Skip to main content

OPEN RESEARCH · MAP

World Accessibility Map

Every point is a website we tested ourselves. The municipal layer shows city, county and state websites traversed with a real keyboard, page by page. The second layer shows institutions and organisations: in the U.S. shops, cafés, clinics and schools checked with axe-core; in the U.S. and other countries worldwide schools, libraries, clinics, services and organisations traversed with a keyboard. Click a point to open the full report for that site and its standing among sites in the same state or country.

Keyboard traversal: show

Country fill:75 % or more50–75 %under 50 %fewer than 100 measured pages

Skip the map and go to the ranking

Loading points…

World country ranking

Every country whose municipal websites we traversed with a real keyboard, plus the United States. The share is the part of MEASURED pages where a keyboard-only user reached the goal. Countries with at least 100 measured pages have a rank. Countries with fewer follow below without a rank: their share is shown but is not stable; the number of pages is in the next column.

How to read the table. “Usable” is the share of measured pages where a person who cannot use a mouse and moves through the site with the Tab key reached the elements they needed (links, buttons, form fields) and could see where they were. For example, 45.6 % means they reached the goal on almost every second such page and got stuck on the rest. “Pages: reached / measured” are the numbers behind the share. “Sites” is how many sites we went through there. The higher the share, the more accessible the sites.

#CountryUsablePages: reached / measuredSites

How to read this map

What the share does not mean. The world instrument, used for almost all countries, has known limits found by a control run on 9 November 2026: on a home page any element with text within the first three Tab presses counts as the goal, so a trap further down the page goes unseen; focus visibility is judged from the outline and box-shadow only, so a focus shown by a background fill is recorded as invisible; in 19.0 % of its section measurements one address is recorded under several sections at once, and some of those sections were measured on the wrong page, which overstates the barrier. In 11 European countries part of the sites was measured by a different, stricter instrument, V5.4.4: the country table on the Europe research page shows which instrument measured which country, so compare them with care. The share describes a measurement of pages at one moment and does not replace a WCAG conformance review.

These are two different instruments. The keyboard traversal asks whether a person can reach the goal using only Tab. The axe-core check counts rule violations in the code. The two layers do not add up, which is why the map never mixes them.

A verdict describes one page at one moment. “Did not open” is not always the site’s fault: we re-measured part of those pages and published a corrected copy of the dataset.

Standing in the state or country is not a score we invented; it compares the site with other measured sites in the same state or country. In 11 European countries that comparison spans both instruments.

The map counts organisations and sites — one dot each. The research pages count log rows, so the commercial “measured” figure on the map is a few dozen lower: those are repeat rows for the same organisation.

The Europe & world layer is a keyboard traversal with a real Tab press: in 11 countries part of the sites was measured with instrument V5.4.4; the other sites of those countries and all sites of the other 185 countries by the traversal of 28 Oct – 9 Nov 2026 over municipal registries (official lists and Wikidata). For those 185 countries the dot is placed by the municipality’s registry name in GeoNames, within its own country. A dot marks the centre of the site’s town: found by the institution’s name in OpenStreetMap (Nominatim), or by the town in the site’s address checked against the institution’s name in GeoNames (CC BY 4.0). It is the town, not the building; a spot check suggests about 5% of locations may be off. Sites without a location are in search and in the country shares.

Government websites research · Commercial websites research · Europe research · Open data · Ask the research · find your site

Author: Maksim Galatin. Self-funded research; the author provides accessibility services; nobody paid to be included or excluded.

Check your own site for free