You still have to support the policy, process claims, customer support, etc. The biggest expense on a unit-economics level, post claims paid, is marketing -- acquisition and retention costs.

If you look at the auto insurers, it's incredibly competitive and everyone is trying to balance those unit costs with the loss ratio. They are all moving targets but premium pricing is heavily regulated meaning your pricing will 100% come under scrutiny from some states (this must be done individually for every state in the U.S.) so any changes to pricing tends to be a complex process that could take months, if not a year+ (in some states) to take effect.

So when you get pricing wrong and are taking a big claims loss, it takes some time to dig out of that and you'll also piss off lots of customers who got in "cheap" and are now getting a rate increase. And when you get pricing wrong and you're loss ratio starts looking better, your competitors may be out-pricing you, making you uncompetitive until your adjustments are improved.

Losses + “Loss adjustment expense” are your costs. Loss adjustment expense is broken into Allocated Loss Adjustment expense - expenses tied to a particular claim (Typically lawyers); Unallocated Loss Adjustment Expense - overhead.

Theoretically two companies with the same policies will pay out the same losses but will differentiate themselves in expense ratios.

In some lines it’s not bad to have loss + expense ratios > 100% because the average time of premium is very far from the average date of loss, so while there is an underwriting loss it is offset by the investment gain.

I'm not so sure. Insurance companies exist based on probabilities. How much margin do you need to make a given profit worth the risk? What about that dollar you brought in where you ended up paying out $10?

Hence Lemonade.