The inversion, laid out plainly
Report 07 published the endpoint count: only a small minority of
indexed CDP Bazaar endpoints cross the ≥5-unique-payer 30-day
threshold. Every reader who quoted that number as
“99.7% of endpoints have ≥1 payer” was
answering a listing question, not a demand question. Report 10 answers
the demand question by putting the SAME cohort next to the SAME
snapshot’s l30DaysTotalCalls
field — CDP’s own count of requests, unaltered by us.
On this snapshot, endpoints with ≥5 unique payers are 762 of 14,675 indexed endpoints (5.19% endpoint-weighted) and they carry 174,008 of 315,911 total 30-day requests indexed (55.1% call-weighted). Endpoints with exactly one unique payer are 10,873 of 14,675 (74.1% endpoint-weighted) and they carry 84,661 of 315,911 requests (26.8% call-weighted).
The two shares live on different denominators — endpoints-with-both-fields vs total 30-day requests indexed — and are stated together on purpose. Endpoint share and call share answer different questions; one without the other misreads the shape. Naked, either half tells a plausible-sounding story that omits the other half.
Why the median matters more than the mean
The exactly-1-payer cohort is where the listing surface lives — 74.1% of endpoints, but only 26.8% of requests. The MEDIAN 30-day call count for that cohort is 1. Read plainly: the typical x402 endpoint has been touched about that many times in thirty days. A mean would be dragged upward by a few outliers with thousands of calls from a single test wallet; the median is what the middle endpoint actually looks like, and the middle endpoint looks like nothing.
This is why the live market map surfaces both denominators next to their numerators in the same block. Every share on this page can be re-derived by the reader from the two numbers immediately next to it — there is no “trust me” percentage on this page.
External corroboration, and the shape of the disagreement
A widely-referenced Chainalysis piece — “Inside x402: 100M Agentic Payments on Base” (Chainalysis Team, published June 3 2026) — is the nearest thing to an on-record third-party read of x402 adoption.
Measured at the Base wallet level, cumulatively through Q1 2026, Chainalysis reports that “agentic payments on Base crossed 100 million transactions in approximately three quarters,” that “transactions of $1+ now represent 95% of total volume transferred, up from 49% in early 2025,” and that a wallet-level tester-to-payer conversion rate has been improving 4x in six months, while the piece also characterises the ecosystem as one where “mass adoption remains distant” and the user base “skews toward newer entrants rather than established institutional players”.
What we measured is different. Report 10 is endpoint-level, rolling 30-day, call-weighted request concentration over CDP-Bazaar-indexed endpoints; Chainalysis counts wallets touching the x402 protocol on Base cumulatively. Different unit of count, different time window, on the same underlying phenomenon.
Where the two agree, directionally. Both find that a large share of early x402 activity is thin and exploratory, and that a smaller cohort carries the meaningful activity. Neither piece treats the ecosystem as mature.
Where the two diverge. The Chainalysis piece is positive on the trend — a wallet-level conversion rate improving 4x in six months, and a $1+ share climbing from 49% to 95%, all measured at the Base wallet level cumulatively through Q1 2026. Our finding is a current-state concentration read, endpoint-level and rolling 30 days on the CDP-Bazaar-indexed sample, and it carries no trend of its own. We do not paint them as saying otherwise.
What we explicitly do NOT claim. We fetched and read the primary source at the article itself on 2026-09-03, not from a search snippet. The article does not contain any characterisation that roughly 50% of x402 activity is exploratory rather than commercial. The article does not cite the higher ecosystem-wide transaction count that appears in downstream summaries. The article does not cite an active-agent count in the tens of thousands — Chainalysis measures wallets and transactions and publishes no agent count at all.
Q. Do Chainalysis’s public x402 findings support the claim that most on-chain x402 activity is exploratory rather than commercial? A. Their published metrics — measured at the Base wallet level, cumulatively through Q1 2026 — are directionally consistent with that reading, but they explicitly frame the trend as improving, they measure wallets rather than endpoints, and the specific “50% is exploratory” characterisation that circulates online is not in their article.
Methodology
Source:
https://api.cdp.coinbase.com/platform/v2/x402/discovery/resources — the public
Coinbase CDP x402 Bazaar discovery API. Unauthenticated, paginated,
JSON. Every entry carries a quality
sub-object with l30DaysUniquePayers
and l30DaysTotalCalls.
Snapshot date: 2026-09-07T09:00:55Z. Every figure on this page is recomputed from that snapshot on each build. The snapshot refreshes daily on the market-map deploy cadence.
N: 14,689 endpoints indexed. Of those, 14,675 have a payer count present (the endpoint-weighted denominator) and 14,675 have BOTH a payer count AND a call count present (the denominator used for every cohort on this page). Total 30-day requests indexed across that set: 315,911. Rows missing either field are excluded rather than silently coerced to zero.
Both denominators, plainly. Every percentage on this page divides by the denominator co-located next to it — endpoints-with-both-fields for endpoint shares, total 30-day requests indexed for call shares. The two are NOT the same denominator and cannot be added, averaged, or otherwise conflated. Stated together on purpose.
Same helper the market-map block uses. Every figure
on this page comes from
_bazaar_call_weighted_concentration,
the same aggregation function the
market-map
concentration block and the
JSON export render from.
The report and the live surfaces cannot disagree by construction.
Scope caveats (read before citing)
Request count, not dollars.
l30DaysTotalCalls is CDP’s own
count of the number of times an endpoint was requested in the last
30 days. It is not settled USDC value; there is no
price applied on this page and nothing here is
“volume” in a dollar sense.
One-payer endpoints and self-dealing.
l30DaysUniquePayers counts distinct
paying wallet addresses; a single address may be an operator wallet,
a funding wallet, or a canary. A 1-payer endpoint is
indistinguishable from self-dealing from on-chain data
alone — not proof of it, and not proof of a real
external customer either.
One catalog, not a census. The CDP Bazaar catalog is one x402 discovery source. It is Base-weighted (endpoints on other chains that don’t register here are out of scope), and both denominators above are conditional on being indexed by CDP. Endpoints gated by 402 that route through other facilitators or aren’t registered in this discovery surface are not in the sample. The finding is about this catalog; the phrase “the x402 ecosystem” is out of scope.
How to cite this
Headline finding. On the CDP Bazaar snapshot dated 2026-09-07T09:00:55Z, endpoints with ≥5 unique payers are 5.19% of indexed endpoints (with both a payer and a call count) but carry 55.1% of indexed 30-day requests, while endpoints with exactly one unique payer are 74.1% of endpoints and carry 26.8% of requests with a median 30-day call count of 1.
Citation.
On-Chain Agent Intel. "The Call-Weighted x402 Demand Cut — Report 10 (2026-09-03)." Snapshot 2026-09-07T09:00:55Z. https://onchainagentintel.io/blog/x402-call-weighted-concentration-2026-09
License. Published under CC BY 4.0 — free to reuse, including in commercial work, provided you attribute “On-Chain Agent Intel” and link back to https://onchainagentintel.io/blog/x402-call-weighted-concentration-2026-09.
Machine-readable primary dataset. Live market map · JSON export · CSV export — same snapshot, same aggregation helper (call_weighted_supply_side in the JSON), refreshed daily. Cite the live URL for current numbers; cite this URL for the 2026-09-03 snapshot specifically.