Future of Finance


How to get the InsurTech revolution happening again

InsurTechs are more numerous than effective. Technological gimmicks and virtue-signalling corporate philosophies capture headlines but customers, revenues and investment dollars continue to flow to a smaller class of large and better-established firms completing conventional tasks such as broking insurance or automating the back office. The InsurTech Revolution has stalled. Start-ups that dreamed of disrupting Allianz or Axa are now pitching to be bought by the same firms. Blockchain is neutered by its diversion into collective infrastructural projects. Artificial intelligence (AI) is full of promise but marooned between specialist InsurTechs and data lakes so deep and wide insuers do not know where to begin. Which is why the best way to get the InsurTech Revolution under way again is to reorientate the incentives of both the insurers and the insured. Insurers need to stop treating risk as financial opportunity and the insured need to stop thinking of insurance as a rip-off. It is an inversion of the status quo that only a truly ambitious InsurTech could actually pull off. One has yet to disclose itself, but if and when it does appear it will be the greatest InsurTech Unicorn of all.

InsurTech is a capacious term. Close definitions soon become so abstract (“Companies using technology to disrupt the insurance industry”) that they drain the words of meaning. Yet any attempt to portray reality is equally soon lost in variety so rich that generalisation is impossible.

Artificial intelligence is full of promise for insurers

InsurTechs are using dashcams, sensors and tracking devices to monitor policyholders, cutting premiums for the prudent and scolding the incautious, and selling time- and usage-based policies. Artificial intelligence (AI) is sifting big data to assess and pick risks that make fewer claims.

The ubiquitous smartphone (itself packable with sensors to initiate cover and expedite claims) facilitates the sale of insurance on demand, especially in markets without alternative means of distribution. The smartphone can also automate the claims process, chiefly via photographs.

Insurance is increasingly “embedded” in sales of other products and services, from household goods, through banking, to travel arrangements. It is one of the ways in which InsurTechs promise to transform the customer experience of insurance. It certainly needs improvement.

Indeed, InsurTechs are now laundering the poor reputation of the traditional industry with premium cashbacks, donations to charity and top-ups to crowd-funding campaigns. Curated social media groups are even rediscovering the mutual model of insurance, with its profit-sharing ethos. 

Price comparison sites are yielding to robo-advisors and chatbots that promise to talk customers through the purchase process. Others are guiding consumers through the claims process unaided by humans or by a hybrid model where a human is always at hand.

Blockchain claims a special affinity with insurance

Blockchain has visited the insurance industry too. Superficially, blockchain is ideally suited to counter the disparate data sources, the inefficient placement and the archaic claims and billing processes that create so many reconciliation mismatches and settlement delays in the insurance industry.

So it is not surprising that, alongside familiarromises to store and share documents and reduce fraud by transparency, blockchain vendors have volunteered smart contracts to automate the claims process. Less predictably, they have advanced the issuance of tokenised catastrophe bonds.

Most intriguing of all are the emergent breed of insurance “marketplaces,” which enable insurers to (and others) to trade underwriting risks or insurance policies or employee benefits or processing technologies or even data – mostly at controlled prices but sometimes at transparent prices too.

Conventional software and systems vendors still abound

But then there are plenty of refreshingly old-fashioned technology vendors that simply provide software applications that promise to improve distribution, enhance pricing, automate underwriting decisions, improve client experience or cut costs in the back office. These too are InsurTechs.

It can seem as if so much is happening in InsurTech that a technological revolution is sweeping through the industry on multiple fronts. But the reality, as a study of the purpose, maturity, size, location, funding and technologies of a group of just over 100 InsurTechs suggests, is much less exciting.

InsurTech is really about increasing sales and cutting costs for incumbents

The activities of the InsurTechs in the sample reduce to three familiar categories: selling, underwriting and operations. Of course the distinction is artificial, in the sense that more efficient underwriting can cut costs, and improve sales through better service and increased capacity.

