FIRST-LOOK NOTICE: Automated checks find some issues and miss others. This is practical guidance, not an audit or legal advice. A clean scan does not guarantee zero barriers.
HTML Snippet or Page Source
Scan Results & Guidance
What Automated Checks Cannot See
Onceover uses a strict two-stage separation of concerns. All diagnostic detection is 100% deterministic, running standard axe-core rules and pattern matchers inside an isolated sandbox. Artificial intelligence never performs the scan, inspects raw HTML, or evaluates legal compliance; its role is strictly bounded to translating confirmed technical findings into plain-language, practical remediation steps for small business owners.
Automated checks evaluate code syntax and known structural patterns, but cannot determine genuine customer access. Crucial areas that require manual human review include:
Whether an image description is genuinely meaningful or merely present (e.g. filename "photo1.jpg" vs "Warm sourdough loaves cooling on a cedar rack").
Whether a shopper can smoothly tab through checkout, modal dialogues, or menus without getting trapped or losing visible focus.
Whether screen readers announce page sections, specials, and pricing tables in an intuitive sequence that reflects visual grouping.
Directions that rely solely on color, shape, or position (e.g., "click the round orange button on the left to place order").
Whether promotional video captions accurately convey speech, speaker turns, tone, and critical sound cues.
Whether policies, allergy disclaimers, return instructions, and form validation errors are straightforward to read and comprehend.