A quarterly account of the record itself: how many systems are profiled, how much of their functional depth is documented, how many deployments are cited, and where the coverage is thin. Every edition stays published, so the change between them is itself a figure.
53
systems profiled, 203 sources cited, 352 deployments documented.
Every system against every module, 53 by 16. 142 of them are still unverified, which the edition says on its own page.
Each cited to the page that names the company. They cover 21 of the 53 systems; the rest publish nothing we could source.
An edition is frozen when it is published, not when it is generated, and then it stays online unchanged. A number in a published edition reads the same next year.
Before each edition there is a stated period in which a vendor may correct a fact about its own record. 2026-Q3 is still provisional, so its window is not open yet: it runs for the 21 days before publication, and the dates are stated here once that date is set. Corrections go through the public queue and are logged. Vendors correct facts, never placement: rows are alphabetical, nothing is ranked, and there is no placement to buy.
Every table behind the report downloads as CSV and JSON, under CC BY 4.0.
Open the dataset