But the linkages between selling, underwriting and operations are a reminder that specialisation between companies – brokers, underwriters, reinsurers and the various categories, such as marine, health, motor and property – does not inhibit high levels of integration within companies.

True, “integration” in practice means internal siloes that separate sales, underwriting, claims, finance and legal departments. But nevertheless, insurance companies are – like any modern corporation – a  lower-cost means of combining services than purchasing them separately in the market. 

In fact, part of the promise of InsurTech is that digital technology can make it economic to unbundle the three components of an integrated insurance contract. The most obvious instance of this is the future the emergent insurance “marketplaces” are groping towards.

But such marketplaces constitute a mere twentieth of the sample, and the realisation of fully functioning markets in insurance risks and components lies many years away. As Table 1 shows, the majority of the sample focus on tasks nearer at hand: making sales and cutting operational costs.

Table 1

The headline-grabbing novelties associated with InsurTech account for a fraction of activity

Much is made of how InsurTechs are taking the industry into new risks (driverless cars, the Gig Economy), adapting it to changing social realities (pay-by-the-mile insurance, personalisation) and improving customer experience and engagement (mobile phone apps, health tips).

Yet less than one in five of the sample are engaged in innovations of this kind. Most are also tiny, with more than half employing less than 50 people and three quarters less than 100. Between them, they account for just 8.3 per cent  of traceable funding, more than half of it going to just four companies.

Likewise, embedded insurance, in which insurance is built into the purchase of a product or service provided by a third party – such as a bank or travel agent or motor dealer or retailer – is a lot less novel than it appears.

Embedded insurance is a new label for a old practice

Compensation paid to holders of Payment Protection Insurance (PPI) policies in the United Kingdom referred in some cases to loan, credit card and mortgage insurance sales made as far back as the 1970s, though the vast majority were sold between 1990 and 2010.

While embedded insurance is routinely portrayed as the insurance equivalent of Open Banking, in which companies source an insurance component from a third party to enhance the value of a product or service, it is less futuristic than this suggests.

For consumers it is an invisible – and therefore less than entirely welcome – evolution of the longstanding but noxious practice of selling one-off plans to protect new products. It has yet to translate into lower premiums. But for insurers it is a useful and growing form of low cost distribution.

Embedded insurance might sound like a whole new distribution strategy but the InsurTechs that make it happen are really providing an outsourced or white-labelled service for the digital age. Re-labelling white labelling as a B2B2C business does not alter the underlying reality.

True, embedding does require technical support in the form of the Application Programming Interfaces (APIs) that allow data to be streamed instantaneously to where the sale is taking place. So InsurTechs have emerged to sell Cloud-based embedded insurance platforms and standardised APIs.

APIs themselves are scarcely a transformative innovation. They originated in the 1940s, and the term was familiar by the 1960s. Their prominence today merely reflects the vast and growing quantities of data  available in digitised form, which insurers are struggling to understand and exploit.

Table 2

Indeed, current demand for APIs in the insurance industry is driven not by the vision of Open Insurance driven in turn by Open Data – in which consumers will own all data relevant to their insurance needs, such as their health and driving record, and consent to its use  –  but by something more quotidian.

The hottest InsurTech topic for investors is … operations

Sales demand a supportive operational infrastructure, and APIs provide one. Likewise, speedier on-boarding of clients, faster underwriting of risks, rapid distribution of policies and efficient processing of claims all require operational infrastructure. InsurTechs are obliging.

Indeed, judging by the flow of funds into the InsurTechs in the sample shown in Table 2, it is in operations that investors see the quickest wins. Investors have noticed that insurance has a well-deserved reputation for inefficiency, with clerks processing paper, emails and spreadsheets.

If greater operational efficiency can cut costs, it also creates room for insurers to add capacity, since  they have more money to compensate the investors that provide them with capital.  But, as Tables 1, 2 and 3 show, a number of InsurTechs in the sample aim to enlarge underwriting capacity directly.

