Commerce Flows

Classified stablecoin transfers for real commerce-related onchain activity — B2B, B2C, and C2B flows identified via protocol matches and behavioural heuristics across EVM chains, Tron, and Solana.

Not every stablecoin transfer is a payment. payments.commerce_flows is a transfer-level dataset that identifies the ones that are — every transfer that is a legitimate commerce transaction, tagged with its flow type (party_pair, purpose), a confidence level, and the protocol where one is known.

The table is intentionally conservative: transfers that don’t match a rule are excluded, not bucketed as “unclear”. It is not a universal stablecoin classifier and does not try to label everything.

Commerce Flows feeds back into the stablecoins dataset — payments are a sub-category of stablecoin activity. Matched transfers have their category updated from unidentified to payment in stablecoins_evm.activity_enriched.

Why this dataset matters

Most stablecoin volume is not payments — it is trading, bridging, and treasury movement. Commerce Flows isolates the slice that is genuine commercial settlement, attaching an explicit rule and a confidence level to every row. It makes it possible to:

• Measure onchain commerce volume by flow type (B2B, C2B, B2C) and protocol.
• Separate protocol-verified settlement (Tier A) from behavioural inference (Tier B).
• Label stablecoin transfers as payment-related for downstream analytics.

Coverage

Chains: EVM chains, Tron, and Solana. Solana coverage is label-based (Tier A) only.

Tokens: Stablecoins only — the dataset classifies transfers from the upstream stablecoin transfer tables. Unlike payments.card_transactions and payments.agentic_payments, which are not restricted to stablecoins, non-stablecoin settlement does not appear here.

Protocol-verified sources (Tier A):

• Crypto cards — sourced from payments.card_transactions.
• Agentic payments (x402 / MPP) — sourced from payments.agentic_payments.
• Stripe, Shopify Commerce Payments Protocol, Coinbase Commerce, Request Network, and stablecoin payment processors (e.g. BVNK).

Behavioural flows (Tier B): B2C payroll-style distributions, C2B merchant aggregation, and recurring B2B commerce.

Output taxonomy

Party pair	Purpose	Meaning
B2B	Commerce	Business pays business for goods or services.
C2B	Commerce	Consumer pays a merchant or service (checkout, subscriptions where detectable).
B2C	Payroll	Recurring business-to-many-individuals payouts.

Some reports group these flows by audience instead. The audiences overlap on the raw taxonomy — agentic settlement is itself a B2B row — so apply them in precedence order to keep each row in exactly one bucket: agents first (B2B rows with heuristic_id = 'agentic_payments', in full protocol detail in payments.agentic_payments); then businesses for the remaining B2B; and retail for the consumer-facing pairs (C2B and B2C).

Methodology

The pipeline runs in ordered stages: a strict exclusion layer removes non-commerce and non-final activity, then Tier A captures protocol-explicit matches (confidence = high), and finally Tier B applies a small set of high-signal behavioural heuristics to the residual (confidence = medium). Each transfer maps to one output row, and Tier A always takes precedence over Tier B.

1. Exclusion layer

Fires before any classification. A transfer matching any rule is dropped from the eligible universe.

• Via upstream stablecoin labels: issuer mechanics (mint / burn), bridge and wrap infrastructure, CEX legs, DEX / AMM / routing, and DeFi vaults / lending / yield.
• Via behavioural shape: same-entity / self-transfers, round-trip and reverse transfers (treasury reshuffling, wash patterns), A→B→C pass-through routing (avoids double-counting intermediary legs), high-frequency bot wallets, and large one-off treasury repositioning.

Eligibility settles 24 hours after block time. Round-trip and reverse-transfer detection looks 24 hours forward, so a transfer’s eligibility is not final until 24 hours after its block time. Rows older than 24 hours are stable; only the most recent 24 hours can still change as later reverse legs land. This keeps both legs of a round trip out of the dataset, at the cost of a short settling window on the leading edge.

2. Wallet classification

Each address is typed from its full transaction history, with a wallet_balance_tier derived from its rolling 30-day average USD balance.

Wallet type	Balance tiers	Profile
Consumer	dust (under $100), micro ($100–$1K), small ($1K–$10K)	Retail-sized transfers, low counterparty diversity
Business	medium ($10K–$100K)	Recurring inflows from many senders, or outflows to many receivers
Institutional	large ($100K–$1M), whale (over $1M)	High-value counterparties, treasury / corporate flow shapes
Unclassified	any	Average balance under $10 or sparse history — excluded from classified output

3. B2B commerce

Tier	Detector	What it captures
A (labels)	Agentic payments	Agent-to-agent settlement (x402 / MPP), from payments.agentic_payments.
A (labels)	Processor deposits	Stablecoin payment-processor settlement legs (e.g. BVNK).
B (behavioural)	Recurring B2B	Business / institutional sender → receiver, ticket ≥ $2K, directional, low recurrence.

