See how we reduced time for testing multiple tarrifs from 2 days to 40 minutes and number of errors from 91% to 5% on average. 

About customer

One of the largest non-life insurers in the Nordics – an association of mutual insurance companies.

Tech Stack: Earnix 

Timeframe: October 2023 – ongoing 

Our Team: 3 Business Analysts (BAs) 

Business challenge

The business need was to build and sustain a testing strategy in Earnix Price-It and to maintain consistency between PC and Earnix. Before Sollers joined the customer, premiums in PC were incorrect (too high or too low) and required manual adjustments by the underwriters. More than 10 integration errors across 11 tariffs occurred.

The customer’s long term goal was to automate as many processes as possible, to use Price-It as a pricing engine without focusing attention on pricing strategy, which was initially developed using R – a well-known programming language for actuaries. 

Our objective

The primary objective was to ensure that the premium calculations were correct, checking for issues such as missing values or dividing by zero. Accuracy over specific return levels was prioritised, except in extreme cases where reporting was required. The aim was also to produce reports that actuaries could interpret, test integration and evaluate performance to safeguard the efficiency and reliability of the system.

It was important to ensure that customers were paying the right premiums to cover the risk, otherwise the customer’s cash flow could have been affected. To avoid this, in-depth testing of premiums in the pricing engine itself and integration with other systems was required.

Implemented solution

Earnix Price-It – a software platform for advanced analytics and pricing optimisation solution for insurers and financial institutions. It analyses data, models risk and optimises pricing strategies. In the project in question Earnix Price-It was used as a pricing engine for PolicyCenter.

Proposed strategy

Our approach

We started by analysing and understanding what happened in the first release. This included working closely with the PC implementation team and the product team. As a result we reviewed all existing tariffs in Price-It and identified errors and process bottlenecks. Based on that we agreed on acceptance criteria and deployment rules and shared responsibilities across the team.

A conducted analysis resulted in creating a first data dictionary, which formed the core of the testing strategy. It was presented in a workshop and implemented afterwards. The initial testing strategy included in-depth testing at three different levels: analytical environment, staging environment, integration. The testing strategy included all prerequisites, naming conventions and sample test scenarios. Testing was carried out at levels of two Earnix environments and an integration level.

Scope of work

What we did 

• PC combinations test. Objective: to verify that Earnix can handle all allowed PC values.
• Old vs. new system premium test. Objective: to ensure that the customer has a good idea of the change that new tariffs will cause.
• Parameters that shouldn’t’t affect the price test. Objective: to check that certain parameters do not affect the price.
• Regression test. Objective: to confirm that the unchanged tariffs remain the same and to check the percentage difference of the adjusted tariffs.
• Performance test. Objective: to validate that the values generated by the online pricing rules meet the expected time responses.
• Mapping test. Objective: to check that the mapping in PC corresponds to the mapping in Earnix and to test whether the same premium is obtained in both systems.

Results

• The number of errors has been significantly reduced. The starting point was 10 errors for 11 tariffs. Now, with 40 tariffs, SME has reported inadequate prices only twice.
• The number of test cases went from zero to 320.
• Testing has been automated. For 20 tariffs it now takes 40 minutes and is done by one person. Initially, testing was done manually, took two days and required three people. Now automation reduced the testing to 40 minut. A separate testing application is being developed that will reduce the testing time to 15 seconds per coverage.

Interesting facts from the project

This was the first Earnix Price-It implementation project carried out by Sollers.

Earnix representatives verified and validated the solutions used by Sollers during the project.

Did you find this article interesting? Get in touch with our experts!

Lennart Imorde – Head of Process Automation