How to Evaluate an Accessibility Vendor's Claims

Vendor pages for accessibility tools sound alike. Each one promises a fast result, a clean report, and a site that works better for everyone. The pages differ in one place that matters: evidence. A claim with a dated report, a sample code change, and a retest behind it can be checked. A claim with only a score, a badge, or a review behind it cannot.

This article shows how to tell the two apart, using the FTC record as a worked example, then gives five checks to run on any vendor before you pay.

Start with the FTC record, read as evidence

On January 3, 2025, the Federal Trade Commission announced a proposed $1 million order settling allegations about accessiBe, maker of accessWidget. The FTC alleged that accessiBe misrepresented the ability of its tool to make any website conform to WCAG, and that third-party articles and reviews were formatted to look impartial without disclosing material connections to the company. The FTC approved the final order in April 2025.

Read that record in two parts. The first part is about a product claim: a statement about what a tool does to a page, made without evidence strong enough to back it. The second part is about a source claim: articles and reviews presented as if they came from outside writers, when the company had material connections to those writers that were not disclosed. Both parts teach the same lesson for a buyer. Check the claim, and check the source of the claim, separately.

Check one: what exactly is claimed?

Write the vendor's main claim in one sentence, in your own words. For example: "This tool changes source code so that named barriers are gone." Or: "This tool adds a panel that adjusts the page while it loads." Or: "This audit tests sampled pages and reports findings for a developer."

Now test the sentence against the vendor's own detail pages. Does the detail page say the same thing, or does it soften the claim? A common pattern is a bold headline claim with a quiet detail page that says something narrower, such as a panel for visitors, a scan of one page, or a report with no fixes. When headline and detail disagree, trust the detail, and ask the vendor to put the headline claim in writing with the detail attached.

Watch for claims that use a result word with no test behind it. A page that says a site will meet WCAG 2.2 AA should say which pages were tested, which criteria were tested, who tested them, and on what date. WCAG 2.2 is the current W3C Recommendation, with testable success criteria at levels A, AA, and AAA, judged on whole pages and complete processes. A claim that names no pages, no criteria, and no date has not yet met that standard of proof, whatever the headline says.

Check two: what changes, in code or at run time?

This is the fastest check in the article. Ask the vendor to show one fix, with the old code and the new code, or to state in writing that no source code changes.

A code fix vendor can show a diff. For sites with a GitHub repository, Axeazy sends fixes as a real pull request with a visible code diff that the owner or developer reviews and merges. That diff is evidence a buyer can read without trusting a score. A script vendor cannot show a code diff, because the product does not change code. That is not a fault in itself, but it changes what the buyer is paying for: visitor controls at run time, not removed barriers in code.

Check three: who tested, and with what?

An automated scan is a useful start, and Axeazy runs 98 automated checks grouped into 14 fix categories. A scan is not a full test. The Department of Justice guidance says automated checkers can help, but a clean report does not mean everything is accessible, and a manual check paired with automated checks gives a better picture. Ask a vendor three things: which pages were tested by a person, which assistive tools were used, and whether keyboard-only use was tested.

A good answer names the method. For example: "A tester moved through the homepage, the contact form, and the checkout with a keyboard only, then checked the same paths with a screen reader, and the findings are mapped to WCAG criteria." A weak answer says testing was done by the tool, or that the tool's own score is the test. A tool testing itself is not a test a buyer can rely on, because the tool cannot see what it was built to miss.

For audits, price is part of the evidence. Accessible.org lists most manual audits at $1,250 to $2,750, at $100 to $250 per primary page and $25 to $100 per light page, with technical support at $195 per hour, checked October 31, 2026. A quote far below that range for a full manual audit should prompt a question: which pages, which criteria, and whether retesting is included. A low price may mean a scan with a summary, not a manual audit.

Check four: who praised the product, and was the connection disclosed?

Reviews, articles, and case studies are evidence only if the source is clear. Ask whether a reviewer was a paying customer, a partner, a staff member, or a paid writer, and whether that connection is disclosed on the page itself.

The FTC allegations about third-party articles and reviews show why this check matters. Articles formatted to look impartial, without disclosing material connections to the company, mislead a buyer about how much outside support a product has. A vendor that discloses connections and separates customer quotes from partner quotes gives a buyer a fairer picture. A vendor that presents every quote as if it came from an outside customer asks a buyer to trust presentation rather than facts.

Check five: what happens after you pay, and what stays if you leave?

A claim about a result should come with a timetable and an exit. Ask for a dated second report after 30 days, a list of fixes merged, and a list of open items. Ask what happens to reports, records, and code fixes when the subscription ends. Merged code fixes stay in your code. A script panel and its settings end with the subscription. A monitoring record should stay available with its dates.

A one hour evaluation routine

Run these steps in order before a second sales call. First, write the vendor's main claim in one sentence and check it against the detail pages, noting any softening. Second, ask for one sample fix with old code and new code, or a written statement that no code changes. Third, ask for the testing method: pages tested by a person, tools used, and keyboard-only coverage. Fourth, read three reviews or articles the vendor shows, and note whether each connection is disclosed. Fifth, ask for a yearly total for your site, the retest terms, and the exit terms in writing.

Read the method on Axeazy's capabilities page and the scan method on Axeazy's methodology page as one worked example of a vendor page that names its fix method, its check count, and its limits, then run the same five checks on every other vendor page.

Related reading

For the head to head that covers the FTC record in a comparison, read Axeazy vs accessiBe: Price, Fixes and Fit, and for the questions to take into a sales call, read how to choose an accessibility tool.

• For a plain buyer checklist to use before you sign, read Accessibility Tool Buyer Checklist.

FAQ

What is the strongest single piece of evidence a vendor can show?

A dated retest report that shows named problems found, fixed in code, and checked again on the same pages.

Does a high price mean a vendor's claims are stronger?

No. Price shows scope and method, not truth. A $1,250 to $2,750 manual audit buys depth on sampled pages, while a $29 a month tool buys repeat scans and a running record. Compare the method and the evidence first, then compare the price for the same scope.

Should I trust a vendor page that shows many reviews?

Trust it only after the disclosure check. Read whether each reviewer was a customer, a partner, or a paid writer, and whether that connection is shown on the page. Praise without disclosure is presentation, not evidence, as the FTC allegations about third-party articles illustrate.

Last checked: October 31, 2026 Sources: FTC January 2025 press release; W3C WCAG 2.2; ADA.gov web guidance; Accessible.org pricing page; accessiBe pricing page; Axeazy capabilities, methodology, and pricing pages.

Start with a dated first read from Axeazy's free scan and keep the report beside any vendor claim you are asked to trust.