Why x402 Numbers Differ
x402 dashboards can disagree because they count different units, follow different money flows, cover different activity, apply different filters, or use different time windows. Double-counting is one possible cause, but a gap between totals does not establish its cause. Compare these five parts of the method before comparing the numbers.
This guide explains differences between measurement methods. For the rules implemented by x402 List, read How x402 List Measures.
1. Counting units and payment identity
One blockchain transaction can contain several payments; one payment can produce several token transfers. A transaction hash, a transfer event, a payment, and an API request are different counting units. Counting distinct hashes can undercount payments in a batch; counting every transfer can count several movements for one payment.
A comparable count needs a stated unit and a payment identity rule. That rule must distinguish payments within a batch, recognize retries of the same record, and normalize aliases for the same network. Preventing a record from being imported twice does not, by itself, identify all the transfers that belong to one economic payment.
2. Money flows and fees
Escrow deposits, seller payouts, and refunds are different movements. Adding a deposit to its subsequent payout can inflate the volume of one purchase. A successful claim in a batch settlement is not necessarily a payment to the merchant; the distribution must be identified separately. The PayAI batch settlement documentation describes this distinction.
State whether volume is the buyer's gross payment or the merchant's net receipt, how intermediary transfers are collapsed, and how refunds are treated. Identify facilitator fees and network costs separately from the merchant price. A gross figure and a net figure can disagree even when both count a payment only once.
3. Coverage and attribution
Totals depend on the networks, mainnet or testnet activity, assets, facilitators, wallets, and endpoints a source includes. USD amounts reported by a dashboard and quantities of USDC transferred on-chain are different units; any conversion or nominal-parity convention needs to be stated. A token transfer alone does not prove that a particular API endpoint was purchased or delivered its response.
Facilitator shares can overlap when the same activity is attributed to more than one facilitator. Adding those shares does not produce a unique ecosystem total. The total needs a deduplication rule across attributions. For example, x402stats' methodology declares its facilitator market shares non-exclusive.
4. Raw volume and filtered volume
Raw volume reports activity within a source's indexed scope. Labels such as organic or real apply additional rules, which can differ between sources. A comparable report shows raw and filtered figures together and states the filter rules, their version, and which categories they exclude. The difference between raw and filtered volume is not automatically all wash trading: filters may also remove developer tests or other categories.
Duplicate records and wash trading are different problems. Duplicate records can count the same payment more than once; wash trading can consist of separate payments between wallets controlled by the same operator. Removing duplicates does not remove those payments. Organic or real labels do not guarantee that different buyer wallets represent independent customers.
5. Time windows and freshness
A comparison needs the same start and end instants, with the start included and the end excluded, expressed in UTC. A rolling 30-day window differs from 30 complete calendar days. If a source does not publish its boundaries or timezone, equal period labels do not establish equal coverage.
Check the last covered block or data timestamp as well as the refresh time. A recent fetch can return old underlying data. Cache delays, missing historical backfills, and changes to the measurement method can also affect comparisons. A published fix does not establish that all historical records have been corrected.
Sources
Evidence basis: public-source research collected on 11 October 2026. These references document methods and historical issues; they do not establish the current behavior of every dashboard or quantify the cause of any particular gap.