Study · October 29, 2026

94% of popular websites split a button across a foldable's hinge

We opened 77 of the most visited websites on a simulated Samsung Galaxy Z Fold and checked what lands under the hinge. Even Google's search box is cut in half. Here is what we found, and how to fix it.

94%split a button, link or form field with the hinge

68%are not foldable-ready (score below 90/100)

74average score out of 100 in book posture

showed a cookie banner sitting on the hinge

Scores are mediocre across the board

No site reached 100. Book posture (screen unfolded like a book, vertical hinge) is where most sites break: the hinge runs down the middle of centered layouts. Tabletop posture (half-folded like a laptop) does much better, at 96 on average.

What breaks

Share of sites with at least one issue of each kind, in either posture.

By category

Average foldable-ready score. Public services and banking do best, travel and search portals worst.

Hall of shame

The lowest scores. Red and orange outlines are the elements under the hinge, as reported by the audit.

Hall of fame

And on the iPhone Fold?

We ran the same audit at the leaked size of the iPhone Fold inner screen (904×640 CSS pixels, assuming @3x like other iPhones) with an iPhone user agent, on the 76 sites that let us in. Apple has not confirmed these specs.

86%put a button, link or form field on the fold line

41%score below 90/100

83average score out of 100 in book posture (74 on the Galaxy Z Fold)

Better than the Galaxy Z Fold, because the wider screen leaves more room around the center, but the same story: Google's search box still sits on the fold (52/100 in book posture).

Two things differ on the iPhone. Its crease is reportedly much shallower, so content on the fold stays readable; the real issue is a wide screen used as a stretched phone layout. And Safari has no foldable API, so pages cannot detect the fold. fold.js handles it with a virtual fold: on an unfolded iPhone Fold in landscape, two-pane layouts split at the center automatically.

Test your own site

The audit behind this study is open source. Run it on your site, your staging server or in CI:

npx foldjs audit https://your-site.com --dismiss-consent
npx foldjs audit https://your-site.com --device iphone-fold

Then fix it with fold.js: two-pane layouts for React, Vue, Svelte, Solid and Web Components, or one script tag that retrofits an existing site.

All results

Click a column to sort. Download the data (JSON).

							Worst issue (book)

Methodology

Tool: npx foldjs audit (fold.js 0.4), headless Chrome with an Android user agent, on October 29, 2026, from France.
Device: Samsung Galaxy Z Fold inner screen, 690×829 CSS pixels in book posture and 829×690 in tabletop posture, with a zero-width fold line as on real devices.
What is checked: the first screen of each home page. An issue is an element whose visible part crosses the fold line: buttons, links and fields, but also clickable elements without proper markup and controls inside web components. Elements hidden or covered by something else do not count, and pages that had not finished rendering after a second, longer attempt are left out. Split buttons, links and form fields are critical; text, modals, floating elements, videos and carousels are warnings; images are minor.
Score: 100 minus 6 per critical issue, 3 per warning and 1 per minor issue, per posture; the site score is the average of both postures. 90 and above counts as foldable-ready.
Cookie banners: a first pass was run as a normal visitor (that is where the 52% figure comes from; on 21% of sites the accept or reject button itself is split). The results above come from a second pass where the audit closed consent banners automatically, accepting as most visitors do (on "pay or accept" sites, rejecting only leads to a subscription wall). A few banners could not be closed and are part of the result.
Sites list: 100 of the most visited global and French websites. 23 sites showed a bot check, an access denied page, a country selection page or a page that never finished loading instead of the real site, and are left out. Every exclusion was checked on the screenshots.
iPhone Fold: same sites and settings at 904×640 CSS pixels with an iPhone user agent, from leaked specs (7.76", 2713×1920 pixels, @3x assumed). The page still renders in Chrome, not Safari. One more site (PayPal) showed a bot check and is left out of this part.
Limits: scores reflect one moment, one location and one device size. Sites change often; run the audit yourself to check today's result.