What it does
Reads step-by-step funnel data, for example product page to cart to checkout to purchase, and identifies the single step with the largest percentage drop-off, then cross-references that step against page load time, form length, or known UX issues to suggest a likely reason. Presents the whole funnel described in plain words with percentages at each step, not just the final conversion rate.
Install
This skill isn't packaged in an installable registry. It's a plain SKILL.md file you can copy directly. Copy it from the SKILL.md tab and save it at the path for your assistant:
Use it in your AI assistant
Claude Code
.claude/skills/conversion-funnel-reporter/ (or ~/.claude/skills/conversion-funnel-reporter/ for all projects)
GitHub Copilot (VS Code)
.github/skills/conversion-funnel-reporter/
OpenAI Codex
.codex/skills/conversion-funnel-reporter/
Cursor
.cursor/skills/conversion-funnel-reporter/ (or ~/.cursor/skills/conversion-funnel-reporter/ globally)
Compatibility
Requires a connected analytics or e-commerce plugin that tracks funnel steps, such as WooCommerce Analytics or Google Analytics goals.
The complete SKILL.md for this skill:
---
name: conversion-funnel-reporter
description: Use when a user wants funnel analytics turned into a plain-English drop-off report, or asks "where are people dropping off" or for a "funnel report".
---
# Conversion Funnel Reporter
## When to use
The user has step-by-step funnel data and wants to know in plain language where people leave the funnel and why, not just the overall conversion rate.
## Procedure
1. Pull step-by-step funnel data for the requested period and list the percentage that continues at each step.
2. Identify the step with the largest single drop-off, that's the priority, not just the overall conversion rate.
3. Cross-check that step against known friction signals, page load time, number of form fields, mobile versus desktop split, to suggest a likely reason.
4. Compare against the prior period's funnel to note whether the drop-off is new or a longstanding pattern.
5. Summarize in plain language where people leave, roughly how many, and the most likely reason, before any recommendation.
## Guardrails
- Don't recommend a specific UX fix as guaranteed to work, frame it as a hypothesis worth testing.
- Be explicit when the sample size at a given step is small enough that the percentage might be noisy.