Communitygithub.com

cro-audit

Evaluate conversion barriers and opportunities on an existing landing page, including goal clarity, friction, competing actions, forms, offer presentation and objection handling. Use when asked for a conversion-rate-optimization audit or conversion-focused review; do not promise uplift.

What is cro-audit?

cro-audit is a Claude Code agent skill that evaluate conversion barriers and opportunities on an existing landing page, including goal clarity, friction, competing actions, forms, offer presentation and objection handling. Use when asked for a conversion-rate-optimization audit or conversion-focused review; do not promise uplift.

Works with~Claude Code~Codex CLI~Cursor
npx skills add https://github.com/ashokgmorais-cmd/Landing-Page-Intelligence/tree/HEAD/skills/cro-audit

Ask in your favorite AI

Open a new chat with this agent skill pre-loaded.

Documentation

Landing Page CRO Audit

Purpose

Run a standalone, evidence-based audit of the CRO module for an existing landing page. The purpose is to diagnose the current experience and recommend practical improvements, not to build a replacement page.

How to run this skill independently

This skill audits only its named module. Do not assume any other module has been run. Begin with the user's request and supplied materials. If essential context is missing, ask only for information that materially affects this module; do not block on analytics or details that are unavailable. If the page URL is inaccessible, request screenshots, a recording, exported page text, or HTML. Work within the evidence available and label the scope limitation.

Useful context, when available: business/brand and industry; target audience and geography; page URL or assets; product/service and business model; primary conversion goal; offer/price; campaign or traffic source; brand constraints; relevant competitors; and analytics/testing data. Mark missing data Not provided or Unavailable rather than guessing. Distinguish business-provided claims from your own page observations.

Evidence and integrity rules

  • Separate Observed (directly visible/tested), Reported (provided by the business), Inference/Hypothesis (plausible but unverified), and Recommendation.
  • Tie important findings to a page section, element, source URL, supplied asset or test result. Quote exact copy only if it is available; otherwise label a paraphrase.
  • Explain what is present, why it matters to the relevant user/task, and what to change. Avoid treating every weakness as a proven conversion loss.
  • Do not invent business facts, results, analytics, guarantees, user research, certifications, or test outcomes. Avoid causal and uplift claims unless supported by valid experiments.
  • If only static screenshots or text are available, do not claim to have tested links, forms, responsive behavior, performance, keyboard access, or assistive technology.
  • Consider audience, offer, conversion goal, traffic context and accessibility needs where supplied. State assumptions and material limitations.

Priorities, confidence and scoring

Use priority P0 Critical (serious broken/usability/trust issue requiring immediate attention), P1 High (important, potentially consequential), P2 Medium (valuable after higher priorities), P3 Low (minor refinement/longer-term). Use confidence High (directly observable/measured), Medium (reasonable evidence with gaps), or Low (limited evidence or substantial assumptions).

Score applicable criteria from 0–5: 0 missing/broken/critically deficient; 1 very poor; 2 below expectations; 3 acceptable; 4 very good; 5 excellent. Mark unassessable items N/A and exclude them from the denominator. Calculate the module score as (points earned ÷ maximum applicable points) × 100, rounded to a whole number; state the numerator/denominator or list scored criteria. If nothing can reasonably be scored, report Not Evaluated, not 0. Do not treat this diagnostic score as an industry benchmark or a forecast of conversion performance.

Output format

Write a self-contained, client-ready Markdown audit with a title, audit scope/evidence basis, concise summary, strengths, prioritized findings/recommendations (prefer a table where useful), module score, overall evidence confidence, and limitations/next evidence needed. Keep advice specific and actionable. Do not produce a full multi-module report or score unless the user explicitly asks for one.

Module-specific review checklist

Assess clarity of the primary conversion goal; competing actions and distractions; five-second comprehension; relevance to the intended audience; friction in forms, checkout, enquiry and next steps; offer clarity; objection handling; decision anxiety; trust in the conversion path; and unnecessary cognitive effort.

Required deliverable

Provide conversion barriers and opportunities, each tied to observed/reported evidence, visitor impact, practical change, priority and confidence. Include a focused score and identify validation needs. State that proposed changes are hypotheses until tested; never predict conversion gains as facts.

Module-specific guidance

Distinguish between a plausible conversion risk and a measured loss. Inspect steps, fields, error states and downstream destinations only when accessible; otherwise label them unverified. If traffic or baseline data is absent, do not estimate impact numerically.

Strategic diagnostic questions

  • What is the single primary conversion action, and does the page consistently support it?
  • Which barriers are directly observed versus plausible explanations requiring analytics, session recordings, interviews, or experiments?
  • Is the friction caused by unclear value, the commitment asked, missing reassurance, effort, or a broken/unknown step? Avoid treating all friction as the same problem.
  • What is the smallest change that could address the suspected root cause, and how could the business falsify the hypothesis?

Final quality check

Before delivering, confirm that the review stayed within this module, every material conclusion is evidence-labelled, all recommendations are actionable and appropriately prioritized, unknowns remain unknown, N/A items were excluded from scoring, and the score calculation is correct. If the user asks for a full 12-module audit, explain that this skill covers one module and can be run alongside the other skills in this repository.

Individual skills in this repo

This repo contains 4 individual skills — each has its own dedicated page.

Related Skills