Table 3

This can be done by picking better risks; more accurate pricing of the risks assumed via more and better data, including information from satellites and sensors; more efficient spreading of risk through reinsurance; the tailoring of exposures to the duration of a risk more exactly; and fraud reduction.

Telemetrics are about helping the insurer not the customer

InsurTechs in the sample are, for instance, fitting cars with dashcams, providing aerial imagery, tracking the weather, and using behavioural data to assess risk, including the risk of fraud. Others are adding capacity by specialisation, in SMEs or mobile phones or Gig Economy or Shariah insurance risks.

These novelties tend to attract more headlines than sales or customers. For example, one estimate is that telemetrics in motor insurance – despite all the vendor-driven press excitement about dashcams and video-driven claims  – own just 5 per cent of the insurance market in the United Kingdom. 

It is not surprising. The current insurance model pools rather than personalises risk, so bad drivers are subsidised by good ones. Higher risk drivers understand this in the same way that large users of  water understand that it is likely to be expensive to install a water meter.

Consumers nurse a justifiable suspicion of “telemetrics,” having learned from Facebook, Google and Microsoft that the term really refers to collecting information about the private behaviour of customers, so it can be sold as insights to third parties.

The old joke has it that everybody considers themselves an above-average driver, but consumers are much savvier than the joke implies. They know telemetrics will not help them but will help insurers pick better risks and price bad risks accordingly. Telemetrics help suppliers, not consumers.

Until consumers own their own data, and insurance companies either pay to use it it or secure the consent of the customer to make use of some of it for the purpose of writing and pricing an insurance policy that the consumer can trust, “telemetric” insurance will struggle to gain traction.

Elon Musk has promised Tesla drivers 20-30 per cent discounts on motor cover thanks to the data drivers share with his company. But there is a reason most insurers must continue to rely not on data about the driving habits of individuals but their age, occupation, credit score, location and vehicle.

Likewise, motor manufacturers may be sufficiently perturbed by the success of Uber and Zipcar to re-brand themselves as a “mobility solutions providers” but InsurTechs that cover part-time taxi drivers and drivers that hire or borrow cars for a few hours are still a marginal force in motor insurance.

Operational efficiency, not transformation, is what investors are backing

Likewise, for all the talk of improvements in customer engagement, the innovations often reduce to an SMS update on a claim (even the UK Passport Office can manage that) or cut-and-paste health tips emailed to customers (as opposed to premium reductions dictated by data from a step-counter). 

Table 4

The gap between rhetoric and reality explains why investors (see Table 4) are more comfortable backing firms whose ambition is no grander than selling a software package that automates existing processes, or APIs that facilitate data exchanges, or serves higher margin specialty insurance niches.

Table 5

Source: CB Insights, State of FinTech, Global 2021

This does not mean “InsurTech” is unpopular with investors – in fact, quite the opposite. According to CB Insights, funding of InsurTechs as a whole has grown at a compound annual rate of 25 per cent over the last seven years, to a record high of US$15.4 bilion in 2021 (see Table 5).

Since 2015, the InsurTechs in the CB Insights sample have raised a total of US$44 billion. In FinTech, only the booming payments, banking and consumer credit industries have attracted more investment. But the opportunity investors see now is not the same as the one they saw four or five years ago.

Scale belongs to not to the banal not the innovative 

Operational efficiency is less glamorous than telemetrics in motor insurance, or genomics in health insurance, or the application of behavioural science to insurance sales, but InsurTechs selling automation or data processing or client-facing software tend to be bigger enterprises (see Table 6). 

Table 6

This lack of scale on the part of the most innovative InsurTechs is evident in the funding they have attarcted so far too. As Table 4 shows, software sellers alone have attracted two and a half times as much investment as motor, life and health and marketplace innovators combined.

In Europe, McKinsey reckons the combined market share of the InsurTechs is probably less than 1 per cent of premium income. In short, the most adventurous InsurTechs have failed to fulfil their ambition of overthrowing the status quo in a manifestly inefficient, unloved and conservative industry.

