Avoid mispricing with a market model

What is your price?

Usually, when someone discovers what I do, I'm asked how insurers come up with their price since there is often a large price difference between the top 10 prices. Then almost without fail, I get challenged on why has their price gone up from last year, which always leads to a rant on having to call up an insurer, mess about with press 1 for this, 2 for.... followed by a 10 to 15-minute wait to finally speak to someone... and then that conversation that nearly always ends up with a discount being offered which matches the cheapest price.

The outcome, the customer feels that all insurers cannot be trusted and that this is now just the annual game, so they always shop around for the best price.

What if you could estimate the market price at the point of the quote?

If you could, there are two immediate benefits:

  1. Improve price optimization - typical margin improvement up to £5 per policy
  2. Reduce the inbound call volume & improve retention

Sounds good, what do I need?

Well, a Market model! (Which is a model that will estimate the cheapest (market) price.)

We've seen many clients with their own market models, and used them as part of their pricing architecture, to varying degrees of success.

At Pebbles, we've built (and replaced) market models for clients that range from large insurers, brokers, MGAs and start-ups.

Each market model we have built is bespoke to that client. We fine-tune our machine learning algorithms to the client's footprint and, work to agree on any overlays to address specific features unique to the client. We are often asked to help with the underlying inflation assumptions needed to support the final prices to meet intended loss ratio targets.

The whole process typically takes us up to a month, depending on the level of integration with the client's processes that is required.

Finally, for those clients that have the capability, we can also deploy some of our propriety data products that have produced a significant uplift in our model's predictive power.

Please, drop Daniel Holland or Sherdin Omar a message to discuss.

And don't forget to follow our Pebbles page for more updates.


Continue reading