(The Arc de Triomphe de l’Étoile, Paris. Image credit: Skitterphoto.)
Few InsurTechs have moved as rapidly and surely toward success as Shift Technology. The company was founded in Paris in 2014, bringing a new approach to insurance fraud detection through its FORCE solution, which replicates, automates and scales the deductive capabilities of insurers’ fraud handlers work through the use of artificial intelligence and large scale/massive processing.
FORCE detects and provides context for a wide range of fraudulent activities, from opportunistic claim exaggerations to schemes executed by organized crime networks. The SaaS solution combines sophisticated AI and human data science expertise, replicating and multiplying the deductive capabilities of an insurer’s best fraud handlers. FORCE applies a wide range of third-party data to rapidly and accurate develop a full and precise picture of each claim. Using data such as weather conditions and satellite imagery, social media monitoring and other sources, FORCE contextualizes and expands claim data. It then uses that enhanced body of data to assign each claim a numerical score representing the likelihood of fraud.
Shift has raised over $100M USD in funding and has signed contracts with over 70 insurers globally—many of them among the largest insurers in the world. In addition to its Paris headquarters, today the company has offices in Boston, Toronto, Tokyo, Singapore, London, Madrid, Zurich, and Hong Kong. Its team consists of more than 200 of the developers, data scientists, and project managers. Having reported on Shift’s new funding round of $60M USD and its first U.S. commercial insurer client, CNA, earlier this month, we thought the timing was right to reach out to the company’s CEO Jérémy Jawish on Shift’s remarkable journey to date.
Insurance Innovation Reporter: Jeremy, tell us how Shift Technology came about. What opportunity did you see to provide the insurance industry something new and useful?
Jérémy Jawish, CEO, Shift Technology: Co-founder Eric Sibony and I randomly ended up working at the special investigations unit [SIU] at a major insurer in France. The investigators we worked with were super expert, but worked manually. We thought we could create something to automate their processes. What we imagined was automating what they were doing when they looked at a large amount of data—but reproducing it on a larger scale, looking at all available databases simultaneously. We asked, “What if this expert could see all the data and links between phone numbers, claims, etc.?”
IIR: How does this differ from available solutions such as link analysis software?
JJ: Link analysis is a much more basic kind of processing; this is looks to all available databases and the Internet broadly, and rather than links and aims to detect subnetworks. When you’re looking over a huge network, you need to apply algorithms to reproducing what the experts do on a much larger scale. We realized that this is not something that an insurer could do internally—it doesn’t make sense for a given company to hire 100 data scientists. So the idea was to build a company and provide the application on a software-as-a-service (SaaS) basis, charging companies a subscription fee.
IIR: What were the challenges of starting up a company on that model? How ready were insurance companies to adopt a SaaS fraud detection model?
JJ: In 2014 selling SaaS to an insurer was a nightmare; the prevailing attitude was, ‘You’re too young, we won’t work with you.’ We held our first investment seed round in 2014, and then we acquired our first customer in 2015. Simply having a customer creates an opportunity for growth—once you detect a network of suspicious claims for one insurer, it involves others. Insurers also talk to their other peers, and if you’re delivering significant value, everybody wants to work with you.
IIR: So from the first customer you were able to develop a reputation. Tell us a little about where you stand today in terms of market penetration.
JJ: In France we have deep market penetration and very solid market share. In some markets, for example, Singapore, we work with the General Insurance Association of Singapore and have access 100 percent of the relevant data under a common agreement with all the insurers. Hong Kong, with the Hong Kong Federation of Insurers, is another example of that kind of arrangement, where we are deploying a platform for all auto and health insurers. We don’t mix data of carriers, especially in Europe, unless we’re working with associations. Once insurance companies realize you’re adding value, it can go very fast.
IIR: How does the SaaS model speed adoption?
JJ: With SaaS, the implementation cycle is very fast. On average, a given insurer can be in production within four months. It’s a very light implementation burden for carrier side. We’ve found that a good portion of clients are achieving very positive ROI in under 12 months. In terms of our growth, since we can get insurers up and running within a few months, we have found that we have been able to achieve significant market penetration in just 8 months.
IIR: Tell us a little about your strategy of growing in regional markets. Obviously, in each jurisdiction, detecting a network of suspicious claims for one insurer, will involves others and spread awareness and interest.
JJ: Yes, and as you working in a country with more carriers you gain experience about the unique ways that market functions. Every country we work in has a dedicated data science team on site—for example, in Boston in the U.S. The more they know about the market, the easier it is to see what’s wrong and right. Going from one country to another can require a lot of adaptation. For example, in Singapore there’s a requirement for policyholders to visit an insurance claims reporting center within 48 hours. In the U.S., there’s no such thing. So doing business can be very different from country to country. It’s important to stress that in every country, we’re developing expertise that no insurance company could develop internally.
IIR: What are other key to Shift’s success?
JJ: One of the most important factors is that the offering is completely SaaS. It’s one product that scales to all insurers, and if our research teams discover new algorithms, we can apply them to all customers. We charge our customers a fixed fee per claim and everything is included. Another key factor is that our team is very rich in insurance expertise. Our only job is to make them successful. This sounds like a cliché but, saying it is one thing, doing it for real is another.
IIR: What’s the focus of Shift’s growth strategy now, in the wake of your latest funding round, and how would you place it in the larger context of your overall growth?
JJ: Our growth of customers and employees has been more than doubling annually. We’ve gone from 20 employees in 2016 to 200 now. By the end of this year we expect to have 350. We opened our Boston and Tokyo offices last years, and we opened our new office in Toronto last month. Our drive now is scaling in terms of geography and having more market penetration in the U.S. and Japan. We’re also investing in data scientists and researchers, and we’re making a big investment new products.
IIR: Shift is known for FORCE, but your recent funding announcement noted that you have also developed a solution called Luke for claims automation. Maybe you could expand a bit on your approach to product development.
JJ: We think about the product development process a little differently than a typical software vendor. Instead of building something we think our clients will want and then introducing it to the market, we work very closely with one or two carriers who have already expressed interest in a set of capabilities and begin there. This is what we did with our Luke solution. You can almost think of these carriers as our Alpha customers. When these initial carriers go into production—like AXA Spain for Luke who we mentioned in the funding announcement—we consider that the “launch” of the technology. We then bring another couple of carriers on board, get them to production and then we consider the product ready for what the software industry typically considers general availability.
We never build on our own, that’s one of our strengths. We partner with carriers because we want to build the right thing for them.