Disrupting Insurance

The Other Financial Services Opportunity

In the last years, most of the focus on innovation in financial services has been, it seems, on banking (investing and lending) and payment. Comparatively, fewer startups have been addressing the insurance space. This is about to change and the opportunity space is amazing. To give a few numbers, non-life insurance market world-wide in 2011 was estimated at $1,877.2 billions. The global life insurance market had total gross written premiums of $2,464.2 billion the same year.

As banks, insurance companies are a balance sheet driven business, more specifically a float driven business. In the past 20 years, arguably the investment side of the insurance business has seen the most innovation. New financial products have been created (not to the most benefit in some cases — AIG and MBS for example), access to markets have become more global and cheaper.

However on the distribution and actuarial front not much disruption has happened, as highlighted by IBM in its Insurance in 2020 report:

Changes in value chain automation, data management, and the use of online mechanisms made over the course of the last several decades were at the tail end of larger technological or societal changes and were directed towards improving existing processes and mechanisms.

This is about to change, and as most digital driven disruption, it is start with the distribution model. Now, insurance distribution has important variations by country but overall the industry distribution model has been built around agents and brokers, a physical distribution network.


What has started happening in the last years to bank branches is happening and will happen to the insurance agent networks. As more and more people spend their life in a digital world and value convenience there over other factors, digital distribution increases. More than with banks, it creates a complex competition dynamic for insurers as a direct digital distribution is at cross against their agent network. A classic case of Innovator’s dilemma. Some insurance companies have responded by branding their direct offer differently and playing with their pricing strategy to ensure their agents will not too negatively affected but in my view, this strategy will not last long as from a customer behaviour perspective, more business will shift online over the next years.

From a startup perspective, this creates an opportunity space in online brokerage, there is a small step from online distribution to online comparison and this type of void is filled quickly.In the UK, comparison websites such as MoneySupermarket have been quick to jump on the bandwagon (or in Switzerland Comparis). Google has seen an opportunity as well and is promoting its own comparison engine in that space. In the US, websites such as CoverPath offer the same type of service for more complex product like term life, and their online experience is much better than any of the existing competitors.

One of the unanswered question is that space is up to which level of complexity can a customer be self-directed? In an insurance world, there is a conflict between the requirements of the underwriting algorithm for being able to price the premium and the optimal UX for users to be able to obtain a product. A term life insurance product for example, cannot be properly priced with just a few field. Also, what if a startup could have its own data sources for underwriting insurance?

The distribution model is also bound to be changed by players that provide value added services out of which insurance is just a component. Trov (an Anthemis investment) is a good example of this new type of potential distribution platforms for insurers. Trov’s objective is to unlock the value of all your physical assets by making them digital. Protection on stuff you deeply care about (as in insurance) is a phenomenal value proposition but creating liquidity can be equally as important. From a customer perspective, there is more value add in a platform that can cater to all their needs than in an industry specific one. Insurer will need let a little go in the direct client relationship to leverage these customer oriented platforms.

Hybrid Model

Insurance is a data and calculation model, ie amassing enough data on events to calculate and price risk. This is an important barrier to entry as the historical data of insurers is impossible to reproduce (you have to bear the risk to get the data). However, from a technical perspective, the cost of acquiring, storing, managing and calculating on data has become lower and lower. The technical moot of the insurance business is disappearing or even reversing (considering their existing technical architecture is expensive to maintain vs new technical solutions).

Take the example of MetroMile based in Portland. As telematics (for end customers) is a new field for the industry, there is much less historical data advantage. Whoever will amass the most data on instant driving behaviour and use it for actuarial purpose will have a strong advantage.

At the same time, insurance pricing is based on average, which means if you are on the low-end of an average, the economic deal you are getting is pretty bad. A person who drives his car on low mileage is probably paying too much insurance. The solution so far has been to reduce coverage, but the reduced risk of low mileage for the insurer does not mean a reduced risk for the driver in terms of his protection. Telematics devices such as Metromile’s, by tracking actual miles driven, allows for more flexible contracts with lower overall premiums but high coverage.

For a new player, a way to enter this field is to propose first a new insurance distribution models, like insurance by the mile. However, my assumption is that the long-term objective from Metromile is to record enough driving behaviour information to be able to underwrite motor insurance based on driving criteria as well as nudge behaviours to reduce the overall risk of Metromile drivers.

Now Metromile is not an insurance carrier and the actual insurance is provided by National General Insurance Company. It is unclear how much the underwriting model is controlled by Metromile or their partner but I am assuming Metromile had input into it. I am expecting to see (and interested to discuss with) more companies adopting this approach. A combination of online customer acquisition and proprietary data sources / data relationships on top of an efficient balance sheet is an interesting business combination and is less capital intensive than a full new carrier.

A new carrier? Alternative to carriers?

Creating a new insurer is expensive. If we take the example of Oscar, a New York based Health Insurance firm. Their minimal capital requirement for operating was around USD 45M (http://www.dfs.ny.gov/insurance/exam_rpt/x9475o13.pdf). As is the case with most balance sheet driven business, hypergrowth is not an easy possibility as capital requirements keep increase with more customers (ie the better you perform, the more your need to raise capital, from a requirement perspective). Not impossible but it takes a certain type of investors to fund this type of startups, from a capital deployed to return ratio, other businesses may appear more interesting.

Another interesting possibility is in peer to peer insurance. Peer to peer insurance is very much different from peer to peer lending. Peer to Peer lending is about increase your investment returns by cutting on intermediary fees, while managing risk exposure through diversification. Peer to Peer insurance is about reducing your cost of insurance by co-managing your own pool of money and claims. Peer to peer lending is about managing several one time low implication relationships (loan agreements), peer to peer insurance is about managing one high implication relationship with many people. Failure in peer to peer lending is on an individual basis, failure on peer to peer insurance is on global basis.

While still very much far fetched, the algorithmic approach to peer to peer insurance is really interesting. Climate Corporation (previously an Anthemis Investment) has proven that with automated data feeds, claim management can be made into a seamless experience: a weather event would be registered by the nearest station and the claim would be reimbursed automatically. Algorithmic and programmable ownership ledger, such as Ripple and Ethereum set an interesting groundwork for expending this to other types of insurance (say your Iphone sensor register a bad fall and trigger an automatic claim reimbursement). Peercover was an interesting player in that space (but has since pivoted).

Other models?

Last, an interesting number of players have emerged in the “behaviour nudging” space. From wearables à la Up to Sherpaa, several companies are tackling the gap between historical health behaviour and digital life. An interesting space indeed.

Leave a Reply

%d bloggers like this: