Assaf Henkin

The through-line

Assaf Henkin's recurring claim about insurance is that there is nothing physical in it to build: "insurance is there you know there's no real product it's all data" [1]. That framing is what let a team of data scientists and developers with no insurance background see the category as adjacent rather than alien, and it drives everything else he argues. If the product is data, then the leverage sits in what data you collect, what you choose to notice in it, and whether the answer you derive can actually be priced and sold.

The second half of the through-line is a moral and commercial preference inside that data view: the industry has built its models almost entirely around what is wrong with a person, and he thinks the upside is unclaimed. Incumbents, as he describes them, are "way too focused on uh understanding what's not so great about individuals," and for every such finding "they penalize you by either increasing the rate or limiting your coverage" [1]. His question was simply why the arrow only points one way: "why is it that you can only you know it can only get worse it can get better" [1].

On rewarding healthy behaviour instead of penalising risk

He takes the analogy from auto insurance, where models already exist under which "if you drive safer if you drive less you'll get benefits for that," and notes that in life and health "not so much we couldn't find too many modern approaches to that" [1]. Sprout Insurance was founded to close that gap: to use data and technology to detect people who take care of themselves and reward them, both in how easy the purchase is and in the coverage, riders and price they end up with [1]. He is careful that this is not a wellness slogan bolted onto a policy. The reward has to show up in two specific places, the customer journey and the underwriting and pricing, or it is not a reward at all [1].

On the Quality of Life Index

The mechanism is what he calls the quality of life index, a way of assessing individuals using lifestyle data: how active you are, how well you sleep, how you live your life [1]. It is collected two ways. First, an adaptive questionnaire with thousands of possible questions that changes according to how you answer. Second, by letting the customer connect a wearable, with Garmin integrated at the time and an intention to connect with "almost anything that you would want to use" [1]. This lifestyle layer is then combined with other sources, including basic demographic data, prescription drug data and credit data [1].

On personalising the buying journey

The first payoff from that combined data is predictive routing. Because the model can predict early in the process how likely a customer is to qualify for a given product, the customer can be sent to the right product with the right coverage amount and the right rate range from the start [1]. He concedes this "might seem you know simple and obvious," then argues that in life insurance it is significant, because most purchase experiences are generic: a person gets a first quote, moves to the application, and "they might be denied or they might you know by the end of it see a price that is double from what they saw and that's all because it's very generic there's no personalization there's no real data engineering" [1].

On underwriting classes and creating new ones

The second and harder payoff is in underwriting itself. He describes the standard ladder of preferred, preferred plus, super preferred and standard as "a very rigid process mechanism that the industry created," one in which everybody is forced onto a small number of rungs [1]. His argument is that an additional data layer changes where a person actually belongs: someone who came out preferred on traditional underwriting may, once quality of life data is added, qualify for preferred plus "or maybe somewhere in between there," and there is no way to express that in the traditional world [1]. So the company works with reinsurance companies to create new underwriting classes that sit in between, alongside a carrier partner for the specific product it files and sells [1]. He is explicit about why this is good business rather than charity: a better rate makes the offer more competitive and more likely to be accepted, a higher value product means lower surrenders and lapses, and a satisfied customer may buy additional products later [1].

On building inside the incumbents versus building in parallel

Before settling on that model, the team spent six months meeting carriers across London, New York, Chicago, Los Angeles and the West Coast, including Hiscox, Aviva, Guardian Life, John Hancock and New York Life [1]. What they found was that every problem raised was "enterprise software or i.t type problems," such as processing claims faster or detecting who is really a smoker, and that these problems sat "within a certain um guard rails that they created" [1]. His assessment was blunt: the team could probably solve them, but "it's not really interesting enough" [1]. Some of those companies offered to acquire the team outright to build internal solutions [1]. He concludes that a newcomer to insurance faces exactly two options, either building solutions for the carriers and existing players, or stepping back and building "something that will be in parallel to what is happening today" from the outside [1].

On teams that outlast any one company

He treats the founding trio as the durable asset. The three founders had worked together for about twenty years across three startups, and when they built the new company as a team of around ten, the other seven were the data scientists and developers from the previous venture [1]. That prior company, a big data platform analysing consumer behaviour across web, social and mobile for marketers, planners and strategists, was acquired by Singtel in 2014 [1]. He describes leaving a year before the earn out expired "because it was a little bit too much for us being part of a 50 000 employee company" [1]. What came next was unusual in sequence: investors approached wanting to back the team "even before we knew exactly what we're going to do," and insurance surfaced from those conversations [1]. His first reaction was scepticism, "what insurance really you know was sounded very strange to us very far" [1], and the team's method for resolving it was to build a consumer data platform first without knowing the application, then let research point them somewhere. They gravitated to life and health specifically because there the analysis of an individual feeds directly into the product, the underwriting, the pricing and the acquisition of those people, which is what they already knew how to do [1].

On where to put the operating team

He is equally practical about geography. The company spent its first year and a half building technology and foundations, launched direct to consumer, and roughly six to eight months later realised it needed customer service staff and licensed insurance advisors to help people who want assistance and advice before buying [1]. Scaling that team in New York "was not that easy," and advisers pointed them to Omaha, St Louis and Kansas City [1]. Kansas City won on two counts: a central time zone and "a very relatively high amount of licensed insurance advisors life insurance advisors," which let them build a team of around thirty quickly [1]. The company runs split across Tel Aviv, New York and Kansas City, with Henkin historically dividing his time roughly half and half between Israel and the United States [1].

Takeaways

• The founding insight is that insurance has no physical product, so the whole business reduces to data: "insurance is there you know there's no real product it's all data" [1].
• Incumbent life underwriting only prices downside, penalising smoking, weight and family history; the unclaimed opportunity is rewarding people who live well, as auto insurance already does for safe drivers [1].
• The quality of life index gathers lifestyle data through an adaptive questionnaire with thousands of possible questions plus wearable integration, starting with Garmin, and combines it with demographic, prescription drug and credit data [1].
• Rewards must land in two concrete places: predictive routing that sends a customer to the right product and rate range up front, and underwriting that can move someone between or above the rigid preferred and preferred plus classes [1].
• New in-between underwriting classes are built with reinsurance companies, with a carrier partner for the filed product; the commercial case is competitiveness, lower surrender and lapse, and repeat purchase [1].
• A newcomer to insurance has two options only: build IT solutions inside the carriers' existing guard rails, or build something in parallel from the outside [1].
• Investors backed the same three founders and their data science team before the idea existed, and the team built a consumer data platform first, then chose life and health because individual analysis there feeds product, underwriting, pricing and acquisition alike [1].