Turning a landing page into a product
Frank Insurance is a digital insurance broker operating in Thailand and Taiwan, acquired by Bolttech. I joined at a very early stage, when it was still a small insurtech startup planning its expansion across Asia. The goal was to create a digital-first insurance experience that felt simpler and more trustworthy than traditional brokers.
As part of the founding design team, I worked closely with product and engineering to turn an initial marketing landing page into a real product. My focus was not visual polish, but defining product direction, user journeys, and a design approach that could scale while allowing fast experimentation.
Too many steps before any value
The experience was designed as a marketing flow rather than a product journey. Users had to complete multiple steps and provide extensive information before seeing any real value, such as pricing or coverage options. This created friction, reduced trust, and led to high abandonment.
My first assumption was that we could reduce the number of interactions before showing a price. Inspired by e-commerce, where users see a product and its price early, I applied the same mental model to insurance to better match user expectations.
“The core issue wasn’t visual design, but a misalignment between user intent, information architecture, and the complexity required to generate a quote.” This became the starting point for iteration. This became the starting point for iteration.
Designed for one hand and half the attention
Before refining visual decisions, we focused on defining the UX foundations of the product. The main challenge was structuring the flow and collecting the required information in a way that felt simple and trustworthy, while still supporting accurate insurance quoting.
I started with early wireframes to map user journeys and explore different ways of gathering data, helping shift the experience from marketing-led to product-first.
I designed user flows with a very practical scenario in mind: someone commuting, using their phone with one hand, potentially distracted and short on time.
This meant prioritising clear hierarchy, reachable interactions, and progressive disclosure of information, ensuring the experience felt lightweight and manageable even in less-than-ideal conditions.
After several iterations we ended up with three main user flows.
We organised a focus group with Thai users to test early concepts and flows with real customers in our target market.
Sessions were fully recorded and translated with subtitles so the entire team could clearly understand the feedback. This helped align product, design, and engineering around real user behaviour rather than assumptions.
Everything testable, nothing final
Even at high fidelity, everything I designed was treated as testable, not final. Key UI decisions were intentionally created with multiple variations, allowing us to validate assumptions through A/B testing rather than relying on personal preference.
We introduced a new typeface for hierarchy and readability, a horizontal logo for digital layouts, and a refined palette without the near-red tone, so the brand felt more consistent and trustworthy. A new iconography system improved comprehension of coverage details and made navigation more intuitive.
A more modern, cohesive interface, clearly positioned as a digital product, not just a marketing extension.
The same principles still run today at bolttech.io, where the platform story is told across product, programs and use cases.
Evidence over assumptions
Working at Frank taught me the importance of separating assumptions from evidence, especially in early-stage products. Strong design is not about jumping to visuals, but about asking the right questions first, understanding user behaviour, and validating ideas as early as possible.
It also shaped how I think about progressive disclosure and friction. Reducing interactions before showing value had a direct impact on engagement and aligned the product with familiar e-commerce mental models.
Operating in a mobile-first, cross-cultural context showed how critical it is to design for real-world constraints: one-handed use, short attention spans, and varying levels of trust in digital services.