AI Funnel QA Checklist: 30 Checks Before You Publish

How to run this checklist

Use a clean browser session, a test email address, and a written set of expected routes. Run the checklist on mobile and desktop. Record the funnel version, date, inputs, expected result, actual result, contact data, messages, and final destination.

Do not mark a category complete from visual inspection alone. Submit the paths and inspect the stored and delivered outputs.

• Name one owner for the release decision.
• Freeze the prompt, rules, and content while testing.
• Retest every affected path after a fix.
• Keep screenshots or exports for material failures and resolutions.

Checks 1 to 6: structure and copy

The funnel should make one promise, ask only what it needs, and lead every visitor to a valid end state.

• 1. The opening states the value and expected effort.
• 2. Every question contributes to a decision or required record.
• 3. Answer options are mutually understandable.
• 4. Required fields and optional fields are clear.
• 5. Every call to action names the next step.
• 6. Empty, error, and unavailable states give a recovery path.

Checks 7 to 12: logic, scoring, and outcomes

Use fixed high, middle, low, boundary, and override cases. A score is not validated until the downstream route matches.

• 7. Every branch reaches a valid result.
• 8. Exact lower and upper boundaries select the intended band.
• 9. Hard overrides take priority over weighted totals.
• 10. Contradictory answers follow a documented rule.
• 11. Each result explains why it fits.
• 12. The result, stored band, and next action agree.

Sources: Score-based outcome method

Checks 13 to 18: contact data and email

Follow the same test identity into the contact record and every message. Redact or delete test personal data when the release check is done.

• 13. One submission creates one expected contact.
• 14. Answers, score, outcome, source, and consent are stored correctly.
• 15. The first email matches the visible result.
• 16. Booking or payment suppresses obsolete reminders.
• 17. Personalization has safe fallbacks.
• 18. Unsubscribe, reply handling, and owner notifications work.

Sources: Result-aware email workflow

Checks 19 to 24: accessibility and usability

W3C guidance recommends explicit labels, clear instructions, understandable validation, and meaningful grouping for related controls. Test with keyboard navigation and at least one screen reader, not only automated tooling.

• 19. Every field has a programmatic label.
• 20. Instructions and error messages identify the affected field.
• 21. Related choices use fieldset and legend or an equivalent structure.
• 22. Focus order follows the visual task and focus remains visible.
• 23. Text and controls meet contrast requirements.
• 24. Mobile layouts avoid horizontal overflow and preserve comfortable targets.

Sources: W3C form labels, W3C form validation, W3C grouping controls

Checks 25 to 30: discovery, performance, and measurement

A public landing page should be reachable through standard links, return useful HTML, declare the intended canonical, and load within current Core Web Vitals targets. Private campaign steps may have a different indexing decision, but it must be deliberate.

• 25. The public page returns HTTP 200 without authentication or crawler challenges.
• 26. Canonical, title, description, H1, and indexability agree.
• 27. Internal links use crawlable anchor elements and valid destinations.
• 28. Structured data matches visible content and validates.
• 29. LCP is at most 2.5 seconds, INP at most 200 milliseconds, and CLS at most 0.1 at the 75th percentile where field data exists.
• 30. Analytics events identify start, meaningful steps, result, and final conversion without collecting unnecessary personal data.

Sources: Google crawlable links, Google Article structured data, Core Web Vitals