When Lemonade (LMND) released its S-1 filing, I saw many naysayers pointing to small revenue scale for an IPO company, and on the other end people rooting on the company for its new approach to insurance. I was skeptical, and frankly, knew nothing about the space. Just looking at the financials at face value I tended to side with the former opinion, but instead I decided to dig a little bit deeper and really understand the business from all angles. Here’s what I found:
Mission and Vision
A good start! Let’s see how they do that.
Value Prop/Business Overview
At the moment, the company primarily sells homeowner’s and renter’s insurance. In their words:
Lemonade is rebuilding insurance from the ground up on a digital substrate and an innovative business model. By leveraging technology, data, artificial intelligence, contemporary design, and behavioral economics, we believe we are making insurance more delightful, more affordable, more precise, and more socially impactful. To that end, we have built a vertically-integrated company with wholly-owned insurance carriers in the United States and Europe, and the full technology stack to power them.
A two minute chat with our bot, AI Maya, is all it takes to get covered with renters or homeowners insurance, and we expect to offer a similar experience for other insurance products over time. Claims are filed by chatting to another bot, AI Jim, who pays claims in as little as three seconds. This breezy experience belies the extraordinary technology that enables it: a state-of-the-art platform that spans marketing to underwriting, customer care to claims processing, finance to regulation.
The primary value prop here is simple - ease of use and digitization. When I first heard about Lemonade it used the keywords “P2P insurance” to describe itself, so it seems the messaging of the company’s primary value prop has changed. Whereas P2P insurance describes a more decentralized model where peer groups directly insure one another to limit risk, Lemonade collects premiums and pays out claims like other insurance companies (though with some business model differences, to be described later).
This shift in messaging hearkens back to some of the large lending fintechs that grew over the course of the last decade, such as LendingClub. The original innovation was dubbed “P2P lending”, essentially allowing retail investors to lend to borrowers, the primary reason being that the banks were less willing to lend during the financial crisis. However, as LendingClub evolved, its model really transformed where it relied less on retail investors, but rather partnered with financial institutions to provide capital. LendingClub chose this direction in order to fuel growth; in 2014, only 15% of LendingClub’s loans were managed by individual retail investors. At that point, the primary innovation became digitizing and automating the lending industry, where at the time much of the space was still manual and in-person. While digitization is positive, LendingClub de-prioritized what I would consider to be a much stronger moat; marketplaces are very tough to build and scale whereas many traditional lenders had begun to digitize with some success, creating much more competition for LendingClub.
Similar to LendingClub, Lemonade has also had to rely on other financial institutions to bolster its growth. While Lemonade offers insurance policies and manages claims, a large percentage of its book (~60% in 2019, 75% is 2020) is shared/sold to other insurance companies through reinsurance, partially making it more of a front-end channel for reinsurance co’s (such as Berkshire Hathaway and Lloyd’s of London), limiting its risk but also revenue/profits.
While I acknowledge that the two companies’ business models and journeys are pretty different by virtue of operating in different industries, I use this analogy to illustrate where my gut tells me Lemonade is going as a company and stock based on its current business model and what I perceive to be a relatively low competitive moat. (Hint hint LendingClub’s stock hasn’t done so well since launch)
Business Model
As opposed to other insurance companies, who get paid by collecting premiums and largely trying to deny claims as much as possible (i.e. attempting to avoid claimant payout), Lemonade has decided to forgo the delta between premiums and payouts and instead takes a flat 25% cut of gross written premiums (GWP) as revenue, with the rest of the premium either sold off as reinsurance or given to the claimant’s chosen non-profit as a donation. According to the company’s S-1, 75%+ of their premiums are covered under reinsurance, meaning they have sold off the claim risk to another insurance company, as well as the premiums, in exchange for a 25% commission to cover acquisition and administrative costs. They’ve done this for a few reasons,
They generally lack the capital/reserves to pay out a ton of claims in the case of a disaster (i.e. to manage risk), and
To stabilize margins and payout rate (though they lose a large chunk of upside)
By capping the revenue on each premium, Lemonade claims to have no incentive to deny a claim, and policyholders are also disincentivized from embellishing/lying about claims, as fraud may take money away from the pool set aside for their chosen non-profit. Both arguments make some intuitive sense, but are challenged in the context of “quick, easy payouts”. These reinsurers are still trying to make money on the premium/payout delta, so if they feel Lemonade makes it too easy to have a claim approved, that relationship might be challenged. In addition, though the company believes earmarking some funds for charity will prevent fraud, offering an easy payout invites additional fraud.
All previous points aside, I’d say the business model is creative and has a fairly strong marketing angle from the policyholder perspective. It seems to be a great way to build a trusted brand and definitely exhibits a more customer-centric culture than other approaches; however, whether it’s a profitable or sustainable model is an entirely different question.
The Market
Property and casualty insurance premiums amount to approximately $1.6 trillion globally (McKinsey). There’s a lot of market value up for grabs and the market is incredibly fragmented, with no real winner in the space. The top 5 players of the US P&C Insurance market only make up 31.6% market share, with State Farm (the leader), not even making up 10% of the market. This suggests there is room for a strong brand to differentiate itself here, though also means there are lots of players to compete against.
Go-To-Market Strategy
Based on this description, Lemonade sounds like a DNVB for insurance - the company prioritizes convenience as well as strong aesthetics and marketing. Not knocking this strategy - as an analogy, many direct to consumer (D2C) brands such as Casper, Warby Parker, and Glossier have been able to outperform traditional retail for exactly these reasons.
