Insurtech Heat -- Prudential pays $2B for 3yr old start-up and Tesla insures own smart cars; plus 14 short takes on top developments
|Sep 9, 2019|| 3|
Hi Fintech futurists --
In the long take this week, I analyze the $3.5 billion acquisition (if including incentives) of Assurance by Prudential, which took many industry observers by surprise given the firm raised no venture capital backing. Combined with other insurtech symptoms, like 3D renderings of home damages composed of selfies, 500 million simulated driver situations for auto claims, and Tesla's entry into the business of insuring its own smart cars, we have a new emerging mental model for risk-taking in the space.
The latest short takes on the Fintech bundles, Crypto and Blockchain, Artificial Intelligence, and Augmented and Virtual Reality are below. Thanks for reading and let me know your thoughts by email or in the comments! Last but not least, these opinions are personal (or maybe made by a robot) and do not reflect any views of ConsenSys or other parties.
Frontier technology is the bacon bits in the Finance salad. It is the most fun to talk about what *could* be, sci-fi style, because what is already there isn’t particularly exciting (depending of course on who you are). So when we think about futurist themes and emerging tech, the picture is inadvertently extreme, with utopian and dystopian highlights. What if we had maximum data collection, and its associated impact on privacy? What if we had maximum usage of Bitcoin and crypto assets, and their associated impact on governments and banking? What if we all worked in virtual reality, and how would that change what we buy and how we pay?
But as an operator or entrepreneur, you don’t have to take the most risk. Rather, you should do what every smart financial advisor will tell you – figure out your risk tolerance and your ability to execute against it, and then find the opportunity with the best risk/return profile. Optimize the Sharpe ratio for your innovation activity! Just as there is an efficient frontier for investment vehicles so there is an efficient frontier for taking technology and business model risk.
This week, we’ve got several examples of compelling value creation opportunities across different tech risk profiles in the insurance industry. The first is a smart repackaging at the consumer layer, automating demand aggregation by removing friction in buying insurance products. Shouldn’t we be able to just get all our insurance on the web as quickly as we buy toilet paper, movies, and sandwiches? Artificial intelligence and big data are involved, but they are supporting characters in the broader story. Let’s take a look at Assurance in the US, WeFox in Europe, and Tencent in China, as representatives of many others.
Assurance came on my radar courtesy of Financial Technology Partners, which was the investment banker on Assurance's $3.5 billion sale to Prudential. Notably, the company is just 3 years old -- which comes out to a cool billion of enterprise value per year, likely a record comparable to the very few Ant Financials. Depending on the details, this is about $25 million of value per employee. So what does the company do? Simple, really. It is a destination website licensed to sell all types of insurance product (e.g., life, health, auto), with a clean onboarding questionnaire like any other roboadvisor, which then matches against policies on offer from third parties. AI and data science are used as the recommendation engine. It is a Kayak or Money Supermarket of insurance, simply designed, cleverly wired, with killer founders.
But even in that formulation, the thing is a head scratcher. I would expect for comps to be somewhere closer to $250 millon for such a business. Are they running at over $200 million in revenue? Unlikely, but they probably do have some compelling marginal economics. Take a look for example at the acquisition of QuoteWizard by LendingTree, the original Fintech lead aggregator. They paid $300 million on $75 million in revenue, in 2018. Another similar company is the German WeFox, which acts as an aggregator of various insurance policies and is mobile friendly. Below I include the FTP profile of Assurance, and the screenshots of Assurance and WeFox.
If you take a look at the traffic below, WeFox has about 150,000 monthly uniques, compared to 700,000 for Assurance -- about 5x larger. That is a big difference! WeFox had raised $125 million in venture capital earlier this year, so let's assume that is an $700 million valuation. Multiply that by 5x, and you get to exactly $3.5 billion. Looks like we unearthed the financial logic of the transaction. But this is silly. If we look at LendingTree, we see a $4 billion public market cap and ... 8 million in monthly uniques, which is 10x better. Of course you can make the argument that Assurance and WeFox are more closely tied to human agents powering a hybrid insurance model, and that they are a newer generation of technology.
