Skip to main content

​
Quality Controls Across the Pipeline

Dune is committed to delivering high-fidelity blockchain data that teams can trust for critical analytics, applications, and business decisions. This page describes the automated controls we run on that data, and what each one covers. Data quality isn’t a one-off test. Blocks are validated as they are ingested, core blockchain tables are monitored on a schedule, and curated datasets are tested whenever their code changes.
Monitoring: Status Page · Data Freshness

​
Trusted at Scale

Dune has been indexing and serving blockchain data longer than any other onchain data platform. Our infrastructure processes millions of queries per day across 100+ blockchains, powering analytics for leading exchanges, DeFi protocols, investment firms, and blockchain foundations worldwide. Years of production operation have exposed us to a wide range of chain-specific edge cases, and the controls described on this page reflect that experience.
“Huge credit to the Dune team for their speed, precision, and commitment to making onchain data accessible and actionable.” Luke Judges, Director at Ripple

​
How Our Pipeline Works

Dune’s data pipeline processes blockchain data through three stages: 1. Ingestion: We ingest raw block, transaction, log, and trace data from blockchain nodes. Blocks are validated against a schema for their blockchain before they are stored, and a block that fails validation is retried rather than stored. 2. Monitoring: Scheduled checks scan the blocks tables of our most widely used blockchains for missing and duplicate blocks, and alert our engineers when they find one. 3. Output: Raw data keeps the block and transaction hashes as received from the node, so any record can be compared against a node or block explorer. Decoded data is produced by applying contract interface definitions (ABIs) to raw data. Curated datasets like dex.trades and tokens.transfers apply standardized schemas and business logic to make blockchain data easily accessible for all your use cases.

​
Our Data Quality Framework

​
Block Validation

Before a block is stored, it is checked against a schema defined for its blockchain and against internal-consistency rules, for example that each transaction receipt belongs to the block and that the block’s transactions and receipts correspond to one another. A block that fails is retried, and our engineers are alerted if it cannot be retrieved and validated. We keep each block’s hash and parent hash, and each transaction’s hash, as received from the node rather than recomputing them. Because these identifiers are published by the network, you can compare any record in Dune against the same record from a node or block explorer of your choice.

​
Gap Detection

Every 30 minutes, automated checks scan the blocks tables of our most widely used blockchains, including Ethereum, Base, Arbitrum, Optimism, Polygon, BNB Chain, Avalanche, Bitcoin, Tron and Solana, for missing block numbers. Any gap found alerts our engineers so it can be investigated and, where the data is available, backfilled.

​
Duplicate Detection

For most of those blockchains, the same checks also detect duplicate block records and alert our engineers.

​
Re-org Handling

Blockchain reorganizations can replace recently produced blocks. For most chains, Dune limits exposure to them by indexing a block only once it is a set number of blocks behind the chain tip, or once the node reports it as safe or finalized. The margin is set per chain based on how that chain reaches finality.

​
Token Metadata

Token metadata (names, symbols, decimals) for ERC-20 tokens is maintained in tokens.erc20, part of Spellbook, Dune’s open-source transformation layer. It is updated as contributions are reviewed and merged.

​
Dataset Tests

Curated datasets carry automated tests on their key columns, such as not-null and uniqueness checks. These tests run whenever a dataset’s code changes, so a change that breaks them is caught before it ships.

​
Monitoring & Transparency

Real-time visibility into data health:

Data Freshness

Check current ingestion latency for all supported chains and curated datasets.

Status Page

Subscribe to alerts for any incidents affecting data availability or quality.
If you discover a data quality issue, let our team know. We investigate every report and prioritize fixes based on impact.
“Dune has by far the largest and most comprehensive dataset on the market.” Dan Smith, Senior Research at Blockworks

​
Next Steps

Enterprise Data Quality Guarantees

For custom SLAs, dedicated support, and priority issue resolution, talk to our enterprise team.

Build Validated Pipelines on Dune

Run your own dbt projects on Dune’s data warehouse. Full incremental model support, automated testing, and CI/CD.