How to Read an Accessibility Audit Report
What each part of the report is for, how to read a single finding, what the severity labels mean, and what to do in the first week after it lands in your inbox.
See a sample audit reportRead an accessibility audit report in three passes. First the summary, which tells you whether people can complete your key tasks and which few barriers to fix first. Then the scope and method, which tell you what was tested, against which standard, and with which tools, so you know what the report does and doesn't cover. Then the findings, one at a time, starting with the most severe.
Most owners don't need to read every finding. Developers, designers and content authors do, but each of them only needs their own share. A good report is built so each reader can find their part quickly.
This guide walks through the parts of a good report, shows one example finding line by line, and explains how a report differs from an Accessibility Conformance Report (ACR) and from an automated scan export. For a complete worked example, open our sample accessibility audit report in another tab and follow along.
What an accessibility audit report should include
There is no single mandatory format. The closest things to a standard are the W3C's Website Accessibility Conformance Evaluation Methodology (WCAG-EM), which sets out what an evaluation should document, and the W3C's template for accessibility evaluation reports. Reports from different auditors use different headings, but a useful one covers the same ground.
| Part | Question it answers | What to look for |
|---|---|---|
| Executive summary | Where do we stand, and what do we fix first? | Plain language, the tasks people can't complete, a count of findings by severity, and the top few priorities |
| Scope | What was tested? | The pages, templates and user journeys by name, the date of testing, and anything left out |
| Standard and method | What was it tested against, and how? | The WCAG version and level, and a description of the manual testing, not only a tool name |
| Tools and assistive technologies | What was it tested with? | Named screen readers, browsers and devices, keyboard-only testing, zoom levels, and any automated tools |
| Findings | What exactly is wrong, where, and how do we fix it? | One entry per barrier, each with a WCAG criterion, severity, location, what happened and a recommended fix |
| Roadmap | In what order should we work? | Findings grouped or ordered by priority, so the first sprint is obvious |
Scope and method
The scope tells you what the findings apply to. Audits usually test a representative sample of templates and journeys, not every page, and WCAG-EM is built around that kind of sampling. The catch is that a report only speaks for what was tested. If your checkout or your booking form isn't listed in the scope, the report says nothing about it.
The method should name the standard (for example, WCAG 2.2 Level AA) and describe how the testing was done. If you're not sure what the letters mean, see our guide to WCAG levels A, AA and AAA.
Tools and assistive technologies
Look for a specific list: which screen readers with which browsers, which phone, whether keyboard-only testing and zoom were covered, and which automated tool was used for the first pass. Our sample report lists keyboard only, NVDA with Firefox, JAWS with Chrome, VoiceOver with Safari on iPhone, 200% and 400% zoom, and axe DevTools. A report that names only a scanning tool describes a scan. The difference is covered in automated vs. manual accessibility testing.
How to read a single finding
Every finding should answer the same handful of questions. The example below is illustrative. It is not from a real client report, but it uses the same fields as the findings in our sample report.
| Field | Example | How to read it |
|---|---|---|
| ID | EX-04 | A stable reference. Use it in tickets, in the walkthrough call and at re-test. |
| Title | Keyboard users can't open the main menu on small screens. | Says who is affected and what they can't do, in one sentence. |
| Severity | Critical | How badly this blocks a person. It is the auditor's judgement, explained below. |
| Where | Site header, the menu button, on every page at widths under 768px | The location. "Every page" and "header" tell you it's a shared component, so one fix clears it everywhere. |
| Who it affects | Keyboard users, switch users and screen reader users | The people who hit this barrier. Useful for explaining the fix to your team. |
| What we found | At 400% zoom the navigation collapses into a menu button. Pressing Tab skips the button and moves from the logo to the page content. NVDA does not announce the button. It opens on mouse click only. | The steps to reproduce: what the tester did and what happened. A developer should be able to repeat it and see the same result. |
| How to fix it | Build the control as a native <button> with a visible focus style and an accessible name such as "Menu", and expose whether the menu is open with aria-expanded. | A recommendation specific to your code, not a quote from the standard. Your developers may solve it another way, which is fine if it passes the re-test. |
| WCAG | 2.1.1 Keyboard (Level A). Part of WCAG 2.0 AA, the AODA baseline. | The success criterion the barrier fails. This is what makes the finding checkable against the standard. |
Three things are worth checking on any finding you read:
- Can you reproduce it? If "what we found" is specific enough, you can see the problem yourself in a minute. Unplug your mouse and try.
