Work / will bank
While Disney+ was in development, I kept looking at the future of insurance. Our research said our customers' pains weren't the ones our insurance covered. So I let them build their own insurance, and watched what they picked.
At will bank I led design for commissioned products: insurance, mobile top-up and capitalization. Card insurance with Disney+ had just proved that our customers responded to the right offer. But it was still an old product with a new hook.
The goal was not just to launch a product, but to build a data ecosystem for future insurance offers. To get there, I organized the team's discovery in levels.
We ran in-depth interviews with simple account customers, a strategic focus for the bank, to map the pains of people with no insurance. For customers who already had insurance, I built a JTBD bot with n8n and Figma Make that asked why they kept it. A second round of interviews went after their anxieties and the unexpected.
The closer to the center, the more it weighs. Family came first: 65% of the customers we heard were heads of household and 61% lived with children.
All of it adds up to one idea the team used as a compass: insurance as a tool to live well, not an end in itself.
If our insurance didn't cover what people feared, what would they pick themselves?
I used the same no-code fake door logic from the Disney+ test. A minicard on the Home screen led to a Figma Make prototype where people built their own insurance, choosing coverages one by one. Clarity recorded the sessions and Supabase stored every pick, price and drop-off.
I grouped eleven coverages by what they protect, and wrote the copy in our customers' own language: for your hustle (broken or stolen phone, motorcycle, work tools), for your family (family protection, funeral support, pharmacy allowance, health discounts, telemedicine) and for your pocket (daily pay on sick leave, scam protection).
25,195 people opened the prototype from Home, and 6,893 went through the whole journey. The URL told me which of four groups each one came from: simple or full account, and whether they had clicked an insurance offer before.
"Select the protections that make sense for your life and create a personalized insurance that fits your pocket."
Each group opened into its coverages, each with its own monthly price, from R$3.81 for funeral support to R$49.90 for the motorcycle.
Behavior confirmed the interviews. People went straight for income and scams, well ahead of the phone or the motorcycle.
Times each coverage was picked in the test
117 people combined daily pay on sick leave with scam protection. 42 built a family set, and 35 a work set with phone, motorcycle and tools.
Pharmacy allowance and health discounts let people add dependents. 156 did, mostly spouses (162) and children (49). Insurance stops being an invisible expense when the whole family can use it.
Times each perk was picked, more than one allowed
Simple account customers who had already clicked an insurance offer were the clear early adopters: 14.6% signed up, and they built more expensive insurance, R$35.56 a month on average. Full account customers dropped off more than 98% of the time and built cheaper insurance, around R$26.
6,893 valid sessions, people who went through the whole journey
The research said health and income. The test showed the same thing with behavior, and gave us the prices, bundles and audience for the next insurance.
The plan was ready when the Central Bank of Brazil liquidated the bank. The fake door and the research bot became the benchmark for other teams at the bank.