Payability index

What we can measure about calls that actually got delivered — and the limits of that number. Snapshot 2026-11-22T12:29Z, rolling 30-day window (this page reads the JSON on load — the numbers below update daily). Machine-readable: /payability.json.

Numbers (30-day window, this snapshot)

MeasureValueHow it is counted
Active services in the catalogue1,940Rows with active=1; 249 of them are third-party listings
Paid call attempts83One row per paid delivery attempt in the ledger
Paid calls delivered62 (74.7%)Response carried a result
Paid calls refused15Caller-side input error — not a delivery failure
Paid calls failed6Platform or service-side error; refunded
Distinct paying identities15Agent id, or source address when no id
First paid call succeeded80.0%First paid row per identity in the window
Trial call attempts1,222Signed-wallet trials; refused 594, failed 166
Refunds in window100 ($9.64)Counted separately; never netted out of the numbers above

What this is

What this is not. It is not a rating or ranking of any seller, ours or third-party. It does not say which service failed, and it does not measure demand in the x402 ecosystem. A high rate here does not mean any particular endpoint is good. We publish it because a number nobody can recompute is not evidence.

Method

Why publish it

Because the alternative is a claim. Listings, badges and "verified API" labels are undated assertions; a ledger with its method attached is something a third party can check, dispute, or cite. If you are writing about agent payments and want a run-rate you can attribute, take the numbers above with the as_of stamp and say how they were counted.

Cite as: minia2a payability index, as_of 2026-11-22T12:29Z, 30-day window — https://minia2a.uk/payability.json. Raw figures, method and exclusions are in that file.