[mriolfi]

phonon you are absolutely right. It's not possible to use A/B tests for admitted insurance use cases in the US, and we don't let our clients do so. This tool was designed with non-admitted use cases or non-insurance products, such as waivers, in mind.

Also, the names on the screenshot are illustrative as the client could vary other factors that are not price in the A/B test, like user verification workflows, deductibles offered, etc. The goal is to find the optimal solution that creates value for the company while matching the risk preferences of the users.

[phonon]

Thank you for relaying that so clearly. So here's the problem.

On Product Hunt[0], 2/8 screenshots clearly show that type of illegal use case ("+10% pricing"), and your own (US based) customer thanks you for "AB test pricing"[1]. You also can't get a product type that is more the personification of highly regulated filed personal lines as "auto insurance".

Literally in a comment from a few hours ago, you tout your ability in "running A/B tests with different pricing logics to optimize profitability."[2] and "your business users can launch and make changes (e.g., increase the price to certain categories of policies"[3].

If you thought what happened with Zenefits was bad, this type of thinking is 10x worse. As in, carrier/you gets fined and has to send rebates to all customers that might have been affected, program shuts down, and you lose your license.

If you're acting as the Broker (MGA?) you have to be extremely careful in whatever type of control you give to an unlicensed entity you are distributing through. What someone from outside of insurance thinks is a routine pricing optimization, "(What if we charge 5% more before a busy weekend?)", can cause untold compliance grief. Plus, if you give the unlicensed entity enough control in how the insurance is presented and sold, they themselves might now need to become licensed producers.

I'm also skeptical you are doing much in the non-admitted space, as marketing restrictions, three declinations, surplus lines stamping etc. is not something you can automate for a consumer (small personal lines) product.

I have no opinion on "non-insurance" products (whatever that is) except that must be a very small percentage of what you are trying to do, as you call yourself "An Insurance Platform".

[0]https://www.producthunt.com/posts/tint-insurance-platform

[1]https://cards.producthunt.com/cards/comments/1231720?v=1

[2]https://news.ycombinator.com/item?id=26026901

[3]https://news.ycombinator.com/item?id=26028605

[mriolfi]

Thanks for the follow-up message, phonon. You are absolutely right and we take these concerns very seriously. Insurance is one of most regulated industries for good reasons.

The A/B test can be used for non-insurance products, such as waivers, trip fees, etc. Definitely agree that this is a small niche and won't change anything in the insurance world, but can be valuable for a subset of the companies.

Our goal when designing the product is to make it modular so it empowers our clients with what they need. We have customers that don't use everything we offer, and that's totally cool!

[phonon]

OK.

Another issue is "AI based underwriting" which is not allowed in California and other states unless the Department of Insurance of that state has a chance to review your algorithms. (And even then, I wouldn't bet on it.)

California in particular requires full upfront public disclosure of all underwriting rules.[0] (What you can do, is gather data for risk optimization, and use that to inform your underwriting rules when they get periodically updated and filed, but insurance companies more or less do that now.)

[0] https://www.faegredrinker.com/en/insights/publications/2018/...