How to use Dune Analytics for NFT insights? (Custom dashboards)

Understanding Dune Analytics Core Architecture

1. Dune Analytics operates as a decentralized data platform built on Ethereum and other EVM-compatible chains, enabling users to query on-chain data via SQL.

2. It ingests raw blockchain events—transfers, mints, approvals—and structures them into standardized tables like nft_transfers, erc721_evt_transfer, and erc1155_evt_transfer.

3. Each NFT collection is mapped to its contract address, allowing analysts to isolate behavior by project without relying on third-party APIs or off-chain metadata feeds.

4. Queries run against immutable historical snapshots, meaning results reflect verified chain state at block height—not estimated or interpolated values.

5. The platform supports recursive CTEs and window functions, permitting advanced cohort analysis such as first-time buyer retention or mint-to-sale latency distribution.

Building Collection-Specific Dashboards

1. Start by identifying the primary contract address of an NFT project using Etherscan or OpenSea’s collection page, then verify it against Dune’s ethereum.contracts table.

2. Create a base query filtering nft_transfers for that address, adding joins to ethereum.blocks for timestamp normalization and ethereum.transactions for gas fee context.

3. Aggregate metrics like daily unique buyers, average sale price in ETH and USD, and transfer volume by wallet tier (e.g., whales vs. micro-traders) using conditional grouping.

4. Visualize outputs with line charts for floor price trends, bar stacks for marketplace distribution (Blur vs. OpenSea vs. X2Y2), and pie slices for holder concentration across top 100 wallets.

5. Pin critical widgets—such as real-time mint count or active listing rate—to a custom dashboard, ensuring live updates every 15 minutes without manual refresh.

Leveraging Community Queries and Forking Logic

1. Browse Dune’s public explorer for queries tagged “NFT”, “Blur”, or “ordinals”, then inspect the SQL syntax used by experienced contributors like @dune_hq or @nft_analytics.

2. Fork a high-engagement query—such as “Top 10 Most Traded ERC-721 Collections Last 7 Days”—and replace the hardcoded contract list with a dynamic subquery pulling from nft.tokens where category = 'art'.

3. Modify time filters to use relative blocks instead of absolute timestamps so dashboards remain relevant across forks and chain reorgs.

4. Add comments inside the SQL using -- to document assumptions, like excluding wash trades flagged by known bot addresses stored in labels.balances.

5. Share the revised query with permission settings set to “Public” or “Team Only”, depending on whether internal trading desk alignment is required before publishing.

Integrating On-Chain Metadata and Trait Analysis

1. Join nft.token_metadata to extract traits, rarity scores, and image URIs directly from token-level storage, bypassing unreliable IPFS gateways.

2. Use string parsing functions like regexp_replace() to normalize trait values (e.g., converting “Background: Blue” to “blue”) before counting frequency distributions.

3. Compute rarity-weighted floor prices by calculating the median sale price of tokens ranked in the top 5% of trait scarcity scores within the same collection.

4. Cross-reference metadata updates with ethereum.logs to detect when new traits are introduced mid-collection via proxy upgrades or dynamic rendering contracts.

5. Export raw trait-level sale histories to CSV for external clustering models that identify undervalued combinations based on historical velocity and bid depth.

Frequently Asked Questions

Q: Can Dune track NFTs minted on non-EVM chains like Solana or Bitcoin Ordinals?A: No. Dune only indexes Ethereum, Polygon, Optimism, Arbitrum, Base, and Gnosis Chain. Solana data requires Helius or Birdeye; Ordinals require Hiro or Mempool Explorer.

Q: Why do some transfers appear duplicated in nft_transfers?A: ERC-1155 batch transfers emit one log per token ID. Dune flattens these into individual rows, which may inflate counts unless grouped by transaction hash and block number.

Q: How often does nft.token_metadata refresh?A: Metadata ingestion occurs once per token on first observed transfer. Updates to existing tokens are not retroactively fetched unless manually triggered via Dune’s metadata backfill tool.

Q: Is there a way to filter out known wash-trade wallets globally?A: Yes. Dune maintains the labels.wash_traders table, updated biweekly. Join it using wallet_address to exclude those addresses in any sales or holder analysis.