Andy Ann, founder of Yas Digital, named the company for its sound of satisfaction, versus the usual frustration whenever customers deal with insurance. The question is, can this insurtech deliver satisfaction to the insurance industry, as well as to policyholders?
Ann’s company is combining a variety of data-sourcing techniques, including telematics, AI analysis and blockchain for provenance. The output is dynamic or streaming data. This stream, if it can scale, becomes a ground-up source of information for insurance companies. It makes small-ticket policies suddenly a lot more commercially interesting.
“Data streaming is possible, but no one is doing it,” Ann told DigFin. “We aggregate data. We’ve aggregated 100 million data sets in our first year. Now we’re seeing some interesting business models.”
Yas has twenty or so projects on the go around Asia. For each one, it collects data, figures out a distribution strategy, understands the claims process, and then goes to insurance companies asking them to underwrite the product.
Only three insurers have agreed to participate so far: Etiqa, Generali and QBE.
Ann says since Yas launched operations in 2021, it has already broken even, and its platform has facilitated more than 1 million policies (Yas is licensed as an insurance agent), in Hong Kong, Malaysia, and Vietnam.
It has enough moment that Ann says the startup is in the process of closing a Series A funding round, which it will announce soon, with proceeds earmarked for expansion into new markets (including in Europe) and more projects.
Startup credibility
How credible is Ann’s vision for streaming data?
Insurtech is littered with the carcasses of simplistic tech fixes that didn’t pan out. Take telematics, which is about sending information in real time over long distances. Sensors on fleets of cars to indicate who’s driving sensibly or recklessly, sensors on watches or apparel to determine who’s exercising regularly.
Yet telematics companies, ahem, crashed and burned. They underestimated nuance (is the speeding driver in a city or in a rural area?), had conflicting objectives (is the idea to lower premiums, or prevent health problems?), and sparked worries of discrimination.
The biggest problem was simply companies didn’t have a good handle on the data coming their way, and lacked the means to turn it into accurate insights.
If Yas is struggling to attract insurance partners, maybe it’s because the startup world hasn’t often lived up to its claims of transformation.
But neither has the insurance industry solved its own, very real, problems. Customers usually have no idea what’s packaged into the policies they’ve purchased. Insurers are clueless about their end customers. Fraud and abuse keep costs high. Attempts by insurers to push their own apps (usually with ‘wellness’ hooks) have yielded few gains, with ‘engagement’ superficial and still unproven as a means of increasing new premiums.
And the world of data and fintech continues to innovate. Insurance companies lack data about their customers. In an age of e-commerce and digital lifestyles, the insurance industry is still reliant on contrived marketing ‘personas’ and demographic guesswork.
In the general insurance market, data-driven business has been relegated to microinsurance, such as flight insurance pioneered by Zhong An, or as add-ons to microlending in developing markets. Ann says insurance executives have come to regard ‘data-driven’ as meaning cheap or for poor people.
“It’s not about asking for someone’s salary or income,” Ann said, “but protecting people on the go through smart contracts and data.”
Whether the end user is rich or poor, embedding digital insurance based on usage data requires high volumes of transactions to make commercial sense.
Sourcing data
The first challenge is that data is simply hard to come by and harder to analyze.
The dynamic nature of streaming data looks good in theory: the data comes from the policyholders themselves. A car insurance cover starts when someone steps inside the car, and ends when they exit. Accident insurance starts when you run in your sneakers at a certain pace (embedded with a chip) and ends when you slow.
Sounds like telematic disasters all over again. But Yan says there are more ways to collect data and correlate it with other data sets, creating a manifold view. This is how Yas develops insights.
It requires an increasingly complex operation, using machine-learning, data analytics, GPS geolocation, telematics, APIs, sensors (the Internet of Things), and blockchain.
Yas also has to diversify its use cases for all of these pieces to fit together. It is doing many things at once: personalizing insurance, providing transparency, enabling decisions based on data and claims and prices based on usage. It integrates into platforms that provide services from health to commerce. And it is designed to serve all kinds of customers – initially in general insurance (home, auto, travel) but also in health, term life, and climate insurance.
“There’s no ‘wow’ factor in insurance,” Ann said, suggesting this is what Yas should deliver. As a platform, though, it is not dealing with the policyholders. Its customers are businesses with reach, and the insurers Yas needs to underwrite its projects. It’s not clear to DigFin who needs to be wowed.
Streaming examples
What does all of this look like?
