What Everyone Is Getting Wrong About Brokers and AI

OK, I finally made some time for my long delayed take on what AI means for insurance brokers. The short story is I think the consensus has it totally backwards.

The perception is small commercial and personal lines are most at risk of disruption while the larger public brokers are fairly insulated. People are focused on the areas that are easiest to disrupt rather than those that offer the most reward for disruption.

Thus, I view the risk as far larger for Aon, Marsh, and peers than it is for small account commercial or personal lines. I’ll start by addressing why the smaller ticket business is of less concern to me before moving on to the core issues in large account.

Direct Disruption

It seems a lot of focus is on the potential for more business to go direct by leveraging AI. The thesis is this only impacts transactional business like auto and BOP and thus the selloff in the major brokers is unwarranted.

Will there be AI advances that make it easier for end users to go direct? Sure. But will they gain material share? Unclear.

There have been direct “marketplaces” for personal and small commercial for a long time. While there have been some modest successes, none are super meaningful. Why is a “new and improved” version of “we’ll pick the cheapest price from 20 markets” going to work better this time around?

The truth is direct has been most successful when it is attached to a brand, e.g. Progressive. People want to buy direct from a single carrier they trust more than they want a machine to tell them which random carrier they barely know is best for them.

One reason small commercial hasn’t made the same progress in direct as personal lines is most direct offerings are “central markets” with multiple carriers. This is not what consumers want (even if it may be better for them).

The other obstacle in small commercial is business owners want a trusted point of contact when there is a problem. Even if AI can guarantee you the lowest price or the most robust coverage terms, a business owner is going to want to talk to a real person who they have some relationship with when there is a claim or other problem with a carrier and they need someone to blame.

Someone who has spent their whole life building a company isn’t going to want to deal with a chatbot when they have a cyber attack and aren’t getting good answers from the carrier about what is and isn’t covered.

So I am very skeptical that we will see much in the way of customer facing automation in small commercial. There will be exceptions I’m sure, but, broadly, I suspect the most likely path to direct small commercial is going to come from a carrier with an existing brand, similar to how Progressive and GEICO leveraged their brand to go direct online.

Oh, and as for the idea that some startup will build this all themselves, good luck with that! Creating a de novo small commercial brand requires a balance sheet. You can not do that as an MGA. We will see MGAs with niche focuses on certain business segments, but that is all.

So, I suspect most developments in personal lines and small commercial will be back office focused to let carriers and agents serve customers more efficiently. That improves their value proposition rather than threatening it.

Large Brokers

Before I get into the “how” large brokers gets disrupted, let’s talk about the “why”.

Margins are too fat.

In the early 2000s, the public brokers had mid teens margins. Now, they are in the mid thirties. Please someone explain to me why this makes sense?

If I am a buyer of insurance, this means about half of the profit stream from my premium is going to the broker, not the carrier ($100 GPW at 12% commission and 33% margin = 4 points which is just as much as the carrier makes if it is writing at a 96 CR).

I can’t think of any other business where distribution reliably keeps half of the downstream profits long term. That doesn’t make economic sense. Therefore, I don’t believe it is sustainable.

Once customers realize this (and it’s shocking mid-large account buyers don’t know already), they will demand lower commissions. How much lower? 10% is an easy first cut (If expenses stay at 65 and revenues drop to 90, that’s still an impressive 28% margin).

If an aggressive Uber-like disruptor comes along and is willing to lose money for long periods to build a franchise? In that case, commissions can easily drop by a third which would leave incumbent brokers at breakeven (and thus unable to service their debt).

Before some of you scream that is unrealistic and would never happen, it’s not hard to do the math. Again, let’s start with $100 of commission and $65 of expense for today’s $35 margin. What happens if a newco can cut costs 20%?

In that case, expenses drop to $52 so if revenue dropped to $67, you still have $15 of EBITDA which is a 22% margin. That’s acceptable.

So, the math maths. The question now is whether an AI-centric newco could service mid and large accounts 20% cheaper than the incumbents. I don’t have the answer to that yet, but I don’t see why it can’t happen. Let me spitball some possible ways it can happen.

Vulnerabilities

There are potential risks throughout the portfolio from mid market to Fortune 500 to reinsurance brokerage to wholesale and even possibly MGAs. Let’s explore each of them.

Reinsurance Brokerage

Why does reinsurance brokerage have such fat margins? The capacity is syndicated so you don’t need help selecting a carrier and the buyers are large and sophisticated so should be capable of figuring out their needs on their own. This doesn’t sound like a market that needs a powerful intermediary.

Now, I know people will jump up and shout about all the value the brokers bring with modeling and analytics and negotiating pricing and so forth. But you know what? Those are all things AI can potentially do.

What stops the next Karen Clark from building a cat consulting practice that does all the analytical functions a reinsurance broker does today and selling it directly to insurers?

The role of the broker would shrink to lining up carriers to participate at the price and structure the outside consultant recommends.

Arguably, one of the brokers could buy said consultant to maintain their role in the process but the new service would be priced well below what Carp or Aon charges today.

Wholesale Brokerage

Similar story – why are wholesale margins so high? There is the myth that it is so much harder to place something wholesale than retail. Sure, there is some truth to that, but it is far less true than it was 20 or 30 years ago.

