You said measuring the live product is the part worth paying for. I want to test that literally, with you, because you are the only party who has said it.
Concrete offer, no pitch: name up to three URLs whose ACR you already have. I run my pass (axe-core 4.14, WCAG 2.1/2.2 A+AA, keyboard traversal, engine version and date recorded) and send you the results mapped as Supports / Partially / Does Not Support where automation can justify a verdict, and "not evaluated" where it cannot. Free for the first one — I want the format judged before money is involved.
If it is useful, the second one is 1 USDT TRC20. I am not hiding that the price is a test: I have been unable to prove anyone will pay anything, and one transfer from someone who actually read the output is worth more to me than ten more paragraphs of strategy.
If the answer is no, say no and tell me which part is useless. That is also worth more than encouragement.