For one example, Yas can put a sensor on a bike to trigger a policy if there’s an accident. But what kind of insurance? Yas’s data science suggests men – in aggregate – want to insure the bike itself, but women want to protect their bodies. Two different products, requiring two separate marketing approaches. Earlier this year Yas rolled out versions of bike accident insurance with Generali.
In other cases, the policy is bundled within a purchase. For example, Yas is now working with Kowloon Motor Bus to put travel insurance inside bus tickets between Hong Kong and Shenzhen. The ticket covers lost or damaged luggage reported within the same day as travel, with claims capped at HK$10,000. This reaches 250,000 travelers a month.
That kind of volume churns out a lot of data, which Yas is using to extend the product to discounts with a bus ticket at dozens of Hong Kong merchants, such as bakeries. But data’s a chicken-and-egg thing: “We need massive amounts of data to support this product,” Ann said.
Because KMB is paying for the insurance product, it gets to keep the customer data. That might be a good deal for KMB, but that creates a barrier to Yas to develop new products from insights into those riders.
Winning trust
A second challenge to telematics is trust. People don’t like being monitored, especially those who feel they might face higher premiums instead of lower ones.
Yas is using blockchain technology to create a provenance and privacy layer to address this.
For example, he cites a use case involving a retail shop that might not wish to purchase insurance to cover a business interruption (from a flood, say). It would have to cough up its annual P&L to demonstrate its revenues, information – secrets – that the insurer gets to keep.
Yas doesn’t rely on backward-looking financial statements. It uses sensors to track real-time action at the register. But the information is stored on a blockchain. The relevant data during the period of outage is only shared with an insurer when the store owner makes a claim.
This pitch – your information is kept private – may appeal to customers, but it’s also aimed at pleasing insurance companies, by giving them confidence in the data, and thus useful in mitigating fraudulent claims.
Big it up
Add it up: AI and machine learning drives customization. Blockchain ensures a closed loop for privacy, but also for transparency when required. Telematics and usage-based models reward behavior and produce fair pricing. IoT devices provide real-time information to better manage risks. Big-data crunching enables dynamic pricing. And Yas’s digital platform is meant to combine everything into a convenient platform for companies and insurers.
And yet: “We reached out to all the insurance CEOs, and they rejected us,” Ann said. Well, all but three. Ann admits to feeling disappointed.
“The roadblocks we face are that most people in the industry just want a lucrative product,” Ann said. They want big premiums – which streaming doesn’t offer. There’s less interest in developing digital-friendly products just to reduce claims fraud or better serve customers.
Another way of putting it is that insurers know that embedded products are very low margin and therefore require scale – massive scale. Yas is trying lots of creative things, but an underwriter might prefer it offer just one big idea.
Scale could come in unexpected ways. For example, data from selling insurance could be used to generate rewards and loyalty programs. The KMB example, letting riders get discounts at local retailers, is one example. An insurer could use a digital platform like Yas’s to offer buyers of, say, car insurance tokens for free coffee. It could be a way to build brand awareness and perhaps customer loyalty; with enough data, it could open the door to an agent attempting to strike up a relationship with the customer.
Another possibility for scale is aggregating many use cases within a single customer. In Malaysia, where sports associations are organized by the government, Yas and Etiqa are providing accident cover. The individual sports and types of protection all vary, but operations can be centralized to deal with a single interface. And sports cover is an entry into health insurance.
Nice niche?
Yas is exploring opportunities in many industries besides travel and sports, including e-commerce, retail, and payment gateways. Ann says he is also developing ideas for crypto, such as insurance for decentralized financial transactions, and for electric vehicles, among other things.
The use cases are many, if not limitless. Real-time data can be embedded and packaged to serve multiple users: insurers, industry, and the end users too. But the flip side of this variety suggests scale remains elusive.
Take one final use case: Rolex watches. In Hong Kong, Yas works with a distributor of Rolex watches to embed insurance in the purchase. It covers incidents of loss or theft. QBE is the underwriter.
But with the purchase information saved in Yas’s blockchain, there’s now a certification for use on the secondary market for luxury timepieces. If someone wants to sell the Rolex, they can consent to a third party to check the record, which includes purchase price, the watch’s provenance, and certification of any insurance or police records.
The retailer also has access to this information, far more than it would get from just the buyer’s credit card details.
It’s a clever idea that’s live in the market, but it’s very niche. The Rolex project shows how picky the underwriters are about choosing what kinds of data to stream. If Yas can identify really scalable data streams, it should find insurance executives ready to write a check.
This story concludes a short series on embedded insurance. See our introduction and our insurance-company case study.