- Is the criterion named? You can look up any success criterion in the W3C's documentation, and our guide to the WCAG 2.2 success criteria explains them in plain language.
- Is the fix about your site? "Ensure all functionality is available from a keyboard" is the criterion restated. "Replace this element with a button" is a fix.
What the severity levels mean
Severity tells you how badly a barrier blocks a person. Most reports, including ours, use four levels:
| Severity | What it means | Example |
|---|---|---|
| Critical | Someone can't complete a task at all. | A keyboard user can't choose a pickup time, so they can't place an order. |
| Serious | The task is possible but very hard, or information is lost. | Quantity buttons are announced only as "button". |
| Moderate | Frustrating, but there is a workaround. | Product photos are announced by their file names. |
| Minor | A small barrier or inconsistency that doesn't stop anyone. | Footer links that fall slightly short of the contrast requirement. |
Severity scales are not standardized. WCAG doesn't define one, so the labels and their definitions belong to the auditor or the tool. Some reports use "blocker" or "high, medium, low" instead. Check that your report defines its own scale, and read the definitions before you compare two reports.
Severity is not the same as WCAG level
This is the most common misreading. Level A and Level AA describe which conformance level a success criterion belongs to. They don't tell you how bad a particular failure is. The W3C's Understanding Conformance page says WCAG 2 does not suggest any particular level of priority or severity, and that how severe a failure is depends on things like how often it occurs and what content or function it affects.
In practice the two can point in opposite directions:
- A Level A failure can be minor. A decorative image with a missing empty
altattribute on one archived page fails 1.1.1 Non-text Content (Level A), and stops nobody. - A Level AA failure can be serious. In our sample report, a menu that scrolls sideways and cuts off prices at 400% zoom fails 1.4.10 Reflow (Level AA) and is rated Serious.
To conform at Level AA, a page has to meet every Level A and every Level AA criterion, so both kinds of failure need fixing in the end. Severity only tells you the order. Don't work through a report "Level A first".
The AODA baseline column
Many audits test against WCAG 2.2 Level AA, while the AODA's website requirement in Ontario is WCAG 2.0 Level AA. A report written for an Ontario organization should mark which findings fail a WCAG 2.0 AA criterion and which fail criteria added in WCAG 2.1 and 2.2. Our sample report does this on every finding. Our AODA requirements guide explains who the rule applies to. This is general information, not legal advice.
See a full report, start to finish
Our sample report covers a fictional bakery's online ordering: scope, what it was tested with, a summary and eight findings from Critical to Minor.
Open the sample audit reportWho reads which part
Nobody needs to read the whole report closely. Split it like this:
- Owner or manager: the executive summary, the scope, and the count of findings by severity. Your questions are: can customers finish the main tasks, what gets fixed first, who owns it, and when is the re-test?
- Developer: every finding that involves keyboard support, names and roles, forms and error handling, or focus. Read "where", "what we found" and "how to fix it", and reproduce each one before changing code.
- Designer: findings about colour contrast, focus indicators, target sizes, layouts that don't reflow, and information shown only by colour. These usually need a design decision before a developer can act.
- Content author: findings about alt text, headings, link text, captions and documents. These are often the quickest to fix, and the quickest to reintroduce, so the author needs to understand the reason and not only the correction.
If the report comes with a spreadsheet, filter it by these groups and send each person their own list. A person handed thirty relevant findings acts sooner than one handed a 60-page PDF.
What to do in the first week
- Read the summary and the scope. Confirm the pages and journeys tested are the ones you asked for. Raise any gap now, not after the fixes.
- Reproduce the critical findings. Seeing a barrier yourself is the fastest way to understand it, and to explain to your team why it matters.
- Hold the walkthrough call. If your auditor offers one, bring the developers and designers who will do the work, and ask about any finding that isn't clear.
- Move the findings into your tracker. One ticket per finding, keeping the finding ID so it can be matched at re-test.
- Assign owners. Every finding gets a named person, even if the fix is months away.
- Start on critical barriers in shared components. These remove the most harm with the least work.
- Set the re-test date. If a re-test is included, find out how long the window is and plan the fixes to fit inside it.
