Auto Insurance Platform — From zero to validated prototype
Full redesign of an insurance platform with a high abandonment rate and a new channel via WhatsApp chatbot. Deep discovery with 16 interviews, in-person shadowing, and two co-creation workshops resulting in a 3-step quote journey.
Data shown is fictional and has been altered to preserve project confidentiality.
Context
A platform with high abandonment and 100% manual lead capture
Stellantis needed to build a new auto insurance platform from scratch — capable of serving multiple profiles in a single cohesive journey: end customers, dealership salespeople, and commercial assistants. Beyond the web channel, the scope included a WhatsApp-integrated chatbot as an alternative entry point for quotes.
The legacy platform concentrated several problems: complex flows, high abandonment rate, and an interface that didn't meet the experience expected by car buyers. Internally, lead capture was 100% manual — spreadsheets and unintegrated tools, with all management centralized in commercial assistants, creating bottlenecks and lost opportunities.
40% of dealerships had their own brokerages, competing directly with Stellantis's internal offer and fragmenting the customer experience.
16
Interviews conducted
8
Players benchmarked
14h+
Hours of interviews analyzed
Research
Interviews, in-person shadowing, and full chain mapping
The discovery process was built to understand all three sides of the problem: the customer buying insurance, the salesperson offering it, and the assistant operating internally. No friction point could be ignored.
01
Qualitative interviews — 16 participants
Interviews with brokers, dealership salespeople, and end customers. Over 14 hours of material analyzed to identify behavioral patterns, purchase motivations, and key abandonment points in the quote journey.
02
In-person shadowing — Dealership in Rio de Janeiro
In-person visit to a dealership in Rio de Janeiro to observe the real insurance sales flow. Direct observation of the salesperson-customer interaction, mapping the moments when insurance was — or wasn't — offered during negotiation.
03
Service Blueprint — AS IS Mapping
Service Blueprint built from internal interviews. Full mapping of online and offline journeys, highlighting operational bottlenecks, dependence on manual processes, and friction points across the entire value chain.
04
Benchmarking — 8 players analyzed
Competitive analysis with direct, indirect, and international players: Auto Compara, Minuto Seguros, Youse, PIER, Porto Seguro, and Lemonade. Evaluation of quote journey, entry form, coverage customization, and results presentation.
Mapeamento AS IS — Jornada de cotação
User flow — legacy platformAS IS
01
Accesses the platformNo clear entry point
02
Confusing home screenNo visual guidance1st abandonment point
03
Extensive formVehicle data filled manuallyHigh friction
04
Coverage without pricingBlind choice, no contextOpacity
05
Results listNo visual comparison between optionsHard decision
06
AbandonsOr calls an external brokerConversion loss
Neutral step
Pain point
6 steps · High abandonment rate
AS IS user flow — 6 steps with multiple abandonment points
Insights
What the interviews revealed about each profile
The data collected revealed distinct patterns for each profile, with direct implications for platform design. No insight was treated in isolation — all of them informed product decisions.
Customers
50%
already had a trusted broker by referral — even before arriving at the dealership.
70% did not receive an insurance offer during financing. Price was the decisive factor for switching brokers.
Salespeople
High turnover
at dealerships made it difficult to spread the value proposition. Insurance was perceived as disconnected from the salesperson's role.
Internal processes
100%
of lead capture was manual. Unintegrated spreadsheets and tools. Management centralized in assistants, creating constant bottlenecks.
The platform's main competitor wasn't another insurance product — it was the inertia of the internal process, which made insurance invisible at the customer's moment of decision.
Co-creation
Two workshops to turn data into product vision
Research insights needed to be transformed into product decisions. The workshops brought together business stakeholders, product team members, and specialists to align vision and priorities before prototyping began.
WS 1
In-person workshop — Building the product vision
9 participants · 7 hours · In-person
Structured session to consolidate research learnings into a shared product vision. Opportunity mapping, design principles definition, and prioritization of features for the new platform.
WS 2
Remote workshop — Chatbot features
10 participants · 2 hours · Remote
Specific focus on the WhatsApp channel. Over 30 feature ideas mapped and prioritized. Definition of chatbot scope, priority conversation flows, and handoff criteria to human support.
Solution
3-step quote, automatic fill, and comparable results
The solution was built from the largest friction points identified: a journey that was too long, manually filled vehicle data, and results that were hard to compare. Every design decision directly addresses a problem from the research.
Redesigned insurance platform — high-fidelity prototype with MUI library
Data shown is fictional and has been altered to preserve project confidentiality.
Key changes
↗
3-step journey: quote simplified to the minimum, reducing the number of screens and decisions required to reach results.
↗
Automatic fill by plate: vehicle data automatically filled from the license plate, eliminating the main friction point at the start of the flow.
↗
Coverage customization with dynamic pricing: the user adjusts coverages in real time with immediate feedback on the quote value, making pricing transparent and the decision more informed.
↗
Comparable insurer cards: results presented in visual side-by-side cards, making it easy to compare options without leaving the screen.
Before & After — Quote flow
User flow — legacy platformAntes
01
Accesses the platformNo clear entry point
02
Confusing home screenNo visual guidance1st drop-off
03
Extensive formVehicle data manuallyHigh friction
04
Coverage without pricingChoice without contextOpacity
05
Results listNo visual comparisonHard decision
06
AbandonsOr calls an external brokerConversion loss
Neutral step
Pain point
6 steps
Quote flow — new platformDepois
01
Enter the plateVehicle data filled automaticallyNo friction
02
Choose coveragesDynamic pricing updated in real timeTransparency
03
Compare and chooseVisual side-by-side insurer cardsInformed decision
✓
PurchaseConversion in-channel, no intermediaries+conversão
Resolved step
Journey reduced from 6 to 3 steps
Before & After — Results screen
Data hidden to preserve company confidentiality.
Data shown is fictional and has been altered to preserve project confidentiality.
Mobile version
Mobile version — responsiveness of the quote journey
Data shown is fictional and has been altered to preserve project confidentiality.
Results
A journey that converts where the previous one dropped off
The prototype was validated internally with the product team and went through rounds of adjustment before final delivery. Expected results were defined based on problems mapped in the research and alignment with business stakeholders.
✓
Abandonment reduction: a 3-step journey with automatic fill reduces cognitive load and the number of drop-offs during the quote flow.
✓
WhatsApp as an entry point: the chatbot opens a new capture flow in a channel where customers already are — without needing to access the web platform directly.
✓
More informed decision: visual coverage comparison with dynamic pricing returns control of the choice to the customer, reducing the opacity of the process.
✓
Wire-flow for development: complete flow documentation with annotations for the engineering team, ensuring continuity and fidelity in implementation.
Project delivered with high-fidelity prototype (desktop and mobile), development wire-flow, and internal validation with the product team — a solid foundation for implementation.
Prototype video
Data shown is fictional and has been altered to preserve project confidentiality.
