The Problem
The product was leaking customers at every stage
A digital car insurance platform in the UAE was struggling with customer acquisition and conversion. The product existed, the pricing was competitive, but users weren't converting, and the team didn't have a structured diagnosis of where and why the leaks were happening.
I applied the Flywheel Model, a framework that evaluates a product across three phases: Attract (are you visible?), Engage (is the experience smooth?), and Delight (do customers want to stay?), to systematically audit the entire customer journey.
What I found was a product bleeding users at every stage of the flywheel.
The Diagnosis
Three broken flywheel stages, five specific pain points
Attract: Invisible where it matters. Searching "best car insurance in UAE" returned zero results for the platform. No blog posts, no articles, no SEO presence. The product simply didn't exist in the discovery phase of the customer journey. Competitors were capturing all the top-of-funnel intent.
Engage: Death by a thousand clicks. This is where the deepest problems lived:
- Information overload: The landing page tried to say everything at once, creating cognitive overload rather than guiding users toward a decision
- Colour psychology misuse: Excessive yellow throughout the UI created a sense of caution rather than trust: the opposite of what insurance needs to convey
- 21-click checkout: It took 21 clicks to reach the cart page. Competitors achieved the same outcome in 16 clicks or a simple 4-step process. The journey started with car model questions, then abruptly asked for them again after entering customer details: a broken information architecture
- Buried USPs: Key differentiators (unlimited roadside assistance, 60-minute policy issuance, free registration service) were hidden in the noise instead of leading the narrative
Delight: No reason to return. No quick online claim settlement, no automated support, no mobile app. Once a user purchased a policy, the relationship essentially ended until renewal time.
The checkout flow wasn't just long. It was architecturally broken. The journey asked for car model details at the start, then re-asked about the car trim after collecting customer information. This wasn't a cosmetic problem. It was a fundamental information architecture failure that eroded trust mid-funnel.
The Solution
Rebuild the checkout as a 4-step flow
I focused on the highest-impact intervention: redesigning the checkout flow by restructuring the information architecture. The approach was to club related questionnaires together and eliminate redundancy:
- Step 1: Vehicle details: All car-related inputs consolidated into one screen
- Step 2: Customer details: Personal and contact information
- Step 3: Quotes: Side-by-side quote comparison
- Step 4: Checkout & payment: Final review and transaction
I designed the wireframes in Figma and proposed validating the new flow through a 14-day A/B test: treatment group sees the streamlined 4-step flow, control group sees the existing 21-click flow. Primary metrics: conversion rate, drop-off rate. Secondary: transaction success rate, average checkout time.
Additional recommendations across the flywheel:
- Attract: SEO content strategy, competitor comparison pages
- Engage: Updated colour palette (trust-oriented blues/greens), prominent USP callouts, car model auto-suggestions from usage data
- Delight: Self-service digital claim settlement, dedicated claim concierge, mobile app