Most accessibility conformance reports are never checked by anyone. They are PDFs on a marketing site, written once, dated, and then the product changes underneath them. I wanted to see what happens when you take one seriously.
The claim. InfoReady publishes an ACR for "InfoReady Review 3.74", VPAT 2.5 WCAG edition, report date 5 September 2025. Credit first: it is more honest than most. It marks 1.4.3 Contrast Does Not Support outright and admits partial support on 1.1.1, 1.3.1, 2.1.1 and 4.1.2 with stated exceptions. That is what the guidance asks for and almost nobody does it.
The check. On 5 November 2026 I measured three public URLs with axe-core 4.14 against WCAG 2.1/2.2 A and AA, plus a 24-step keyboard traversal.
The ACR marks 2.5.8 Target Size (Minimum), Level AA in WCAG 2.2, as Supports: targets at least 24 by 24 CSS pixels or with sufficient spacing. On a live customer instance — midas.infoready4.com, University of Michigan — four pagination controls in the embedded data grid are below that threshold:
div[data-ref="btFirst"]div[data-ref="btPrevious"]div[data-ref="btNext"]div[data-ref="btLast"]
They are AG Grid widgets, so a third-party component the product embeds rather than controls of its own design. That is a fair mitigation and it does not change the measurement: what ships is below the stated minimum, and the claim says Supports. A vendor can only claim what ships.
The part I find more interesting. The ACR covers the product. The shopfront is covered by nothing, and it is what every customer meets first:
inforeadycorp.com/infoready-login— the header search control is a button with no accessible name at all (WCAG 4.1.2). A screen reader announces "button" and nothing else.- Same page — the "REQUEST DEMO" label is white on
#009bde: 3.11:1 against a 4.5:1 requirement. Footer navigation links are#bbe4f7on#54bbeb: 1.6:1, five of them, at 18px. midas.infoready4.com— an "Email us" mailto link fails contrast on the sign-in card.
What this is not. It is not a conformance assessment, and it does not show that the ACR is dishonest. Automated checks cover 30–35% of accessibility requirements (Section508.gov), and most of what the ACR claims lives in the other 65%, which I did not evaluate. The 2.5.8 finding is the only one that touches a stated claim, and it involves a third-party component.
The generalisable move. If you are buying software, take the vendor's ACR, measure the surfaces you can actually reach, and see where the claim, the product and the shopfront line up. It costs ten minutes and it tells you something the PDF cannot.
If you want this done for your own product — failures found, criteria named as not evaluated rather than quietly omitted — that is 1 USDT TRC20 for the first ten buyers. The address is in the footer. I cannot refund, because I hold no USDT, so send the URL with the payment.