Insurance

Simplifying insurance with the help of Design Thinking

December 7, 2017
In order to continue its digital transformation, in 2017, Bâloise Assurance decided to incorporate the Design Thinking methodology into its design process for new products. And to try out the method, the IT, Innovation and Marketing teams chose the GoodStart project, the very first 100% online home insurance policy in Luxembourg.

A realistic assessment of the facts

Starting from the observation that 18-35 year-olds account for around 150,000 people in Luxembourg, 41% of whom live with their parents due to the high cost of property, Bâloise wanted to address the other 60% who predominantly rent apartments and for whom online insurance, although commonplace, had until then been unavailable.

But instead of a traditional top-to-bottom approach, the Design Thinking process made it possible to avoid major errors in product placement by removing preconceptions and prejudices with regard to customers, mistakes which a traditional approach would have made it impossible to spot until after the launch.

Cédric Rochet, the Chief Innovation Officer at Bâloise, has led the adventure from its very beginning, and states, "For example, the feeling which was shared within the teams was that price was THE most important issue for young people. However, our interactions with customers on the ground, and our subsequent iterations, taught us that for many of them, the speed and responsiveness of the insurer, together with the simplicity of the product, are at least as important.

The 6 key stages of success

So here are the 6 stages which prevented us from taking the wrong direction:

  • First of all, educate yourselves! In order to be thorough and gain the necessary know-how, we brought some experts in Design Thinking into our premises, in particular Rémi Rivas from the company Ignited Kingdom, one of the French leaders on the topic. After 2 days of training, our mixed team of developers, marketers and project managers were supercharged and were talking of nothing but UX (user experience), UI (user interface), and "iterations".
  • Then, ask the target. Thanks to a good consultation with an auditing firm which is a major employer of young people, and with a Telco player which particularly specialises in newcomers, we carried out several sets of customer surveys, focus groups, user labs, and discussions with people representing our target of YoungStarters. It was there that we realised that price came after practicality.
  • Then, it was time for prototyping in Inno Lab: starting from a mock-up in cut paper, we continued to make progress by means of "Balsamiq" type wireframes, then a dummy integration on several "click dummies", the goal always being to come face-to-face with the future reality as quickly as possible.
  • Finally, during the summer, we took advantage of the lull to carry out our "silent launch"; the purchase of traffic on Adwords and Facebook Ads enabled us to confirm our hypotheses by means of several A/B testing iterations, recorded meetings and physical focus groups and feedbacks. We also regularly faced cold feedbacks from partner start-ups, counting on their views to discover UX concerns.
  • Then, at the very beginning of October, the public Go Live moment arrived. We rolled out the whole of the campaign, but stayed flexible throughout, every blockage, every user problem providing the opportunity to modify the web and mobile interface.
  • And the results are there, the customers too. GoodStart has now become established in the landscape as the first 100% online home insurance policy in Luxembourg and we continue to adjust it in light of the regular analyses and tests to which we subject the interface.

Conclusion: it must be tested by real users, not just at the beginning and end, but regularly throughout the project, by following an iterative 'test & learn' logic. GoodStart, as launched today, does not look like the initial idea of the product… and it is all the better for it, as it has been adjusted to the needs expressed by users, which differed considerably from our first assumptions.