I find the high direct traffic to the Assurance site very curious, whereas both WeFox and LendingTree are primarily search engine optimizers. Bringing us back to the core narrative, the main observation from this spectacular transaction is that innovation around customer acquisition -- whatever is hidden behind these numbers -- can be done while retaining traditional product and industry structure. If AI and data science are laser focused on making the Direct channel work better, maybe you get to walk away with a billion or three. That is certainly what Tencent is doing with selling insurance through its social chat channels.
The second example, which is at a higher level of technology risk, is applying artificial intelligence and augmented reality to the underwriting and claims management processes. This is a manufacturing story for the financial product itself. Examples here include PLNR for modelling damages from phone photos on a 3D model, and Greater Than’s car driving data for underwriting auto insurance.
The former, illustrated above, is one step above taking a selfie of car accident damage and sending it to the insurer. Instead, you take multiple pictures of some damage in your home, as shown by the camera positions in the rendering, and the software constructs what looks like a Mixed Reality version of the house. The composite is analyzed by a human being, who manually measures the size and severity of the damage. Thus the actual human judgment sits with a "micro-worker", while the technology essentially teleports that worker to the needed location. This type of innovation changes how the sausage is made, but it also requires for the behavior change will be adopted.
The latter, illustrated below, is a novel way to think about insuring auto accidents. Instead of looking at the population broadly, the model underlying this software analyzes each driving trip in real time against 500 million driving situations modeled out using AI. I assume (but don't know) this is sort of like Simudyne, where you can simulate out various behavioral agents and run massive environments that generate data from a single real data point. The driver can see how their behavior actually impacts insurance premia -- whether or not you get into an accident, being aggressive may lead to a tangible quantitative penalty. Note that such a change in underwriting approach could logically become the default for any smart car.
As a final point, you can increase the risk one more time by introducing a business model pivot, which is what we see with someone like Tesla starting to offer their own insurance. While the last example suggested using frontier technology within an underwriting software solution from an insurance technology provider (similar to banking-as-a-service), the case of Tesla offering insurance on their own cars is slightly orthogonal. Tesla is the store for the car, and the financial product is there only as a side-feature, or a mere additional monetization engine. Alternately, it's like an air conditioner store making money on warranties.
I am also starting to analogize financial products sold by non-financial destinations, which are often attention platforms like Apple or Amazon, to generic brands in a supermarket. Yes, the Bank of America bank accounts might have more associated “trust”. But that is like saying that Colgate is a better toothpaste than just any other toothpaste because it is branded. The reality is that generics are a close substitute for price sensitive consumers, and in the Retail and Pharma sectors generics are always biting at the heels of established brands. As financial APIs get better, allowing for any company to easily spin up bank, investment, and insurance accounts, competitive moats start to erode and financial generics become much more common place. This hurts a brand’s ability to extract economic rents, but helps the distributor grow its footprint and the consumer retain some of the surplus. Value to the people, my people!
Featured Interviews, Podcasts, and Conferences
How should fund managers use AI to streamline operations? I was interviewed for this article for the Hedge Fund Law Review, which provides a deep and thoughtful discussion.
See you at the DeFi Summit, with my keynote discussing the macro Fintech and DeFi trends, on September 10th, London.
Check out my Keynote at the RoboInvesting Summit, focused on the rebundling of digital wealth, banking, lending and payments, September 11th, London.
Very excited for my first ConsenSys event, introducing a major announcement, in Tel Aviv, September 15th.
SIBOS, speaking on the Discovery stage about tokenization and the evolution of finance, on the 23-26th of September in London.
Lendit, speaking on the potential of payments cross with Augmented Reality on the 26th of September in London.
