themselves, it is that commitment to the staff and the strategy that is a little bit more invisible behind the scenes.

But we've seen that at pretty much now every Tier One carrier.

29:35 Adam

OK.

So would you say digital revenue, in general, digital revenue is up in the third quarter year over year and then sequentially, yes or no?

29:49 David

Yes, yes to both.

29:52 Adam

Both OK. On a year-over-year basis, if you don't have an exact, you know, is it more like up 5 to 10, 10 to 20, somewhere in between?

30:01 David

Yeah.

I would say for us, more than that, but I don't want to use us as a proxy here for total industry demand.

And so I would say it's hard for me to tell straight up.

What I will say is they are just as, if not more, aggressive in terms of bid price than they were.

And so that is usually a pretty good indication that growth is being chased by a lot of people.

And so I would expect that growth to be up as a result.

30:29 Adam

I've already got some questions coming in, but please put questions in the chat.

We'll move to that next after one last question for me.

David, what are insurers saying about 2027? What are early indications?

30:47 David

So early indications have been enthusiastic.

You know, we spent time with a lot of our carriers in person last week out at ITC in Vegas, and all of the talk kind of centered around how do we grow more next year.

And so I'd say it is a lot of talk about targeted rate reductions.

There's a lot of talk from the Tier Two carriers about what states can we open up, how can we open the aperture to let more in next year.

And so it would seem that there have definitely been points in time over the last, you know, six years where the talk is more on how do we consolidate losses, how do we focus on retention in the book?

It's not that those things aren't discussed. They always are.

But I would say the conversation is anchored much more on growth this year.

And I'd say, again, people are getting what sounds like fairly ambitious targets passed down to them that they have to go out and hit for 2027.

31:41 Adam

Great.

So, I'll just throw out questions, and we've got a bunch.

So maybe just trying to do these, we've got a little time, but try and do it over the next 10 minutes or so.

We talked about this a bit, but, you know, to the extent you know, I mean, how much actual AI binding is going on, and which carriers are actually allowing AI binding?

And if you don't want to say which carriers, you could just give some color on that.

32:07 David

Yeah.

So, AI binding is an interesting thing because it can happen.

I think it is happening, but I'm not sure if the carriers, I won't speak for them, I don't know if they want it to or not right now.

Like, I don't know if it should be happening because I think the AI binding that is happening right now is people on the early-adoption curve of Muse typing things like, "Go find me better car insurance."

And it's just going and shopping Progressive and Allstate's direct websites.

And, as we talked about before, just kind of marching through the flow.

I think that that is exactly what needs to be kind of corralled and directed in a better and more compliant way right now because I don't think the carriers are crazy about that.

I won't speak for them, again, because we haven't asked them point-blank about that, but that would make me nervous as a carrier.

And so the AI binding that's taking place is probably not the answer for the industry long term.

The answer for the industry long term has to be something that checks all the boxes from, like, a presentations and compliance standpoint.

33:08 Adam

OK.

Insurance carriers use data analytics providers like VERA Risk and RELX to price policies.

Does AI lead them to reevaluate these relationships and bring some of those tools in-house?

33:23 David

That is an excellent question.

I mean, those are the expenses we talked of, right?

I mean, some of them, at least, that go into the quoting process.

And so I think a lot of this answer depends on can you find a way to properly moderate your quote volume.

That is something that we focus on a lot with our carrier partners, is things that we have built internally to predict when to pull a rate for a carrier and when to match them with a consumer.

And so a lot of the answer to this question, I think, will have to depend on, you know, are they working with intermediaries or distribution partners or in an ecosystem that has any concern for the quote volume?

If the answer is no, then you have to move this, I'm guessing, farther back in the quote process.

If it used to happen one-third of the way down the funnel, then you kind of move it almost to the end.

Otherwise, you're going to see these costs go through the roof.

I think the interesting part of this question, I'm not an underwriter, I'm a distributor, is if you don't use those tools, you still have to pull those data points somehow, some way.

Like, how else do you go about getting that information?

Is it feasible to build that in-house?

And even if you do, what is the expense load of running it there?

And so you need the information one way or the other.

And so I don't know how cost-efficient you can get in building it yourself versus getting it externally.

I think what you see first is more a shuffling of where in the process it gets pulled and hopefully a shuffling of where they try to get volume from and whether or not it caters to their desires as it relates to overall quote volume, to, you know, essentially a look-to-book ratio.

35:02 Adam
How do you see AI affecting carrier-agent-policy relationships beyond the shopping process?

And I'll dive in there for a second.

I mean, digital shopping of auto has made a land-based carrier less required for a number of years, and we've seen somewhat of a shift away.

It does AI accelerate that on auto.

But then on the bundle, where still, you know, just traditional agents, you know, that's their bread and butter.

Does AI begin to change that paradigm where you can be less reliant on them?

35:40 David

It's a good question.

I think a lot of it, especially at that bundled end of the market, as you're talking about, Adam, there is a different trust element there.

Just because the dollars get bigger.

It also insures what is most Americans' most important asset, which is their home.

Like, that matters a lot.

And so, there's a big trust curve.

