October 18, 2016 Philip Aston

Building the Usage Based Insurance App – pt 1

This will be a four-part blog, describing how we can change the world of auto insurance using realtime analytics and optimized communication for devices connected via an unreliable mobile network. We’ll be looking at the opportunities to leverage realtime data using Push Technology’s Diffusion Cloud, IBM’s Watson, IBM’s Streaming Analytics, and IBM’s Cloudant product.

As with every good business idea there first needs to be a problem waiting to be solved. In this case we are exploring how traditional auto insurance can be transformed with a usage based model that not only reduces expenses for safe drivers, but also reduces expensive liability for insurers.

When applying for an insurance policy you are typically asked a bunch of questions related to your driving history, and then you are expected to make some predictions; how many miles do you expect to drive this year? Where will you park your car? Are you driving socially or commuting?

These questions are a bit tricky and if your circumstances change you are expected to call your insurance company to let them know.

There are also the techniques that don’t appear to make much sense. For example changing your job title slightly you can reduce the cost – a UK comparison site even provides a tool for this. It is also common to get a cheaper insurance quote by increasing the annual mileage, which is also counter-intuitive.

Most people insure their car on a yearly or semi-yearly basis solely because it is the cheapest way to do it. Paying monthly leads to an 11% higher cost on average.

This whole process is clearly very inefficient and wasteful. There must be a better way?

Usage Based Insurance

The solution to this problem is to collect as much of the data as possible on the fly. Instead of predicting how far you will drive, an app will record this as you drive in real-time.


Taking this idea to its logical conclusion you should no longer need to buy a yearly insurance policy and could instead pay per mile. Depending on your location the insurance may cost more, for example it is expected that driving through the centre of New York City would cost more per mile than driving on the highway. A route with known weather or traffic conditions ahead could also increase or decrease the risk and therefore the price. The insurance company could significantly increase their insurance cost per mile and depending on the amount driven the customer would still save a lot of money. It seems like a win-win for both sides.

In order to answer some of the more complicated questions such as which job title you should use, some of the tools available on the IBM Bluemix platform can be utilised; Tradeoff Analytics can be used to determine which insurance policy is more suitable for the customer, Streaming Analytics can be used to get a price that is relevant to the current location of the car, or suggest a lower risk route.


Check out part two of this blog to see how we actually built the demo application, the components from Bluemix used and how Diffusion Cloud easily brings all these realtime data streams together.

Like this idea? Or have another use case you think realtime technology can enable? Get in touch! (@push_technology).

Want to see how easy it is to get started with Diffusion Cloud? Sign up for a free demo here.



Enjoy the rich functionality of Diffusion 6.7 as part of your event-driven application.

Quick Start Guide

Step-by-step guide to getting started.

Diffusion Cloud

SaaS offering that focuses on business.

Diffusion On-Premise

A pub-sub platform for real-time applications.