That gets you moving. Sequencing the rest of the work, writing tickets and tracking progress is covered in our guide to building an accessibility remediation plan.
Audit report vs. ACR vs. automated scan export
These three documents get confused, and they do different jobs.
| Audit report | ACR (completed VPAT) | Automated scan export | |
|---|---|---|---|
| Who it's for | Your own team | Your buyers and their procurement teams | Your own team, as a first look |
| Organized by | Barrier: one entry per problem found | Criterion: one row per requirement in the standard | Rule: one entry per automated check that failed |
| What it tells you | What is broken, where, for whom, and how to fix it | How well the product meets each criterion, using terms such as Supports, Partially Supports and Does Not Support, with remarks | The issues a tool can detect in the code of the pages it loaded |
| Produced by | A person testing with a keyboard and assistive technology, plus tools | The vendor or an auditor, ideally from audit results | Software |
An ACR is a completed Voluntary Product Accessibility Template (VPAT), the form published by the Information Technology Industry Council. It states conformance criterion by criterion; the US government's Section 508 site explains the conformance terms. It is not a fix list, and it is not a certification. An audit report is usually the evidence an ACR is written from. See what a VPAT and an ACR are, or our VPAT and ACR service if a buyer has asked you for one.
An automated scan export is a list of rule failures from a tool. It is quick and useful for catching problems that are detectable in code, and you can run a free automated scan on our site. It can't tell you whether someone can complete your checkout with a keyboard, or whether your alt text makes sense, because no one tried.
Signs of a weak accessibility audit report
- No scope. You can't tell which pages or journeys were tested, or when.
- No named assistive technologies. If no screen reader or keyboard testing is mentioned, it probably didn't happen.
- Findings without a WCAG criterion. You can't verify them against the standard or use them as evidence.
- Findings without a location or steps. "Some images are missing alt text" leaves your team to find them all over again.
- Fixes that restate the standard. Generic advice copied from WCAG, with nothing about your pages or code.
- The same finding repeated for every page. Hundreds of entries for one template bug make a report longer, not better.
- Only tool-detectable issues. If every finding is something a scanner reports (contrast, missing alt, missing labels) and nothing describes what happened with a keyboard or screen reader, it's a scan with a cover page.
- Severity with no definitions, or everything marked high.
- A whole-site verdict from a small sample. WCAG-EM is explicit that a conformance claim for an entire website can't be made from evaluating a selected subset of it. Be wary of "your site is compliant" or a badge. The AODA has no certification scheme for websites.
- No re-test. Without one, the report is a list of problems and never becomes a record of what was fixed.
If you're comparing proposals before you buy, ask each auditor for a sample report and check it against this list. What an audit involves, and what you're paying for, is covered in what a website accessibility audit is. Our own audits are fixed-price; see what each package covers.
Frequently asked questions
What should an accessibility audit report include?
A plain-language summary, the scope (pages and journeys tested, and when), the standard and method, the tools and assistive technologies used, and one entry per finding with the WCAG success criterion, severity, location, what the tester observed and a recommended fix. The findings should be in priority order or come with a roadmap.
Does a critical finding mean we're breaking the law?
Not by itself. Severity describes the effect on a person using your site, not your legal position. Whether a legal requirement applies depends on your organization and jurisdiction, and on which standard the finding fails. In Ontario, the AODA's website rule is based on WCAG 2.0 Level AA, so look at which findings fail that baseline.
Should we fix Level A findings before Level AA findings?
No. Use severity, and whether the barrier sits on a key journey or in a shared component. WCAG levels say which conformance level a criterion belongs to, not how serious a given failure is. You need both A and AA criteria met to conform at Level AA.
Does a report with no findings mean the site is fully accessible?
It means no barriers were found in the pages and journeys tested, on the date they were tested, against the standard named. It doesn't cover pages outside the scope, or changes made since.
Can I send our audit report to a customer who asked for a VPAT?
Usually not. A buyer asking for a VPAT wants an Accessibility Conformance Report in the ITI template, with a conformance level for each criterion. The audit report is the evidence behind it, and it also contains detail about your code that you may not want to share.
How long is an audit report valid?
There is no fixed period. A report describes the site as it was on the test date. It goes out of date as the site changes, quickly after a redesign or a new feature, and slowly on a site that rarely changes.