Table 7

Early expectations of blockchain are unfulfilled

This (at least in part) reflects an early focus on blockchain as the disruptive technology. It was understandable, given the theoretical ability of blockchain to address the major source of inefficiency in the industry: the sharing and storage of reliable data between multiple parties.

Of the InsurTechs in the sample, more than half were founded during the blockchain-induced boom of 2015-18 (see Table 7), when post-Ethereum expectations that InsurTechs could disrupt and even displace incumbent insurance companies were at their height.

Blockchain promised results in all the areas in which investors are placing bets in InsurTech today: efficiency and cost-cutting in premium collection and claims handling, enhanced customer experience and engagement and improved and timelier information on which to base decisions.

This would be achieved by data-sharing. For example, insurers were going to cut customer due diligence and on-boarding costs by sharing encrypted Know Your Client (KYC), Anti Money Laundering (AML), Countering the Financing of Terrorism (CFT) and sanctions screening data through a blockchain.

The same logic promised reductions in fraud as insurers could share information about claims, reducing the risk of exaggerated or multtiple claims for an event; establish ownership of high-value items through shared digital certificates; and eliminate bogus sales by insurance brokers.

A shared, blockchain-based database was expected to expedite reinsurance flows too, thanks to smart contracts automating exchanges of information and money. Risks were to be ceded and claims made with notifications, settlement and reconciliation of payments all automated.

In its most visionary form, smart contracts tied to AI algorithms sifting through reams of written and visual data were predicted to take over nine tenths of underwriting, claims assessment  and settlement processes, transforming the cost base of the industry.

Smart contracts fed by “oracles” were predicted to transform the cost even of the chief innovation of the traditional insurance industry: the parametric policies that pay out not specific sums calculated by loss adjusters but amounts indexed to a measurable event, such as a month of drought or flood.

Blockchain-based peer-to-Peer (P2P) insurance, in which groups insure each other against loss, voting to make pay-outs against losses and sharing unspent premium income, promised to counter the poor reputation of the retail insurance  industry for value and customer service.

Blockchain in insurance is now the preserve of collective infrastructural projects

In 2017 a group of major insurance and reinsurance companies were sufficiently perturbed by the disruptive potential of blockchain across such a large swathe of activities that they set up a consortium to devise collective applications of blockchain technology.

That consortium (b3i) still exists, and retains the support of 21 of the largest insurance and reinsurance companies in the world, and 43 “nodes” on its network. Its flagship product (B3i) remains an evolved version of the excess of loss reinsurance placement product with which it began.

The B3i strategy is to provide a blockchain infrastructure (Fluidity) that insurers can use to build in-house blockchains, but which can support third-party products and services too. The Institutes RiskStream Collaborative has a similar ambition to build a standard infrastructure for the industry.

Both B3i and The Institutes RiskStream Collaborative (whose 40 members outwigh the 21 of B3i) have followed the example set by virtualy all blockchain initiatives launched by incumbent financial institutions and opted to run closed, permissioned blockchains on R3 Corda technology.

Chainthat (an information-sharing, workflow processing and reconciliation platform for insurers which launched initially on Ethereum) and Claimshare (which combats duplicate pay-outs on the same policy) followed suit. Incumbent insurers, like incumbent banks, have ditched the public blockchain.

They have done that because the information-processing inefficiencies in the industry lie between insurance companies as well as within them (where the conventional Software as a Service (SaaS) vendors are doing work that blockchain vendors rarely see).

Solving problems between companies requires infrastructural networks of the kind B3i and The Institutes RiskStream Collaborative are developing. They enable incumbents to leave internal systems untouched but do also represent a retreat from the initial vision of blockchain-in-insurance.

True, public blockchain challengers have survived. Etherisc, for example, still sells off a public blockchain oracle-fed smart contracts that cover crop failure, hurricanes, flight delays, digital wallet theft, loan collateral haircut coverage, life assurance and critical illness risk.

Scalability is an elusive quality

Munich-based Etherisc continues to pursue a compelling vision of decentralised insurance applications based on blockchain, but it is not yet a large company. According to Crunchbase, it has so far raised US$100,000 in a seed round.