#WinnersOfWealthTech Ep 24: Rich Cancro, CEO of AdvisorEngine. Shout out to Rich Cancro, with whom I co-founded AdvisorEngine. In this podcast, he walks through early experience with online brokerages in the 1990s, what went right and wrong with Wall Street in the 2000s, and what it is like to start a Fintech company in the 2010s. This advice and journey are generalizable to senior people in the financial industry considering the start-up route.
Charting Monzo's journey from coral millennial accessory to 'grown up' bank. If you don't get neobanks, this is a helpful read. Two million people use Monzo, 35 thousand new customers signing up per week. And yet, few people use it as a primary bank account with direct deposit, leading to smaller balances and harder economics.
Beware of hidden fund management businesses from Big Tech companies. I knew Apple had lots of cash, but I didn't know that it had an internal hedge fund that helped it grow $100 billion into $240 billion over the last 7 years. What is striking to me in those numbers is that Apple could buy ailing Deutsche Bank, with its customers, branches, revenue, and apps, 20 times over. And it doesn't.
A firm owned by billionaire hedge fund manager Alan Howard plans to launch a $1B crypto venture. The billion in question would be dedicated to a crypto fund of funds, which is pretty contrarian market timing. While it is always good to see more capital entering the space, I still get more excited about operating progress.
What You Should Know Before Putting Half a Million DAI in Compound. One of the quickest growing lending protocols in town, Compound is increasingly on the radar of the crypto industry. This article does a great job of exploring the risk of a "bank run" on the system. If I read things correctly, nearly 95% of the overall committed capital is being lent out. In bank land, that's like ... pretty high leverage. Where is that regulatory crypto capital?
Ethereum DApps will be available on Telegram’s new blockchain project. Telegram made big promises that led to it raising over $1 billion at the top of the market. The due date to deliver has come -- supposedly launching in October -- and initial focus is on interoperability with Ethereum. Telegram is no Facebook, but it still has meaningful footprint and mindshare. I see this primarily as a compliment to Ethereum, rather than a threat.
One of the Oldest Quants Is Going All-In With Robots. It used to be that you hire data scientists to write mathematics to create algorithms to beat the markets. Then you hired computer scientists to write robots that created algorithms to beat the market. Now you get AI architects to create robots that write robots that try to compete with other AI robots on arbitraging robots that make indexes. So totally normal finance stuff, $7.5 billion of it. Here's another side of the same coin: The Commoditization of Information, written by my friend Geoff Yamane.
Amazon Trials Hand Recognition Payment Method At Whole Foods. We already see facial recognition as a method of payment that replaces QR codes in China. These trials supposedly use visual imagery based off your hand (not just prints, but the volume of it) to charge your account. In the long run, things that forge identity out of your body make *natural* sense.
Wells Fargo: Artificial intelligence and machine learning a 'double-edged sword'. In other news, sky is blue and sun is bright. One interesting tid-bit is that for a giant like Wells, the use a multi-cloud approach (e.g., AWS, Microsoft) in order to get benefits from each of the vendors in their area of best practice. Here's another in the same vein: ASX approaching artificial intelligence with caution.
Some numbers: Cognitive artificial intelligence is expected to be a big business as research firm IDC forecast that firms will spend approximately $77.6 billion on the technology by 2022, and Juniper Estimates 3.25 Billion Voice Assistants Are in Use Today, Google Has About 30% of Them.
Snapchat Adds New Landmarker Targets & Templates to Lens Studio. Creators can tag geographic landmarks with 3D rendered objects and animations. When will the first financial advisor or insurance sales person tag a competitor's store front with a web app that launches their website? This could be like buying Google AdWords against your competitor names.
Airbus Previews Military Sandbox App for HoloLens. Check out the visualization. I can see a similar tool being developed for visualizing a financial life path, for example.
Police team up with Avast to make an 850,000 device botnet self-destruct. It is always fun to read about a futuristic showdown between a web of old computers networked together against their will, assailing some Minecraft server. In this case, however, they were hard at work mining cryptocurrency. Alas, the Feds got 'em with some counter-hacking.
Looking for more?