We talked today about, will consumers be trusting enough in an AI agent to just share their personal information on an ongoing basis?

There's another leap beyond that too.

Do you trust it to proffer good enough advice to cover your home?

I mean, that's yet another leap.

And so, if we do get there, I think it'll be a while.

I do think that it is easier for the lines of lesser magnitude, like auto, to see that degrade first.

But I don't think that there are a ton of super-meaningful monoline auto insurance agent relationships out there.

It's not to say that they don't exist. They absolutely do.

But they're famously difficult to maintain because they're economically disadvantageous for the distributor.

And if you have to spend a lot of time on a person you're writing a monoline auto for, it doesn't make a ton of sense.

And so I think that people will continue to have a lot of trust in hearing a licensed person on the other end of the phone say, "Hey, you can trust me. I'm a licensed individual. This is the right thing for your home."

That feels like it holds on a bit longer for me, in my opinion, versus the lines of lesser magnitude like auto.

37:11 Adam

To that point, I've always understood that, you know, the auto insurance have been reticent to buy online.

Home is even a whole different world.

The, you know, risk, most home insurance companies don't allow, you know, online binding.

Is that correct?

And does that hold back, you know, an AI agent from saying, you know, someone's got a house and home, just go get that for me?

Is the home piece even tougher?

37:36 David

Potentially yeah, I think so.

It's not that there are none.

You know, Kin is a great example of a company that is kind of built for this moment, right?

They're ready to bind online right now.

But a lot of the more traditional home insurers prefer to bind with an agent in the loop.

It's a good place to make sure that an advisor can ensure that a customer knows everything that is happening because when those claims come through, they cost a lot of money and there's a lot of liability thrown around.

And so that's a nice security blanket, to have a licensed individual in the loop there.

So yes, it is going to be tougher to wedge into that market than it will be for auto.

38:13 Adam

OK, good question.

Do you have any thoughts on AI liability coverage? This is from the questionnaire.

Seems that many retail brokers and primary brokers are not on board offering AI coverage.

I've heard there actually isn't a lot of AI coverage.

But yeah, any thoughts on that?

38:36 David

Not my specialty, I'll say that, a little bit outside my line.

But I think that, just like cyber coverage before it, I think that if there is enough capacity out there that is interested in it and can back it, then it will proliferate quickly in the market.

So far as I know much about that.

But that's a little bit outside my expertise.

39:00 Adam

Let's jump into a different question.

Who do you think are the most innovative insurers on the market today?

39:08 David

Man, that's a good question.

Innovative is a very loaded question because I think that innovation comes in a lot of different forms.

And so, you know, I would say that innovation in distribution can look very different for a lot of different people.

I think GEICO is doing a wonderful job right now of entering the digital market and has done a great job over the last 18 months of really carving out a real foothold in what they are doing in terms of kind of marketplace-based lead sourcing and binding.

I think that Progressive runs probably one of the most regimented and innovative bidding-for-ad-clicks operations in this entire industry.

I think they do a wonderful job.

I think that Allstate, the data science work they do on bidding for ad clicks is some of the most innovative work.

And so there's so much to be said here that goes further into claims management, and first notice of loss with AI, and underwriting and AI, and underwriting forms that I can't touch on so much.

But I would tell you the big carriers that might feel slow-moving at times, they do a lot of work in the background that you cannot see, especially as it relates to the marketing and distribution engine, that is really, really impressive.

And so I think sometimes the most impressive part is they don't have to shout it from the rooftops, and they still make enormous gains year over year, which is kind of testament to that.

40:36 Adam

OK, what are typical lead fees as a percent of premium that carriers will pay Zebra, and what has been the recent trend?

40:50 David

I won't speak to the exacts here, but what I will say is you can kind of follow this trend very logically over time.

When loss ratios are better, there's more room in the combined ratio to go out and pay more, especially when they feel good about how the loss ratio will trend over the next six months.

And so I'd say right now they're aggressive.

And so when Adam asks about, what is the growth posture of the carriers? A lot of it boils down to the basis of this question, which is just like, how much are they willing to spend on a single insurance lead as it relates to how we expect it to convert and turn into premium for them?

I'd say right now, and for the last year, we've been in an aggressive stance as an entire ecosystem.

That was not the case in 2024.

You know, there weren't as many competitors.

It was easier to eke out a win.

Much more difficult now.

Everybody's in the market, everyone's competing, everyone's looking for growth.

And so I'd say it's higher right now than it has been in the past.

And that's why we say growth posture has been pretty aggressive recently.

42:02 Adam

And, sorry, question for Max, EverQuote, QuinStreet.

What happens to them longer term in the context of, you know, AI and greater use of agent chatbots?

Is it positive, negative?

I'm sure they're both.

But yeah, just some thoughts there.

42:23 David

It's a really interesting question.

I think that a lot of this boils down to how much of the American population is going to go use these tools.

I mean, most commerce is still done, like in our industry, via Google today. Google AdWords still dominates in terms of what drives traffic in this industry.

And all these folks that you've named here do a really wonderful job of helping to cultivate that traffic and send it to the carriers.

