The short answer
Inclusely opens your store in a real browser and runs axe-core 4.14, the open source engine most scanners use, plus our own checks for WCAG 2.2 AA. Automated checks find only part of the problems, so we mark which ones need a person and guide you through them. Every result is dated and kept.
Inclusely is in early access
What we check
Each scan opens one or two pages of every template in your store: the home page, collections, product pages (including one with video and one with options), the cart page and cart drawer, search, a blog post, a regular page, the contact page and the 404 page. We find these through Shopify, not by guessing links.
Every page is tested at three sizes: a desktop screen (1440 by 900), a phone (390 by 844) and a narrow 320-pixel view for reflow. Popups are tested twice, once when they appear and again after they close.
On each page we run two layers of checks:
- axe-core 4.14, an open source engine by Deque, with its WCAG 2.0, 2.1 and 2.2 A and AA rules and its best-practice rules. We turn on its target size rule (2.5.8), which is off by default.
- Our own checks, which press keys, open drawers and measure things axe-core does not test. They are listed in the next section.
We also read your whole product catalogue through Shopify. That finds missing alt text (a short description of an image for people who cannot see it) and weak alt text, such as file names, product codes or the word “image”, on every product, not only the sampled pages. We check every colour scheme in your theme settings for readable contrast.
We never scan checkout, customer account pages or anything behind a login.
Our own checks on top of axe-core
Of the criteria added in WCAG 2.2, axe-core 4.14 automates only target size. These are the tests we add. Each one asks a plain question.
| Criterion | What the test asks |
|---|---|
| 2.4.11Focus not obscuredNew in 2.2 | Is the focused link hidden behind a sticky header or chat button? |
| 3.2.6Consistent helpNew in 2.2 | Is your help or contact link in the same place on every page? |
| 2.4.7Focus visible | Can a shopper see where they are when they press Tab? |
| 2.1.2No keyboard trap | Can you Tab out of every menu, popup and drawer? |
| 1.4.11Non-text contrast | Are input borders, colour swatches and focus rings easy to see? |
| 2.2.2Pause, stop, hide | Can a slideshow that moves for more than five seconds be paused? |
| 1.4.10Reflow | Does the page work 320 pixels wide without sideways scrolling? |
| 4.1.3Status messages | Does a screen reader hear that an item was added to the cart? |
We also check text over images by sampling screenshot pixels, text spacing (1.4.12), form fields missing autocomplete (1.3.5), repeated vague link text (2.4.4), and whether the page language matches your market (3.1.1).
What automation cannot check
Automated checks find some problems, not all. In Deque's 2021 study, automated testing found about 57% of issues by volume, but could fully test only about 16 of the 50 WCAG 2.1 AA success criteria (Deque, March 10, 2021). A person still has to judge things like these:
- whether alt text describes the image well, not only whether it exists;
- whether a screen reader user can find a product, pick options and reach checkout;
- whether video captions are accurate;
- whether error messages say clearly how to fix the mistake;
- whether the page still works at 200% and 400% zoom.
Inclusely marks every result as automated, needs review, or manual. It includes eight guided tests for the manual parts, with steps and a description of a pass. A result is saved with the tester's name and the date. Our score shows automated checks and manual tests as two separate numbers, never one.
For a criterion by criterion view, see our WCAG 2.2 AA guide.
How fixes are reviewed
- Alt textSuggestions are shown in a grid. You edit or approve them, and only approved text is saved to Shopify. Alt text you wrote is never replaced unless you choose that for a batch. The old text is kept so you can undo.
- Theme settingsWe suggest the nearest colour that is easy to read, in the same hue, and show where to change it in the theme editor. You make the change.
- Theme codeChanges come from a tested library of fixes for Dawn and Horizon. Each is checked for valid Liquid and Shopify Theme Check, kept small, and never changes the settings part of a theme file. You apply it on a copy of your theme, or our developers do.
- Optional storefront fixesOff until you turn them on, under 10 KB in total, and each listed as temporary next to the permanent change.
A problem is marked fixed only after a new scan confirms it is gone. Until then it stays open.
How we date and keep evidence
Every scan, finding, fix, approval, undo and manual test becomes an entry in your log. Each entry records who did it, what changed, the date and time, what it looked like before, the scan it came from and the engine version.
Entries are only ever added. Each one stores a fingerprint (a SHA-256 hash) of the entry before it, so if an old entry were changed, the chain would show it. Screenshots are cropped to the problem element.
You can export the log as a PDF or as a data file. How long it is kept depends on your plan, as shown on the pricing page.
We never record whether a shopper uses a screen reader or other assistive technology.
When a result is wrong
Every automated tool sometimes flags a problem that is not real. We keep a list of known cases and skip them, for example a native browser dialog that a theme already uses correctly. When a result needs a human eye, we mark it “needs review” instead of counting it as a failure.
You can mark any finding as wrong, with a reason. Those reasons go into your log, and we use them to improve the list.
Engine versions and how we update them
| Part | Version | Note |
|---|---|---|
| Standard | WCAG 2.2 Level AA | Reports can also be filtered to WCAG 2.0 AA (Ontario) and 2.1 AA |
| Rules engine | axe-core 4.14.x | Pinned, and saved on every finding |
| Browser | Chromium | Desktop, phone and 320-pixel views |
| Theme fixes | Dawn and Horizon | More themes later |
We test each new axe-core release on our own test stores before we switch to it. When an engine update changes the rules, the log notes it, so a change in your count is not mistaken for a new problem in your store. We review this page at least every three months and whenever an engine changes.