Every stablecoin transfer classified by economic intent, every holder labeled by venue type, across 38 chains and 200+ stablecoins.

The dataset has two label dimensions.

Address-level labels — every address in the enriched balances table is bucketed into one of 11 economic-function categories: DEX, Lending, Bridge, Yield, Issuer, Treasury, CEX, Burn, unidentified smart contract, whale, and unidentified.

Activity-level labels — every transfer in the enriched activity table is classified into one of 9 activity categories (flashloan, dex_swap, issuer, yield, cex, lending, bridge, dex_liquidity, unidentified) using a deterministic precedence waterfall.

These are not Dune's general-purpose public labels. They're purpose-built for stablecoin analysis — distinguishing dex_liquidity from dex_execution, lending_liquidity from lending_execution, and issuer_operations from issuer_treasury, distinctions that generic labels don't make.

Labels are recomputed on every pipeline run from per-protocol models (Uniswap, Curve, Aave, Morpho, major bridges, the CEX address universe, and more), so new activity flows through automatically. Entity labels and manual attribution are updated periodically as new relevant activity is uncovered.

Coverage goes beyond a single category tag:

• Whale EOAs (unidentified externally-owned accounts holding more than $10M in stablecoins) are flagged as a distinct unidentified_whale category, separate from generic unlabeled EOAs.
• Treasury and issuer treasury wallets are tagged separately from operational addresses.
• Named protocol contracts — specific Aave markets, specific Uniswap pools, specific bridge contracts — are individually labeled, not grouped as “DeFi.”

Named-entity attribution for centralized exchanges and large institutional holders (e.g. identifying a specific Coinbase hot wallet) is in active development. Until that ships, customers should expect category-level tagging plus whale flagging, with named-entity coverage rolling out incrementally.

Foundation tables give you raw transfer events and daily balance snapshots. The enriched tables add two layers of context on top of that infrastructure.

balances_enriched adds an economic-function category and subcategory to every address holding a stablecoin. Every dollar of supply is bucketed by where capital is sitting and what role it serves.

activity_enriched classifies every individual transfer by what action it represents — a DEX swap, a lending supply or repay, a bridge deposit, a CEX deposit/withdraw, a flashloan leg, an issuer/PSM flow, a yield deposit/withdraw.

The practical difference: raw transfers tell you that value moved. The enriched tables tell you why— someone supplied to a lending market, bridged out, swapped on a DEX, or deposited to a CEX. That's the difference between net balances and an analyzable economic dataset. Foundation and enriched tables are both part of the Enterprise stablecoin dataset.

Steakhouse Financial was the embedded research and methodology partner on the dataset. Their work covered:

• Deep protocol knowledge for the issuer category — Maker's PSM, ALM, and transmuter mechanics, and the correct way to classify issuer flows generally.
• The activity taxonomy itself, including the edge cases that separate a dataset that looks right from one that is right.
• QA and testing throughout the build.

Steakhouse's role was methodology, classification design, and validation.

Yes — this is one of the dataset's core strengths.

The dataset is intentionally unopinionated on adjusted volume. Other providers ship a single pre-baked formula. Dune exposes the activity classification underneath, so you define what counts as “real” volume — for example, excluding dex_swap, flashloan, and intra-CEX transfers.

Because everything sits in DuneSQL with full composability, the same methodology applies to every issuer in the dataset at once — USDC, USDT, RLUSD, PYUSD, EURC, BRL stablecoins, and more — for genuine apples-to-apples comparison. With opinionated providers, you get the vendor's number per token and have to trust it. With Dune, you own the methodology.

The building blocks are all in the dataset today:

• The activity table lets you decompose a balance move into its drivers — for example, “supply on chain X dropped $200M due to bridge_deposit and cex_deposit flows on these dates.”
• The address labels identify which protocol or category drove each leg.
• Everything is queryable in DuneSQL and can be wired into dashboards, alerts, or scheduled briefs.

A packaged monitoring layer — alerts, named-entity attribution, and leadership-ready briefs — is on the roadmap and is directly tied to the named-entity work described above.

Where Dune is distinctly stronger:

• Transfer-level activity classificationacross 9 categories. No other provider does this natively — Artemis classifies at the app/address level; Allium and Blockworks don't classify transfer activity at all.
• Chain breadth — ~37 chains with full data, versus ~17–24 for competitors.
• First-class daily balance snapshots with category breakdown, not just supply charts.
• Composability — the enriched stablecoin tables are queryable directly in DuneSQL and joinable with any other Dune dataset: DEX trades, lending events, bridges, prices, decoded contracts. Competitors deliver fixed tables or API endpoints, which constrains what analysis is possible.
• Methodology transparency — every classification rule is documented and inspectable.

Where other providers are currently ahead:

• Allium and Artemis ship a pre-computed adjusted volume number out of the box. Dune doesn't — by design, since adjusted volume requires opinionated choices customers may want to make themselves. It's derivable from the activity categories.
• Artemis offers a survey-based B2B/P2P/B2C payments classification framework — a different lens (payment type) than Dune's (onchain activity type). Similar classifications are in development.

Yes. The address labels, activity categories, and economic-function taxonomy are all embedded inside the enriched tables — there's no separate labels SKU. An Enterprise subscription covers foundation and enriched tables, including the full classified view.

Note: any third-party named-entity label set layered in later — for high-confidence attribution to specific institutions — may be packaged as a private add-on, since upstream licensing flows through separately.