UIPrompt Library
Start Pro $14Account
Back to Income Lab
goalDraftReady now

Conversion-audit service

Captures current evidence, identifies friction, and produces five prioritized recommendations with explicit limits.

Setup capital
Under $100
Monthly cost
$0–50
Setup time
3–6 hours
Weekly effort
5–10 hours
First dollar
1–4 weeks
Sales burden
Written outreach

User-selected scenario, not an expected outcome.

Every consequential external action pauses for explicit human approval.

Autonomy by business phase

Current level 3/5. Potential level 4/5 in 1–3 months.

1

Validate

AI-assisted

AI inventories evidence; the human confirms the business goal.

2

Build

AI-executed with approval

AI drafts findings and prioritization.

3

Acquire customers

AI-executed with approval

Outreach is drafted but never sent automatically.

4

Deliver

AI-executed with approval

A human reviews the audit before client delivery.

5

Collect and operate

Human-led

A human owns invoicing and client follow-through.

AI responsibilities

  • Capture observable evidence
  • Find friction
  • Prioritize recommendations
  • Draft the report

Human responsibilities

  • Confirm access
  • Interpret business context
  • Approve delivery
  • Avoid unsupported performance promises

Approval gates

Approval applies to the exact action once. It is not blanket permission.

Explicit approval

Outreach

Approve every service pitch before sending.

Acquire customers
Explicit approval

Use client data

Confirm authorization before viewing analytics or private data.

Validate
Explicit approval

Publish

Approve the exact audit before client delivery or public sharing.

Deliver
Explicit approval

Invoice

The human reviews and sends the invoice.

Collect and operate
Runnable goal

Copy the complete asset

/goal

Goal: produce a bounded, evidence-linked conversion audit for one approved page or flow. First confirm the business goal, intended user, primary conversion, access authorization, baseline evidence, and exclusions. Capture only observable or authorized evidence across message clarity, trust, information order, interaction friction, accessibility, mobile behavior, performance, and measurement gaps. Never invent analytics or promise conversion lift.

Create an annotated issue inventory with evidence, severity, affected user, and confidence. Select exactly five recommendations and rank them by expected decision usefulness, implementation effort, reversibility, and evidence strength. Provide proposed copy or interaction changes only where the evidence supports them. End with a client-ready brief, limitations, and a measurement plan. Wait for explicit approval before outreach, client-data access, delivery, public sharing, or invoicing. Stop on unauthorized access, missing evidence, regulated claims, or any request for guaranteed results.

MILESTONES
1. Authorize and baseline: Goal, user, conversion, access, and available evidence are recorded.
2. Audit friction: Issues include screenshots or direct observations and confidence.
3. Prioritize: Five recommendations have rationale, effort, and limits.
4. Prepare handoff: Report and measurement plan pass human review.

COMPLETION CRITERIA
1. Every finding links to evidence
2. Exactly five prioritized recommendations
3. No invented analytics or lift promise
4. Human-approved client handoff

BUDGET CEILING
$50 USD

EXTERNAL-ACTION GATES
1. Outreach — Explicit approval: Approve every service pitch before sending.
2. Use client data — Explicit approval: Confirm authorization before viewing analytics or private data.
3. Publish — Explicit approval: Approve the exact audit before client delivery or public sharing.
4. Invoice — Explicit approval: The human reviews and sends the invoice.

STOP CONDITIONS
1. Unauthorized access
2. Insufficient evidence
3. Unsupported performance claim
4. Budget ceiling reached

Evidence record

Projected scenarios never appear here as observed outcomes.

Evidence stateDraft

Goal has not completed a recorded audit run.

Experiment count0 recorded

No model test date recorded

Supported modelOpenAI GPT-5

Catalog baseline; exact version must be recorded at first test

Stop and escalate

  • Site or analytics access is unauthorized
  • Evidence is insufficient
  • A recommendation requires an unsupported lift claim
  • Privacy or security issue
  • Regulated industry claim
  • Contract or accessibility-liability question

Known failure modes

  • Generic checklist output
  • Invented analytics
  • Claiming guaranteed conversion lift

Prerequisites and changelog

A freelancer or agency productizing a bounded website audit. · Produce a defensible, client-ready conversion audit without promising a lift.

Before you start

  • Permission to inspect the site
  • Defined conversion job
  • Any analytics must be client-authorized
  • UX analysis
  • Copy review
  • Client communication

Changelog

  • 1.0.0 · 2026-07-12Initial launch workflow.