Key Highlights
20 different pricing models in just 9 months
New policy creation cut to 5-10 mins from 1 hr
Challenges
Aviva had relied on Excel for their pricing tools but had become frustrated by its slowness and instability. For underwriters, it was taking 20-30 minutes to simply open the software and download the data from the company's policies and systems database. For pricing actuaries, too much time was also being spent investigating and fixing bugs across their 20 separate rating tools in Excel. These challenges meant that underwriters were delayed in responding to brokers requests.
Solution
Aviva initially considered building their own solution but determined the complexity of the project and the resources, expertise, and skills needed would detract from the pricing team's ability to add value to the business. It was also likely to take too long to build.
Instead, Aviva looked to the market and selected hyperexponential. The GCS team worked closely with hyperexponential on an initial design phase for successful implementation. Key to this phase was integrating Aviva's existing underwriting workbench systems and replicating and improving models that had been created in Excel. This included building single modules that are used across rating tools. The exercise took just nine months.
Getting up to speed with Python
The new platform leverages Python computing language. Python is relatively easy to understand, and so will enable fast onboarding of new users. Not all of Aviva's pricing actuaries had experience with it, but hx delivered tailored Python training to get the team up to speed.
Results
The shift from using excel to hx's modular model development framework saved 75% of the build time for certain models. The team can make a change to the rating once and see it automatically rolled out across tools. With no bugs and errors to check, the pricing team can focus on how to improve the models and better support the underwriters.
New policy generation is now much faster, averaging under 10 minutes compared with up to one hour previously. As a result, pricing and underwriting decisions are made quicker and more accurately, making underwriters more confident and competitive. This has considerably improved relations between the two teams.
Improved access to data is reducing time spent on data processing and powering portfolio analysis. The team are exploring batch testing in hx, enabling fast impact analysis. The deeper analysis is helping the teams make more informed decisions.
What's next?
Aviva is now looking to leverage hx's capabilities to unlock deeper pricing insights, delivering richer analysis and driving improved data-led decisions. To do so, Aviva aims to plug in its other core systems, such as CAT and risk modelling software. It is also looking to add Machine Learning capabilities, which will be easier to enable with Python. The combination of improved data, process automation and deeper analytics will ultimately support underwriters to service their clients better.
