Redesigning a campaign landing page to make insurance feel human — and turning a 4% CTR into 14% through data-led iteration.
Context
The campaign landing page was visually dated — heavy with content, cluttered in layout, and built at a time when the priority was information density, not conversion. The numbers reflected it: a 4% CTR, an average session time of just 55 seconds, and a bounce rate sitting at 36%.
Users were arriving with real intent — these were paid campaign visitors — and leaving before the form even registered. The problem wasn't the product. Insurance is needed. The problem was that the page failed to make it feel relevant, accessible, or trustworthy in the time it had.
Problems Identified
We started with a data audit before touching a single wireframe — session recordings, heatmaps, funnel drop-off data, and a heuristic review of the existing page. Four clear problems emerged.
The copy led with policy terms, coverage clauses, and product names. Nothing on the page connected to how ordinary people actually think about money, risk, or their family's future. The language was written for agents, not for someone scrolling on a phone.
A single-page, centre-modal form presenting every field at once. No progression, no context, no feedback during fill. Asking for phone number and OTP upfront created immediate distrust. Completion rate was 4% — most users saw the form and left.
The primary call to action wasn't sticky, wasn't prominent in the visual hierarchy, and competed with secondary links and navigational elements. On mobile, users had to scroll significantly just to find where to act.
The page was a monolithic design — you couldn't swap sections, test variations, or personalise for different campaign sources. Every iteration meant rebuilding from scratch, which meant iteration barely happened.
The Central Idea
The design concept started with a single question: what if insurance didn't feel like a financial product? What if it felt like something you already understand?
"A health cover starts at less than a pizza a month." The pizza is highlighted. Not because it's clever — because it's true, and suddenly insurance stops being abstract.
We built the communication architecture around everyday reference points. Each section of the page was designed to demystify a specific anxiety — cost, complexity, claim hassle — by replacing jargon with something relatable. The visual design followed: warm, human, accessible. Not corporate blue everywhere. Not policy documents dressed up as marketing.
This wasn't just copywriting. It was information architecture. We restructured the entire page hierarchy around what a first-time buyer actually needs to feel before they'll trust you with their phone number.
What We Built
Instead of one fixed page layout, we designed the landing page as a system of modular sections — hero, trust strip, plans, form, testimonials — that could be reordered, swapped, or removed without rebuilding. This is what made iteration fast. When heatmap data told us something, we could act on it in the next publish, not the next project.
We rebuilt the form from a wall into a conversation. Instead of every field at once, each step asks one thing and responds intelligently. Enter a location → nearby cashless hospitals appear in real time. Enter a sum insured → a dynamic suggestion shows what's ideal for your profile. The form became a tool, not a gate. We also tested phone + OTP first vs. last, and focused single-page flow vs. hero-embedded form — shipping the version that cleared each A/B test.
Every section of the page had one action it was trying to drive. The primary CTA was made sticky on scroll so it was always visible on mobile without scrolling back. Secondary actions were visually deprioritised. The result: users knew exactly what they were supposed to do next at every point on the page.
The client was running on an older tech stack — high-end graphics and animations were off the table. Rather than accept slow load experiences, we designed skeleton loaders that displayed short, conversational insurance-related text while content loaded. Phrases like "Did you know? 1 in 4 hospitalised patients pays out of pocket." The constraint became a micro-engagement touchpoint.
How We Iterated
We followed a structured but adaptive loop: data audit → wireframes → design → A/B test → publish → measure → repeat. No phase was final. No decision was sacred.
What made this work was that we tracked everything from day one. After the first iteration shipped, heatmap data showed users were spending disproportionate time engaging with the plans comparison section — more than any other part of the page. It was buried in the third section.
Heatmap finding: Plans section had the highest engagement on the page, but it was the 4th section users reached. In the next publish, we moved it to position 2 — immediately after the hero. Session depth increased and form starts increased in the same cycle.
We ran multiple form variations in parallel: a focused full-screen flow, a hero section with the form embedded directly, and a modal triggered by CTA. Each was tested internally first, then shipped to a traffic segment. The hero-embedded form with contextual feedback won — and became the production version.
Tech constraint, design solution: The client's legacy stack couldn't support animated or heavy graphic elements. We didn't fight it — we built leaner, and designed the skeleton loaders to carry conversational value. The page actually felt faster as a result.
Outcomes
The plug-and-play component system became the template for the client's subsequent campaign pages — other product verticals used the same architecture and form framework. The progressive form pattern, the contextual feedback model, and the skeleton loader approach were all carried forward.
Worth noting: even with significant technical constraints on the client side that limited graphic fidelity, we still moved every metric meaningfully. Simplified communication and reduced form friction mattered more than visual production quality.
Confidential Work
This project is under NDA. The case study above reflects the full thinking — all actual wireframes, design iterations, A/B test variants, heatmaps, and before/after screens are in a private presentation I can walk you through directly.
Request full case study