One of its ideas proved interesting enough to attract an imitator, in the shape of an incumbent insurer. Axa launched a similar product (Fizzy) in September 2017 that used smart contracts to pay policyholders for flight delays.

Fizzy was scrapped it three years later because it had attracted insufficient business to make money. Again, even the best, most focused ideas with a large mass market to attack in the shape of airborne travellers proved hard to scale, even for a major insurer with an installed retail client base.

Likewise, the re-discovery of mutual (or P2P) insurance by blockchain start-ups have also proved hard to scale. One such start-up, Friendsurance,  has since extended its services into traditional disciplines such as running a a managing general agency, brokerage and bancassurance.

Artificial intelligence is more promising for insurers than blockchain

AI algorithms sifting through Big Data is proving a more powerful tool than blockchain for disrupting the insurance industry at scale. The rapid success of Lemonade, for example, which moved from foundation in 2015 to an IPO in 2020, is built on AI.

Instead of using blockchain to win the trust of customers by mutualising the risk and rewards, Lemonade adopted conventional corporate answers: BCorp status plus allowing policyholders to nominate a charity to which the difference between premiums and claims-plus-costs can be sent.

Its focus – renters, homeowners, pets, car and life insurance – is narrow. Using AI-driven chatbots to sell cover (Al Maya) and pay claims (Al Jim), the model works in terms of customers (1 million plus) if not profits (in 2021 revenue was US$128.4 million and expenses US$362 millon).

Table 8

It is not alone. Allstate, which has revenues of US$42.2 billion and makes gross profits of US$6.4 billion, uses AI-driven chatbots with customers. Indeed, AI is used by a majority of InsurTechs in the sample in a variety of ways: sales, risk assessment, underwriting, claims and fraud prevention (see Table 8).

AI is far more important than blockchain, where as many sampled firms insure blockchain businesses or connect to blockchain networks as actually use blockchain in their business. Nine out of ten do not touch it at all (see Table 9). When it comes to innovation in insurance, AI is what matters now.

Table 9

It is not hard to see why. AI is a misnomer. What it really describes is computational statistics: algorithms sifting through large quantities of structured and unstructured but digitised data in pursuit of information that might be useful. The value of this technique to insurers is obvious.

AI makes sense for insurers because insurers use data but badly

Since life assurance was invented in the 18th century, using primitive mortality data, insurers have used statistics to gauge the risk of an event occurring. The digitisation of data, from official databases to social networks, has simply expanded the quantity and quality of the data they can draw upon.

In expanding the availble data sets, the major change of the last decade is the ubiquity of the smartphone. It can tell insurers where people are, what they like and how they live. In addition, insurers can draw on satellite and sensor information about the weather, shipping, cars and buildings.

Insurers do of course collect vast quantities of data of their own, but inefficiently. Even on-line, policyholders must complete lengthy questionnaires asking for information insurers should know already:  name, address, age, location, occupation, claim history, driving record and so on.

So it would be an exaggeration to claim that the insurance industry has mastered data sufficiently well to understand customers completely or make policy and claims handling more efficient. In reality, anybody who buys an insurance policy faces much the same process today as they did 30 years ago.

No insurer has a holistic understanding of its customers. Indeed, far from organising their businesses around the client, insurers continue to run operations, let alone data analyses, that are siloed between the personal, property, motor, marine, commercial and other lines.

As a result, the insurance industry is failing to capture data, or organise it, or ask the right questions. Its members are refusing to share data with competitors in ways that would enable both parties to gain insights. Many insurers are simply overwhelmed by the quantity of data.

AI algorithms eating data are a major industry in InsurTech but …

Whch is why investors see an opportunity in the room for improvement. In the sample, data innovations account for a seventh of funding (see Table 4) and nearly a fifth of all the InsurTechs by number. The data specialities advanced by the firms in the sample are numerous and varied.

