WCAG 2.2 success criterion, Level AA
WCAG 3.2.4 Consistent Identification
A control that does the same job on several pages must be identified the same way on each of them: the same name, the same icon, the same label. Not "Search" on one page and "Find" on the next.
- Number
- 3.2.4
- Level
- AA
- Added in
- WCAG 2.2
- Guideline
- 3.2 Predictable
- Principle
- Understandable
What it asks
When the same function appears on more than one page in a set, its name, its icon and its text alternative should match from page to page. A search button called "Search" on the home page and "Find" on the product pages does the same thing under two names. A magnifying glass icon with the text alternative "Search" on one page and "Go" on another is the same failure in a different place.
The other direction matters too. Two controls with different functions should not share a name. A "Save" that saves a draft and a "Save" that publishes are confusing in a way that this criterion is meant to prevent. The rule does not require that every page has the same controls, only that a control which repeats is recognisable as the same control.
Consistency is about identification, not about matching words exactly. A "Download PDF" link whose text alternative also names the document it downloads is still consistently identified, because the pattern is the same. The criterion is Level AA, applies across a set of pages, and is one of the easier criteria to fix once the inconsistent pairs are listed.
Its official name is success criterion 3.2.4 Consistent Identification, at Level AA. It was added in WCAG 2.2, under guideline 3.2 Predictable and the principle Understandable. The W3C’s normative text says, in part:
“Components that have the same functionality within a set of Web pages are identified consistently.”
From WCAG 2.2, W3C. Full explanation: Understanding 3.2.4 Consistent Identification.
Who it helps
Screen reader users, who learn a control by its name and search for it by that name on the next page. If the name changes, the search fails and they have to work out whether "Find" is the same thing.
People with cognitive disabilities and learning difficulties, who are helped by things that behave and look the same each time. People with low vision, who recognise an icon and expect it to mean what it meant last time.
What passes and what fails
These pass
- A search button labelled "Search" in the header of every page.
- A magnifying glass icon whose text alternative is "Search" wherever it appears.
- "Sign in" used on every page, with "Sign out" as its opposite, and never "Log in" as a variant.
- A "Download" link on each document page whose text alternative follows the same pattern: "Download annual report, PDF".
These fail
- "Search" in the header on most pages and "Find" on the help pages, for the same search.
- "Log in" on the home page and "Sign in" on the checkout, for the same action.
- The same cart icon with the text alternative "Basket" on product pages and "Cart" on the checkout.
- A "Submit" button that sends a message on one form and saves a draft on another.
How to test it by hand
- Choose a set of pages across different templates, as for 3.2.3.
- On each page, list the controls that repeat: search, sign in, basket, menu, contact, download, share, and any icon-only buttons.
- For each one, write down its visible text, its text alternative if it is an icon, and its accessible name as the browser’s developer tools report it.
- Compare across pages. A control with the same function and a different name, or the same name and a different function, fails.
- Note the inconsistent pairs with the pages they are on. That list is the fix.
What automated testing can decide
What a machine can decide
Given a set of pages, automated testing can find the controls that repeat from page to page and compare their names, listing each pair that differs and the pages it differs on. Any honest comparison has to match controls by what they do rather than by what they are called, since the name is the thing being tested.
What it cannot
Some differences are not failures and need a person. "Pricing" in the header and "Plans" in the footer may be two links to one page that are each consistent in their own place. A control with no name and no clear place in the page is harder for any tool to match with its counterpart elsewhere. And the comparison needs several pages: one page cannot be inconsistent with itself.
How Dreven handles it
Dreven looks at the controls that repeat across the pages in an audit and shows a reviewer any whose name differs from page to page, with the pages each appears on. The reviewer decides each result, because some differences are deliberate.
The result covers the set of pages in the audit. Where there are too few pages to compare, the report says so, and an audit of a single page records that consistency across the site was not assessed.
No automated check makes a site conform. In Dreven, a person reviews the results and signs the report.
Related criteria
- 3.2.3 Consistent Navigation (Level AA). Repeated navigation must keep the same order on every page.
- 3.2.6 Consistent Help (Level A). Help that repeats across pages must sit in the same place in each page’s order.
- 4.1.2 Name, Role, Value (Level A). A control must have a name at all before its name can be consistent.
- 2.4.4 Link Purpose (In Context) (Level A). Each link’s purpose must be clear from its text.
Questions people ask
Does every page have to use exactly the same words?
The same function should be identified the same way, so a search button should be called the same thing on every page. Words can vary where the function varies. A pattern such as "Download annual report, PDF" and "Download price list, PDF" is consistent.
Is "Log in" versus "Sign in" really a failure?
When both lead to the same sign-in for the same site, yes. A person searching the page for the name they learnt will not find it. Pick one and use it everywhere.
Can an automated tool test 3.2.4?
Given several pages, it can match the controls that repeat and list the ones whose names differ. It cannot tell on its own whether a difference is deliberate, so the list goes to a person. On a single page there is nothing to compare.