And there’s a very simple way I can prove that. Look how much higher wholesale brokerage margins are vs. 20 years ago!

Btw, you know what’s not up in that time? Carrier underwriting margins. Combined ratios during the “E&S Golden Age” are fairly similar to what they were last century. The benefit of the growth has disproportionately fallen to distribution.

Thus, there is a golden opportunity for a new entrant to develop a product that makes it easier for retail agents to access wholesale capacity. This could be as simple as a fee per submission which, in practice, would act like a commission, but it would be a fixed $ amount likely equal to less than 1% of premium.

In theory, wholesale carriers would love this platform. They would likely have less pricing pressure from an independent platform than the arm twisting that goes on where carriers have to take lesser business as “tribute” for being able to quote desirable placements.

In reality, it would be interesting to see if carriers are too chicken to sign up for a platform like this in its infancy due to fear of retribution from their main source of premium. But if that hurdle can be overcome, providing more transparency for retail customers with wholesale needs is in the client’s best interest and should happen, though once again, it will flatten margins along with quoting times.

Fortune 500

I think people are far too dismissive of the risk to large corporate accounts. In many ways, I think these have the most disruption risk.

Why? Because a new AI offering can go straight to the risk manager and bypass Aon and Marsh. The newco only needs to convince one person to win a big account. Sure, the legacy brokers will still be needed for executing placements and for claims handling but all the analytics can be stripped out and done by a challenger.

If I’m a sophisticated RM and I can share my data with an AI-centric vendor who can do all the analysis Aon does for me but at half the cost, why am I not taking that deal?

The savings are more germane to a large corporation that is clinically bottom line focused than to a small business where insurance is a necessary evil. If a RM can show the CFO he or she can cut brokerage costs materially without sacrificing any coverage, that’s going to lead to a bigger bonus. There is thus more incentive for RMs to try outside solutions than at smaller companies.

Now, some of you will say, but wait, Aon and Marsh can develop that capability in house (or buy the vendor who develops it). Yes, they probably can, but they’re going to have to charge less for it than they do today.

Middle Market

I think mid market is less at risk than the other areas above because it is going to be harder for a newco to reach enough companies at once with a broad offering, similar to with small commercial.

However, like with large corporates, you only have to convince one person here. It’s the CFO instead of the RM and insurance isn’t as big a priority for them and they may be more hesitant to share data, but there will be some interest, especially for those where insurance is a bigger part of their expense structure (real estate, trucking, etc.).

While I don’t believe we will get the much desired “super app” where the agent can enter data once and get all the quotes they want, I do suspect AI will make that repeated data entry easier (for both small and mid market).

That will also likely bring an “auction” style quoting system closer to reality which, in theory allows for bypassing the agent if a newco can get appointments for its “commercial rater”.

Like with some of the other advances, time will tell whether these features originate inside or outside the industry and, if the latter, whether they gets acquired by one of the large brokers to eliminate the threat.

MGAs

This one might have to be a whole separate article later cause I know it will generate strong reaction so I’ll give the short version for now. AI could kill the MGA golden goose that generates so much commission for brokers.

No, I’m not saying MGAs will go away. I do however see a path where new MGAs go “direct” to the carriers.

Imagine I’m a carrier who provides a lot of capacity to MGAs. Why wouldn’t I ask Claude to research emerging businesses with unique insurance risks?

Before one of my underwriters can quit and form a MGA that then asks me to provide the paper while they get rich because they’re now “worth” 20X “EBITDA”, I can find this niche and then ask Gemini for a marketing plan to place ads direct to these small business owners on Google.

Now, I’ve identified the customers, I’ve built the product for them and I write them through my captive “MGA subsidiary”. If I don’t want to keep 100% of the risk, I lay the rest off on the same reinsurers who would have backed the MGA if one of my employees created it.

I’ve now cut the brokers out of the loop and retained my underwriting talent. That’s a pretty huge win.

Probabilities

I have no idea what the probability is of any of the above. I do know it’s not zero which is why the stocks of the public brokers are down. Markets discount possible future events and the scenario where the future is worse has grown in probability.

One other thing to consider when handicapping probabilities is the effort required of the disrupter. While it’s true that the “TAM” of a challenger model to local agencies is huge, the effort to achieve it is significantly greater than taking on reinsurance brokerage.

If you develop a solution for the latter, you only have to convince a few dozen buyers (and really far less, because once you convince say five prominent ones, the rest will follow). The effort to convince thousands of agencies to adopt your technology is expensive and time consuming. The effort to convince millions of businesses to dump their local agent for you is Herculean.

So, if you’re handicapping the odds of disruption, the most likely success isn’t the one with the biggest TAM or the least operational complexity. Rather, it’s the one where there is a highly successful niche where clients pay a lot for information beyond the actual placement of the risk.

It just so happens those are the ones with the fattest margins which makes it that much easier for a newco to extract favorable rents.

It’s Not All Bad News

You’ll note nowhere did I suggest brokers are going away. Rather, the innovations they have brought in recent decades that expanded margins will be targeted. Maybe the brokers will develop the solutions in-house first and forestall the challengers. That is certainly possible, but it is no guarantee.

The core bread and butter of binding, servicing, claims handling, etc. will likely remain. In other words, agents will look a lot more like they did last century. That isn’t a bad thing. It’s just less profitable.