Importantly, though, some of these also serve as the underpinning and the rails for driving some of the rate quotes in.

And so they have a lot of really valuable rails in the ecosystem.

I think it'll take a long time for them to become, you know, not important to what is happening.

It would take a lot of adoption of AI tools to move these guys out of the driver's seat in terms of being an important part of facilitating the ecosystem.

But I'd also say to keep an eye on, they're constantly innovating from what we can tell.

You know, they are looking at rated marketplaces. They are looking at how they act as, you know, the Rate 1 rails for a lot of the carriers.

All these things will continue to matter so long as there are intermediaries and distributors in the ecosystem and people need traffic driven to their site.

They'll continue to be important.

But I think that the ultimate answer to this question is, like, if everybody has an AI agent in America, will it dampen EverQuote some?

Absolutely.

Just because they're not facilitating the ultimate transaction and an AI agent has no interest in clicking on an ad.

And so the closer you can get to the bind, I think the more valuable you will be over the longer term.

44:13 Adam

And a follow-up, and if not appropriate, don't answer.

But of these three, which do you think has longer-standing power?

44:22 David

That's a good question.

I feel wholly unqualified to answer that question, but I can just, you know, guess wildly if you want.

But no, I probably won't take a stab at that one.

But I would say, if you know who has more proximity to the bind, you know who has the means to consummate a transaction, to produce the rate, to make themselves relevant not just to a consumer who's searching, but to the agent internet, not the consumer internet.

If you are relevant to the agent internet and you don't just produce ads, but you can also produce a transaction, I think the greater staying power you will have.

44:59 Adam

OK.

So, following that chain, how are you changing use of AdWords today, or do you think you'll change that going forward?

45:08 David

We're going to follow the money, to tell you the truth.

Like we said before, we are going to build completely for an agent future pretty quickly.

We are a really great size because we have tons of carrier partners, but we can still move really nimbly.

And so our opinion is you can't just shout at the floodwaters.

You kind of have to build a canal and direct it to where you want it.

And so we will build all the Muse connectors and then Instinct connectors that we can, and we will have a place to direct that traffic.

But in the meantime, Google is still where most Americans are transacting today.

As long as there is activity there and as long as it is cost-efficient, we will continue to do so.

I think that what you will see over time, should AI agents become more prevalent in this process, what that will naturally do is reduce the real estate that you can go and purchase, the number of leads that you can purchase out of a general search engine marketing ecosystem.

And as a result, it's going to get relatively more expensive.

That is just supply and demand at its heart.

And so there will be a point where it no longer feasibly makes sense for us to go and purchase that traffic.

I don't think that's anytime soon.

We still spend a ton of money on Google.

We still spend a ton of time on SEO and AEO, and we think that'll still have some legs for a while.

And so I think it'll be less of a hard-switch-over at any point in time.

I think it'll be a gradual shift.

I think it'll be all about the preferences of Americans.

And it is our job as a distributor to go and meet them where they want to transact.

And so we're going to keep all the lanes open.

46:40 Adam

And final question, as far as market competition, auto, our sense is that, you know, in the last six months the market's getting increasingly competitive, there's more rate decreases in the last few months.

Have you seen that just continuing, the market getting more and more competitive, more rate decreases, or are you seeing more stabilization?

47:04 David

I would say, in general, especially among Tier Two carriers, we continue to see more targeted rate decreases.

I think that, especially heading into 2027, you will see some Tier Two carriers try to drop rate to drive growth in some targeted spots.

We're still seeing some rate drops here and there.

It's much more stabilized than it was.

Things are very competitive, though.

I would say people are fighting hard for share right now.

I think that's why the Tier Two carriers are, with the lack of a brand, trying to find a way into the market and compete on price a little bit more.

And I think they're having to get more aggressive at this point if they want to buy into the conversation.

47:45 Adam

One more on that. Is the competition on price, and that's typical?

In the last couple years, deductibles went up, up, up.

Is there a move back down on deductibles, or are the insurers hanging on to the higher level of deductibles even as they lower price, to the extent you see that?

48:03 David

I think the deductibles will stay high.

If you just look at what premium is as a percentage of the average American's income.

That's something that we track at TheZebra really, really closely.

We call it TheZebra Premium Pressure Index.

It's essentially, how affordable is insurance?

And then we also track, you know, how many people are uninsured or underinsured in America.

All these things keep going up.

You know, rate inflation was really tough for a couple years.

And even with some targeted rate reductions, it remains pretty high.

I think that, putting on my general economist hat, you know, inflation elsewhere in the market leads to pressure everywhere else.

And so I think that people are looking to scrape dollars where they can.

And so I think for the foreseeable future, you continue to see deductibles stay higher.

And I think, Adam, as you have pointed out in your reports, that's leading to less frequency right now.

And so that lessening of the frequency probably sticks for a little bit longer, as long as the relative inflation pressures of the market persist and as long as premium as a percentage of income persists at the level it's at right now.

49:18 Adam

Well, David, I'd like to thank you, and I thought that was a great, great view.

And thank everyone for joining.

And you'll be hearing more from Zebra.

They're definitely the company to watch.

So thanks, guys.