Harley Alexander


Case Study

Key Skills

  • User Research
  • Conversation Design
  • Chatbot development
  • React/Redux
  • Serverless
  • API & Integrations

Health.com.au are committed to putting their customers on the policy that best suits them – not what makes the most money. This starts at the beginning of their journey when a customer gets a quote. We researched, designed and created a new way for customers to get the right policy at the right price.

Health’s target demographic is tech savvy, 18-30 year olds who want to cut through the crap. In their words, health want to be “simply helpful”. Facebook Messenger presents a new sales channel that fits this demographic. This was the perfect way to disrupt a stagnant industry with careful conversational design and leveraging new technologies.


Our user research showed us two clear personas: those trying to avoid penalties at tax time (cheapest) and those who were looking for particular coverage (specific).


We workshopped potential conversation flows with the client mapping out how phone conversations go. It looked something like this:

Regardless, we ran a few user testing sessions with a prototype mocked up in bot society. The results were what we knew already; this was long-winded and unnecessarily complicated.

Based on our research, we knew that this could be simplified. Messenger provides us with a suite of tools to create rich, engaging experiences. We ended up with a much simpler flow:

Further user testing confirmed that this flow was much simpler and much more satisfying from the user’s perspective. They felt they understood the cover, what they were purchasing, and no time was wasted.


The last phase of design is scripting. We’re huge on the importance of script and personality within a bot. It’s what can make or break an engaging experience. We worked with health to produce microcopy that was both entertaining and spoke to the brand values


Time to build.


We leveraged the awesome platform meya.ai to expedite our development process. Meya provides the building blocks such as conversational flow, bot API, integrations and user storage – so you can focus on the important things.

Initial stack planning. Note the MVP area. We also made several tech decisions to move closer to AWS than the GCC stack.

That’s just one piece of the puzzle though. We built out three other core components of the chatbot in-house; a serverless API, a micro-interaction react app and several python components to tie them all together.

The serverless API was designed with flexibility in mind. A requirement is, obviously, to tie into health’s own APIs to provide product and pricing data. It also has to verify that requests are indeed coming from Messenger to provide contextual information. The API was designed to be able to drop in ‘adapters’ that would allow any API change, should the company update their stack in the future.

Chat interactions are designed to be short, sharp and to the point. To supplement the conversation, we designed and built a handful of these interactions to seamlessly integrate into the experience.

The result? You can sign up for health insurance entirely from inside Messenger, where you spend most of your time anyway.