MarginPilot landing page — from prompt to working page

Prompt → Research → Plan → Conversion architecture → Generated visuals → Final copy → Working HTML page

The prompt

Create a complete high-converting landing page for a fictional B2B SaaS product called MarginPilot — a chargeback automation tool for Shopify and Stripe ecommerce merchants. It helps operators automate dispute response, assemble evidence, keep human review before submission, and measure net recovery instead of vanity win rate. Think strategically: explain the conversion logic behind each section and which objection it handles.

The workstream

Research — competitive landscape + ecommerce chargeback/recovery context
Conversion architecture — section map vs. objections (conversion-copywriter)
Image generation — the page's supporting visuals in one round
Final copy — every section (conversion-copywriter)
HTML build — one self-contained page (frontend-engineer)

Specialists on this run

@plan (planner) · conversion-copywriter · frontend-engineer

@plan assembled the role-fit specialists and routed each step to the right one.

Conversion architecture

It wrote the conversion architecture before any copy — mapping every section to the objection it handles along a “Problem Recognition → Mechanism Revelation → Capability Proof → Decision Commitment” arc — and named the alternatives it set aside.

What it produced

It grounded the work in real research on the competitive landscape and ecommerce chargeback/recovery context, generated the page's supporting visuals, then wrote and built the page. Hero: “Your margins don’t leak during the month. They leak in the 96 hours after a dispute hits.”

Artifacts

Working landing page (HTML)
Generated visuals: dashboard, connected data sources, before/after

A reviewable first draft — not a production guarantee. Every element is AI-generated and can be wrong; verify anything before you ship it.