Approximately 70% of their customers are under 35, which is an attractive demographic with respect to building a long term trusted relationship with the customer. The company states:
Our renters insurance customers steadily increase spending on their renters insurance policy with us over time. Customers who bought Lemonade renters insurance three years ago spend 56% more on their renters insurance now than when they first joined. This germination can be observed across our book of business, spurred by the accumulation of wealth people typically enjoy during their working years. According to the Federal Reserve, the median net worth of an adult American under the age of 35 is about $11,000. By their 40s and 50s, their net worth has grown by 10 to 15 times, and that growth peaks at 25 times after the age of 75.
This is also somewhat illustrated by their customer “graduation” rate (i.e. percent of condo/homeowner’s policy holders who were originally renter customers and upgraded). Renters pay on avg $150 per year while homeowners pay $900, though this graph doesn’t tell me what percentage of renters graduate (which would be a much more useful metric).
Despite my generally positive views on the company’s go-to-market strategy, it is not yet enough to convince me that the company has a long-term competitive advantage.
Team/Operations
The company touts a pretty sophisticated front and back-end tech stack, as shown below. They claim that all of their operations are incredibly data and AI-driven, and personalized to the customer.
AI Maya: Our playful onboarding and customer experience bot sells our policies and handles everything from collecting information and personalizing coverage to creating quotes and facilitating payments.
So I wouldn’t really call Maya a new and innovative AI - I’ve seen many off-the-shelf chatbots that have similar functionality. She isn’t truly conversational - she asks only basic questions and has a pre-set list of potential questions I may ask her. Anything more sophisticated like “What is a deductible”, which I would argue isn’t a super-sophisticated question for an insurance co, can’t be answered by Maya.
AI Jim: Our claims bot handles the "first notice of loss" for 96% of claims as of March 31, 2020, and in approximately a third of cases can manage the entire claim through resolution without any human involvement.
AI Jim is definitely more interesting - whereas today many insurance companies require you to file a claim and wait weeks to receive a decision (as most claims are processed by humans), AI Jim creates a much more efficient and customer-centric claims experience. Again, I’m not sure how much real AI is really here, given Jim is only capable of fully resolving 1/3 of claims, but I will admit that the experience is far beyond what I’ve seen when filing claims with other companies.
CX.AI: Our bot platform built to understand and instantly resolve customer requests without human intervention. About a third of of all customer inquiries are handled this way. Customers often require assistance pre- or post-purchase, ranging from coverage questions to performing changes to their policy, such as adding a spouse, updating coverage amounts, changing payment methods, and adding newly purchased items.
Lemonade runs its operations with automation in mind, both on the front and back-end. While I imagine there are many insurance organizations now working to create efficiency by incorporating machine learning into their decisioning process, they seem to be behind and have not yet fully exhibited this technology or automation in the front-end experience. In addition, I imagine the decision engine, the Forensic Graph, that AI Jim and CX.AI run on will become more sophisticated over time and further lessen the load on human agents. As a result of the work the company has done to automate the process, their NPS score is first-class. In addition, it can now handle about 2500 claims per employee, where the largest player can handle about half that.
On the back-end, the company uses Blender to fully manage the insurance process, from underwriting and claims to marketing, as well as Cooper to automate repetitive tasks for employees such as processing paper checks and collating regulatory filings.
All of these technologies combine to create a much more efficiently run organization, and I truly commend the founders and leaders, Daniel Schreiber (former mobile and media tech executive at Sandisk, Alechemedia) and Shai Wininger (formerly co-founder and Chief Product Officer at Fiverr) for the machine they’ve built.
Financial Traction
The numbers look to be trending in the right direction in all 3 charts, but increases in written premiums don’t necessarily tell us if the company is in good financial health. Per the second chart, the company still loses money on each written policy, which is very obviously not a good signal of profitability. The third chart seems to paint a rosier picture, as the company has finally been able to bring down its gross loss ratio below 100% (gross losses are smaller than premiums collection), though the company is still far away from industry average loss ratios of 40-60%. What these 3 charts tell me is that the company has grown incredibly fast, unprofitably, and wants us to believe that they are getting closer to profitability.
The company’s largest spend item is marketing, and they claim they receive $2 of gross premium for every $1 of marketing spend. However, given the majority of their premiums are sold to reinsurers, my math tells me that they keep at most 25% (flat fee) +(75%*25%) (commission from reinsurer) = 44% of the $2 as revenue. If my math is right, they’re already earning less in revenue than spent in marketing with this take rate, which is exemplified by the company’s financials ($89.1m spent in S&M and $63.8m in net earned premium in 2019).
For this company to become profitable, you essentially have the believe that customers who were already acquired will continue to spend more money with LMND and will become more valuable over time, covering the cost of acquiring them. The company’s retention data has shown little evidence that past customer cohorts’ lifetime value has exceeded the cost of acquisition.
Using the same methodology as Nima Wedlake from Thomvest Ventures to calculate LTV/CAC, the number for 2019 was <1x (0.74x). Not a pretty picture.
Based on this info, I’d say the future with respect to LTV is a bit murky. Not hopeless, but murky. And given that it seems to be one of the single most important factors in driving profitability for the company, I believe it presents serious risks going forward.
My Conclusion
While I believe the insurance market is ripe for some change, with Lemonade best positioned to disrupt, I also see structural competitive issues ahead for Lemonade, as well as financial challenges based on the business model. To summarize:
Pros
Large market with no clear leader and room for a trusted, branded player
Incredibly strong marketing and customer-centricity
Strong operational backbone
Cons
Relatively low competitive moat - largest differentiation is marketing
Inefficient and unprofitable customer acquisition
Flat-rate business model and reinsurance severely limits profitability
Ability to meaningfully increase LTV is still unproven
Thanks for reading this far and let me know your thoughts, and please subscribe!