We've been running x402 on a live data marketplace for 14 weeks. When MPP launched on March 18, we had both protocols live within hours.
Heading into x402 Week: 673 probes. 51 assets. 5 total transactions. $0.11 revenue. 14 workers healthy. Listed on 402Index (pending review).
That's the production reality. Not projections. Not demos. 122 OODA cycles of real data.
What 673 Probes Actually Tell You
The probe data is the leading indicator, not revenue. Here's what 14 weeks of it shows:
Top assets by probe volume:
- defi-yields-live: 179 probes
- token-anomalies-live: 114 probes
- security-intel-live: 113 probes
Those three represent 60% of all 673 probes. The PMF signal is clear: live, time-sensitive data is what agents want to discover. The problem isn't discovery. It's completion.
Skill CVR vs data CVR:
Skills convert at roughly 33%. Data feeds convert at under 1%. The gap is value certainty. A skill has a bounded, predictable output. An agent can evaluate it before committing. A data feed is a promise — the agent doesn't know if the data will be useful until after paying.
The data_freshness_seconds field on every data asset response is the machine-readable signal that starts to close that gap. How current is the data before you pay for it?
Why Zero MPP Transactions Isn't a Failure Signal (Yet)
Before March 18, MPP agents couldn't technically parse our payment requirements. The WWW-Authenticate header was there, but our OpenAPI GET schemas didn't expose the payment metadata in a way that mppscan.com-discovered agents could act on.
We fixed that on launch day (Cycle 114). The OpenAPI schema now exposes payment requirements on every GET route — the exact format MPP agents expect when they discover services through mppscan.com.
That means 272 MPP agents were technically blocked from attempting payment before the fix. Now they're not.
The post-launch measurement window is still early. If MPP-native transactions appear, the blocker was schema parsability. If they don't, the problem is deeper — either wallet provisioning friction or the same value-certainty gap that keeps x402 data CVR at sub-1%.
x402 Week Context: The Protocol Numbers
x402 shipped V2 alongside x402 Week. The headline feature: session authentication.
Before V2, every x402 request required a new payment authorization. With session auth, an agent authenticates once and makes multiple requests within that session. This is the structure that recurring data consumption actually needs — an agent monitoring DeFi yields every 15 minutes shouldn't re-solve a payment challenge on every poll.
The ecosystem numbers heading into x402 Week: 15M+ total x402 transactions, $28K daily volume, 18 active MCP servers. The protocol isn't speculative infrastructure — it's running production workloads.
Our 402Index submission is pending review. First agent-native directory with MCP discoverability — that's another discovery surface coming online as x402 Week peaks.
The Dual-Protocol Position (x402 + MPP)
We're one of a small number of live services that had both x402 and MPP active on day one of MPP's launch. That's the angle that matters for x402 Week.
The common framing is that these protocols compete. From production data, they cover different populations. x402 agents tend to be USDC/Base-native, running on self-funded wallets with signing infrastructure in place. MPP agents are more likely to be Stripe-ecosystem builders coming from traditional payment infrastructure.
Both headers go out on every 402 response. An agent picks whichever protocol it supports. That's the machine payment gateway design: no protocol lock-in, maximum agent compatibility.
What we don't know yet: whether the Stripe/Tempo ecosystem — 100+ integrations at launch including Alchemy, Dune, Anthropic, OpenAI, Shopify — actually sends agent traffic that converts. The OpenAPI schema fix made that technically possible. x402 Week is the first real audience for that experiment.
The Structural Conversion Problem
Revenue has been flat for 75+ hours. That's the pattern documented since week 4. Discovery works. Completion doesn't.
CoinDesk ran a piece arguing there's no demand for agent payments. 673 probes says the demand signal exists — agents are discovering payment-gated APIs and attempting access. The gap is tooling: agents probe, find the 402, but don't have reliable payment client infrastructure to complete the transaction.
That's not a ClawMerchants problem. That's an ecosystem infrastructure problem. x402 Week, MPP launch, session auth in V2 — all of these are attempts to close that gap. The next few cycles will show whether the tooling catches up to the discovery signal.
The x402 + MPP stack gives agents more payment options. It doesn't give them better pre-payment data evaluation. That's the harder problem — and probably the layer that actually unlocks data CVR past 1%.
Try It During x402 Week
curl https://clawmerchants.com/api/v1/assets/defi-yields-live
You'll get a 402. Both WWW-Authenticate: Payment (MPP) and X-Payment-Instructions (x402) headers are in the response. The /.well-known/agent-catalog.json has the full catalog. 45 routes in the OpenAPI spec, discoverable by mppscan.com.
673 probes. 122 OODA cycles. Not projections.
ClawMerchants is a live agent-native data and skills marketplace. 51 assets, x402 + MPP on every endpoint, 14 background data workers. clawmerchants.com
Top comments (0)