Checkout Barrier Index

How hard real online stores are to use with a screen reader or a keyboard on the way to checkout, measured with automated accessibility checks and a keyboard-only walk on each product page. Stores are reported by platform, never named.

Crawl finished Nov 8, 2026, 10:07 PM UTC · analysed at commit 0406ebe

Open the results as a dashboard

Headline results

• Median axe-core violations per product page (distinct WCAG A/AA rules failed)

  4

  Middle half of stores (interquartile range): 2–6, n = 60

  How this was counted

  axe-core 4.14.0, tags wcag2a, wcag2aa, wcag21aa, wcag22aa, one product page per store. Median failing elements per product page: 13. Home page median: 3 rules.

• Stores where keyboard-only traversal could not reach add to cart within 150 Tabs

  9% 3 of 33

  95% Wilson interval: 3%–24%, n = 33

  How this was counted

  Tab only from the top of the page, no Enter, Space or clicks; Escape pressed once if focus was trapped. Counted over product pages with an enabled add-to-cart control named "add to cart/bag/basket" or "buy now". Focus traps among the failures: 1. Not in the denominator: 12 pages where a cookie or sign-up dialog held focus and Escape did not close it, 7 whose add-to-cart button is disabled or hidden until a size is chosen, 8 with no control using that wording, 0 traversal errors.

• Stores where Tab alone could not reach add to cart, counting a dialog that held focus as blocked

  33% 15 of 45

  95% Wilson interval: 21%–48%, n = 45

  How this was counted

  The same traversal as the row above, but the 12 pages where a cookie or sign-up dialog held focus and Escape did not close it count as blocked instead of being left out: a keyboard or screen-reader user meets that dialog first. Added on 9 November 2026 after the first analysis, from the same crawl; no store was crawled again.

• Stores with at least one button that has no accessible name (home or product page)

  35% 21 of 60

  95% Wilson interval: 24%–48%, n = 60

  How this was counted

  axe-core rules button-name, input-button-name and aria-command-name. A screen reader announces these as "button" with nothing else.

• Most common failing rule: color-contrast

  60% of stores (36 of 60)

  95% Wilson interval: 47%–71%, n = 60

  How this was counted

  Elements must meet minimum color contrast ratio thresholds. Counted once per store if it fails on the home, product or cart page; 697 failing elements in total. Tied on store count with target-size (480 elements); ranked first by failing elements.

• Sampled stores that refused or timed out for an honest, rate-limited, read-only crawler

  24 of 100

  95% Wilson interval: 17%–33%, n = 100

  How this was counted

  Blocked at the home page: 21; at the product page: 3. Also excluded: 0 disallowed by robots.txt, 16 with no product page found from the home page, 0 crawler errors. Blocked stores are left out of every other number.

By store platform

The most common failures

1. color-contrast: Elements must meet minimum color contrast ratio thresholds. Found on 36 stores.
2. target-size: All touch targets must be 24px large, or leave sufficient space. Found on 36 stores.
3. link-name: Links must have discernible text. Found on 35 stores.
4. button-name: Buttons must have discernible text. Found on 20 stores.
5. image-alt: Images must have alternative text. Found on 15 stores.
6. nested-interactive: Interactive controls must not be nested. Found on 14 stores.
7. listitem: <li> elements must be contained in a <ul> or <ol>. Found on 13 stores.
8. list: <ul> and <ol> must only directly contain <li>, <script> or <template> elements. Found on 12 stores.
9. meta-viewport: Zooming and scaling must not be disabled. Found on 12 stores.
10. aria-required-children: Certain ARIA roles must contain particular children. Found on 10 stores.

How Sightline avoids them

Sightline never asks a blind shopper to use the store’s pages, so none of these failures sit on its path: product facts come from structured catalog data, the readback is spoken text, and payment happens through PayPal with a spoken confirmation.

Limits of this measurement

• Convenience sample: 100 well-known stores selling everyday goods, picked from public knowledge and platform showcase pages, then grouped by the platform their pages actually showed. Not a random sample of all online stores, and platform groups are small, so per-platform numbers are indicative only.
• Crawled between 2026-11-08 20:14 and 2026-11-08 22:07 UTC from a single location in India with headless Chromium at 1280×900. Stores change their pages often, and geo redirects, currency conversion and consent banners differ by country: several large US retailers answered with a country selector or a global corporate page, which shows up here as no product page found.
• Stores that timed out or gave no product page on the first pass were crawled once more after product-link discovery was widened (a second category page and flat product URLs). Stores already measured were not crawled again, so a fresh run of the current code can pick different product pages.
• 24 stores refused or timed out and are excluded, so the measured set leans toward stores without aggressive bot defences.
• Automated rules catch only part of WCAG. Zero axe violations would not mean a page is accessible; these counts are a floor, not a full audit. No screen reader was run.
• One product page per store, found from links on the home page (or up to two category pages). It may not represent the rest of the catalogue.
• Keyboard traversal presses Tab only (plus Escape once when trapped). It never presses Enter, Space or arrow keys, so it does not open menus, choose a size or dismiss dialogs that need a button press. Pages where a consent or sign-up dialog held focus, and pages whose add-to-cart button is disabled or hidden until a size is chosen, are left out of the reachability rate rather than counted as failures.
• Add to cart is matched by its accessible name in English ("add to cart/bag/basket", "buy now"). Stores that use other wording are counted as having no matching control, not as failures.
• Visible focus is a heuristic: computed styles of the focused element, its label, parent, next sibling, first child and pseudo-elements are compared focused versus blurred. It misses indicators drawn elsewhere and can count unrelated style changes.
• Product images, the price region and size or colour selectors are found by heuristics on class names, the main heading and structured data, and can miss or misclassify elements.
• Cart pages are opened by direct URL only where robots.txt allows: 32 stores had an (empty) cart page audited and 29 were skipped because robots.txt disallows it, so cart numbers are partial. Checkout pages are never opened.
• The Sightline path line is a design property of the product, not a measurement on these stores.
• Stores appear only as unsalted 12-character sha256 prefixes of their domain. That keeps names out of the published data but does not stop someone who guesses a domain from confirming it; the per-store rows exist for reproducibility, not for naming stores.

How it was measured

• Read-only: no add-to-cart clicks, no forms, no accounts, no personal data.
• robots.txt honoured, one page every five seconds per store, an honest user agent.
• Automated checks find only part of the barriers people meet; real numbers are worse, not better.
• The script and the aggregated data are in the repository: evals/barrier/.