They include the eye-catching (such as using sensors and videos and standardised data-sharing between motor insurers to accelerate claims) and the predictable, such as boosting sales (by cross-selling or personalisation) and profits (using it to pick better risks and cut losses to fraud).

InsurTechs in the sample are offering to improve risk management by integrating historic loss, personal health and  fitness, corporate history, real estate, satellite and metereological data and even to analyse scientific papers in search of the next asbestosis catastrophe.

There are novelties (such as personalised or pay-by-the-mile motor insurance, checking the insured status of counterparts, digital identities for on-boarding and discounts on health insurance for good genes and a step-counter) but the striking innovation  is the infrastructural.

The fact that insurers need help automating data exchanges and aggregating and presenting data is the clearest symptom that they understand the value of data but not what data they have or could obtain, or what it can tell them, or how they can turn insights into streams of revenue.

… InsurTechs are not rising to the transformative potential of data in insurance

The innovative InsurTechs are not helping. They are specialists, in most cases trying to sell an AI and data-driven product or service (or indeed their entire business) to incumbent insurance companies that want to improve their existing modus operandi. They are not trying to transform the industry.

This is a consequence of the difficulties most InsurTechs have encountered in scaling their businesses. CB Insights has identified 34 “Unicorns” (companies valued in their latest funding at US$1 billion or more)  in InsurTech. It is not many: they make up just 3 per cent of the total number of Unicorns.

And the most valuable is not a FinTech but a 28-year-old conventional insurance broker worth US$5 billion. It is a minnow by comparison with established firms such as AON (capitalised US$68 billion) or Willis Towers Watson (US$28 billion), Travelers (US$44 billion) or Allstate (US$37 billion).

Disappointment has prompted InsurTechs to scale back their ambitions. Five years ago, at the height of the blockchain-inspired start-up boom, they dreamed of overthrowing the incumbents. Now they would be grateful for a contract from or a partnership with an incumbent, or a purchase offer.

The US$44 billion sunk into InsurTechs tracked by CB Insights has exhibited a consistent bias to maturing rather than start-up InsurTechs. This is evident in a rising number of deals (including acquisitions) and the growing value of deal size (which averaged US$33 million in 2021).

The way to break out of this pattern is not for incumbents to embrace novelty instead. They must do harder, more ordinary things: ditch legacy systems, adopt Cloud, train employees, share data with competitors and invest in new knowledge and technology.

A true InsurTech revolution hinges on changing the inecntives of insurers and insured

As for InsurTechs, their responsibility is to recover the ambition of five years ago, because the traditional insurance industry is as broken now as it was then.  Not just because it has failed to control its costs, or adapt to the digital age, or even because its customers actively distrust it.

In reality, the insurance industry is ripe for disruption because it is impossible for an unreconstructed industry to deliver what it promises: redress for losses caused by the realisation of the risks it ostensibly covers, such as car accidents, fires, earthquakes, floods, disease and death.

Conventional insurers sell policies as a form of risk management. But the only risk they really manage is the risk of being put out of business by a pay-out. The entire paraphernalia of the industry – capital, excesses, no claim bonuses, reinsurance, parametrics and so on – is designed to avert that outcome.

The result is a form of cognitive dissonance on the part of the insured, who believe they are fully insured against an external eventuality, yet never draw the obvious conclusion from a claims process that always disappoints and almost never covers the whole amount of the loss.

There is a fundamental conflict of interest at the heart of the insurance industry as it is structured today. For the insured, a risk is a potential disaster. For the insurer, a risk is a financial opportunity.  Formally, insurers assess risks in terms of probabilities. In fact, they assess them in terms of profit.

The question which a truly disruptive InsurTech should address is this: why not measure risks accurately and price insurance against them accordingly? Digital data and computational statistics make it possible. The InsurTech that made it happen would launch a revolution.

If the revolution is to succeed, it is one consumers must support. A cynic would say that consumers are complicit in the current pretence, preferring to pay less for a mythological product to paying a full price for an honest one. That is a claim which only an InsurTech has the incentive to test.