WCAG 2.4.5 Multiple Ways

What it asks

People find pages in different ways. Some browse a menu. Some search. Some scan a site map or an A to Z. Some follow links from related content. This rule asks that a site offers at least two of these, so someone who cannot use one has another. The W3C’s list includes site search, a site map, a navigation menu, a table of contents or index, links from related pages, and breadcrumbs.

It applies to a set of pages, not to one page. A single page cannot show whether the site has a search as well as a menu. It is about finding a page within the site, so the pages that are steps in a process are exempt: you are not meant to reach step three of a checkout from the site map. The exemption is for those steps, not for the whole site.

Most sites pass without thinking about it, because a menu and a search box are both normal. The sites that fail are usually web apps built around one search box with no browsable navigation, and sites whose only route to deep pages is a chain of links from the home page. It is Level AA.

Its official name is success criterion 2.4.5 Multiple Ways, at Level AA. It was added in WCAG 2.2, under guideline 2.4 Navigable and the principle Operable. The W3C’s normative text says, in part:

“More than one way is available to locate a Web page within a set of Web pages except where the Web Page is the result of, or a step in, a process.”

From WCAG 2.2, W3C. Full explanation: Understanding 2.4.5 Multiple Ways.

Who it helps

People with cognitive disabilities, who may find a deep menu hard and a search easy, or the other way round. People with low vision, for whom a site map is often quicker than a large menu. Screen reader users, who may prefer search to working through a long navigation region on every page.

Anyone who knows what they want but not where the site keeps it.

What passes and what fails

These pass

• A site with a main navigation menu on every page and a search box in the header.
• A documentation site with a table of contents in the sidebar and a search.
• A shop with category navigation, a search, and breadcrumbs on every product page.
• A small site with a menu and a site map page linked from the footer.
• A checkout whose steps are reachable only in order. Those pages are exempt.

These fail

• A knowledge base with a search box and nothing else: no menu, no index, no site map.
• A site where product pages are reachable only by following links from the home page, with no search and no site map.
• A web app whose only navigation is a search field and a recent items list.
• A site map that has not been updated and omits most of the site, leaving the menu as the only real route.

A site map or a search that does not cover the pages people need does not count as a second way to those pages. And the navigation menu that omits the deep pages is the common hidden failure: the site appears to have two ways, but only one of them reaches most of the site.

How to test it by hand

1. List the ways the site offers to find a page: menu, search, site map, index, breadcrumbs, related links.
2. Pick a few pages deep in the site and try to reach each one by two different routes from the home page.
3. Check that the search works and returns the pages you tried, and that the site map lists them.
4. Identify the pages that are steps in a process, such as checkout or sign-up, and set them aside.
5. A set of pages where you can only find a page one way fails.

What automated testing can decide

What a machine can decide

Given several pages from one site, automated testing can look for each mechanism on each page: a search form, a navigation region, breadcrumbs, a link to a site map, lists of related links. Counting the mechanisms present across the set is routine once the pages are in hand.

What it cannot

Whether a mechanism is real is harder. A search box that returns nothing useful, or a site map that lists a tenth of the site, looks the same in the markup as a good one. Which pages are steps in a process takes judgement. And a scan of one page cannot answer the rule at all, because the rule is about the set.

How Dreven handles it

Dreven treats this as a question about the set of pages in an audit, not about one page. A reviewer sees which ways of finding a page were found across those pages, such as a search, a navigation menu, breadcrumbs and a site map link, and decides whether the site really offers more than one.

Where there are too few pages to answer the question, the report says so rather than guessing. An audit of a single page records that this criterion is about a set of pages and that site-level navigation was not assessed, so a Not Applicable row can never be read as tested and clean.

No automated check makes a site conform. In Dreven, a person reviews the results and signs the report.

Related criteria

• 2.4.1 Bypass Blocks (Level A). Skipping repeated navigation, which is the other side of having navigation at all.
• 3.2.3 Consistent Navigation (Level AA). Navigation that repeats across pages must stay in the same order.
• 3.2.6 Consistent Help (Level A). Help that repeats across pages must stay in the same place.

Questions people ask

All WCAG 2.2 criteria, in plain words