Every figure with its date and the document it was measured with.
The canary exercises the product in production every 30 minutes. Last round: 2026-11-08T12:35:23+00:00, 33 quality checks, green
Core Web Vitals p75 (n = 40): LCP 1.20 s ✓ (≤ 2.0 s), INP 64 ms ✓ (≤ 150 ms), CLS 0.012 ✓ (≤ 0.08)
50/50, 20/20
how it's measured 50 trap questions on 5 documents we know word for word (contract, policy, payslip, minutes, quote): it must say “it isn't there” without giving any figure. And 20 answerable ones: the data and the right page.
7/7
how it's measured A contract and two invoices through the 7 tools that extract amounts, dates or deadlines: every figure must be in the document or follow from it by a calculation we redo.
227/227
how it's measured OpenPEPPOL's official unit tests (BIS Billing 3.0), each with what the official validator must say.
0, BR-CO-16 ✓
how it's measured The rules published by the European Commission, run as they are: the official example passes and, with the total tampered, the right rule fires.
3/3
how it's measured The official examples from the Spanish Tax Agency's hash specification (record, chained record and cancellation).
161, 182, ✓
how it's measured The official example of Gipuzkoa's Foral Order 521/2020; and a consistent forgery —the same amount altered on paper and in the QR— that the check code exposes.
PAdES-BASELINE-LT
how it's measured Signed in production and validated with the European Commission's official validator (DSS).
1000, 5.9 s, 37/37
how it's measured A ZIP with 1,000 invoices, 37 of them tampered: time in production and whether the report flags exactly the tampered ones.
0
how it's measured All 159 tools run in Chinese and compared with Spanish: interface texts left untranslated.
159/159, 159/159
how it's measured All 159 tools, end to end, in Spanish and in Chinese, also listening for browser errors.
25/47, Word→PDF ✓
how it's measured A conforming PDF/UA-1 document (title, language, headings, list, table, image with alt text) goes through every tool that returns a PDF, and veraPDF —the reference validator— judges the result. Plus Word→PDF from an accessible Word file.
20/20, 0
how it's measured The 20 most used tools, end to end with the keyboard only in a real browser: search, open, pick the file, fill in what it asks, run and download. The second figure: controls on the way without an accessible name (what a screen reader would say).
87/87, 0
how it's measured Orca 46 —GNOME's free screen reader— on Chromium, no mouse: the home page and the 22 most used tools, in Spanish, English and Chinese, walked with the Tab key. What Orca SAYS at each stop is what gets judged. The figure: walks with no problem at all (a control with no name, a silence, text in another language, a repeated name or a hidden focus).
10/10
how it's measured Ten UNTAGGED PDFs —from three different programs: text with headings, images, links, two columns, a form, a scan, symbols— go through “Make accessible” and veraPDF judges whether the result meets PDF/UA-1.
0.6 s
how it's measured Largest paint of the home page with the CPU 4× slower and slow 4G.
0
how it's measured WCAG 2.2 (A and AA) violations according to axe: 30 site pages in the seven languages and all 181 tools opened, each with its panel.
17/17, 3/3
how it's measured 20 real exclusions confirmed by procurement tribunals (TACRC, Madrid, Andalusia and the Basque Country), each with a document reproducing its defect: Sobre listo must detect the 17 the tribunal upheld and not block the 3 it overturned.
2627/2627
how it's measured 2627 real awards from the Public Sector Contracting Platform, from the last few days: each body's tax ID is looked up in the official directory of contracting bodies (23514 bodies) and all 2627 are recognised, each with its exact body. Legal entities and joint ventures (UTEs) only.
888, 15280
how it's measured Preliminary market consultations and prior notices (888) and expiring contracts (15280) that we flagged and that were later tendered: same body, same subject matter. Each case keeps the date of the notice, the date it came out and the link to the tender; the document shows a sample of the latest 5 of each kind, with no personal names.
96.9 %, 85.9 %
how it's measured 15 real tender documents from the Community of Madrid, read without having looked at them before the reader was frozen: it finds 85.9% of the exclusion grounds and performance conditions marked by a human reading (140 of 163) and, of the sentences it gives, 96.9% are right (375 of 387, judged by hand one by one). Every quotation appears word for word in the document (390 of 390).
9/10, 0
how it's measured A bank of 20 test technical proposals, 10 with a planted contradiction (for example a deadline promised in two different ways) and 10 without: it finds 9 of the 10 and raises no false alarm across all 20. Without tuning against this bank it was 7 of 10 at the first attempt; after correcting two plantings that were not real contradictions, 9 of 10.
Download the documents and what should come out, and run them through PAPYR or any other product.