4. B2C payroll

Tier	Detector	What it captures
A (labels)	Request Network	Protocol-mediated payouts through Request payment proxies.
B (behavioural)	Payroll batch	Business-tier sender distributing to ≥5 recipients on a regular cadence, mid-size tickets, no return flow.

5. C2B commerce

Tier	Detector	What it captures
A (labels)	Crypto cards	Card purchase settlements, from payments.card_transactions.
A (labels)	Stripe	Stablecoin checkout settlement into the Stripe MerchantFactory contract.
A (labels)	Shopify CPP	Final settlement legs of the Shopify Commerce Payments Protocol (captures / charges only).
A (labels)	Coinbase Commerce	Final settlement into the merchant / operator.
B (behavioural)	Merchant aggregation	Merchant-like receiver aggregating many payers, then settling out to 1–2 wallets.

Table schema

Each row is one classified stablecoin transfer. Non-applicable fields are NULL. Every row carries tx_hash + transfer_evt_index + transfer_unique_key to join back to the upstream stablecoin transfer.

Column	Type	Description
blockchain	varchar	Chain name
block_month	date	Block month (partition key)
block_date	date	Block date
block_time	timestamp	Block timestamp
tx_hash	varbinary	Transaction hash of the classified transfer
activity_evt_index	bigint	Event index of the protocol anchor event for Tier A matches; NULL for Tier B
transfer_evt_index	bigint	Event index of the underlying token transfer
transfer_unique_key	varchar	Unique transfer identifier; merge key and join key back to upstream
token_address	varbinary	Token contract address
token_symbol	varchar	Token symbol, e.g. USDC, USDT
amount_raw	uint256	Raw token amount
amount	double	Normalized token amount
amount_usd	double	USD value at block time
sender	varbinary	Raw transfer sender
receiver	varbinary	Raw transfer receiver
payment_uuid	varchar	Identifier grouping the legs of a single payment; populated only where a multi-leg payment can be reconstructed, NULL otherwise
sender_wallet_type	varchar	consumer, business, institutional, or unclassified
sender_wallet_balance_tier	varchar	dust (under $100), `micro` ($100–$1K), `small` ($1K–$10K), `medium` ($10K–$100K), `large` ($100K–$1M), `whale` (over$1M)
receiver_wallet_type	varchar	Same vocabulary as sender_wallet_type
receiver_wallet_balance_tier	varchar	Same vocabulary as sender_wallet_balance_tier
party_pair	varchar	b2b, c2b, or b2c
purpose	varchar	commerce or payroll
heuristic_type	varchar	labels_based (Tier A) or activity_based (Tier B)
heuristic_id	varchar	Specific rule that matched, e.g. crypto_card, stripe, shopify_cpp, coinbase_commerce, request_network, agentic_payments, b2b_processor_deposits, c2b_commerce_h3_aggregation, b2c_payroll_h1_batch, b2b_commerce_h_recurring
confidence	varchar	high (Tier A) or medium (Tier B)
protocol	varchar	Protocol or provider for Tier A matches (e.g. x402, stripe, shopify_base_cpp, coinbase_commerce, request_network, or the card brand for crypto cards); NULL for Tier B
dataset_version	varchar	Methodology version; bumped when classification logic changes
_updated_at	timestamp	Pipeline load time (when the row was last written)

Sample queries

Commerce flow volume by party pair and purpose

SELECT
    party_pair,
    purpose,
    heuristic_type,
    SUM(amount_usd)        AS volume_usd,
    COUNT(*)               AS tx_count,
    COUNT(DISTINCT sender) AS unique_senders
FROM payments.commerce_flows
WHERE block_date >= NOW() - INTERVAL '30' DAY
GROUP BY 1, 2, 3
ORDER BY 4 DESC

Protocol-verified flows only

SELECT
    protocol,
    blockchain,
    SUM(amount_usd) AS volume_usd,
    COUNT(*)        AS tx_count
FROM payments.commerce_flows
WHERE confidence = 'high'
  AND block_date >= NOW() - INTERVAL '30' DAY
GROUP BY 1, 2
ORDER BY 3 DESC

Related tables

• payments.card_transactions — Tier A source for crypto card matches.
• payments.agentic_payments — Tier A source for x402 and MPP matches.
• stablecoins_evm.activity_enriched — downstream consumer; exposes category = 'payment' for matched transfers.