How to Use Dune Analytics to Read On-Chain Data

What Dune Analytics actually is

Dune Analytics is a web-based platform that indexes public blockchain data from networks like Ethereum and Solana, then lets users query that data using a SQL-style language and publish the results as interactive charts and dashboards. Instead of running your own blockchain node and writing code to parse transactions, you log in, type something close to English with table names, and Dune returns a table you can turn into a bar chart, a line chart, or a map.

The model is community-driven. Dune's team maintains the underlying data pipelines and a set of decoded contract tables for major protocols, but the actual dashboards you read are written by independent analysts, protocols, researchers, and curious users. That is the same reason Dune is useful and the same reason it is dangerous: the platform scales because amateurs contribute, but you have no native guarantee that any chart you find is correct, complete, or current.

Think of Dune as something between a search engine and a spreadsheet. The blockchain is the raw spreadsheet of every transaction, and Dune is the tool that lets people write formulas against it and then publish the resulting chart. When someone shares a Dune link on X or in a research note, they are pointing you at a saved query plus a visualization on top of it.

Risks and limits of reading Dune dashboards

The biggest risk on Dune is treating a chart as truth because it looks polished. Dashboards are only as good as the SQL behind them, and SQL is unforgiving: one missing join, one wrong join, and a chart that looks like it shows protocol revenue can quietly be counting the wrong subset of transactions. There is no official scoring or audit step before a dashboard goes live, so a dashboard with thousands of views can still contain a bug.

Latency is the second limit. Dune refreshes its data on a schedule, often within minutes to hours depending on the chain and table, but it is not a real-time terminal. If you are watching a token launch or a liquidation event, the numbers you see on Dune may be 15 to 60 minutes behind the actual chain. For long-horizon charts that does not matter; for trading decisions it does.

Labeling is the third limit and the one most beginners underestimate. Dune has tables that map wallet addresses to human-readable names, like labeling the Ethereum Foundation multisig or a known exchange hot wallet. Those labels come from a mix of official submissions, community contributions, and sometimes self-reporting. A label is a hint, not a verdict. The address tagged "Binance 7" might actually be a market maker that happens to receive deposits from Binance, and the address labeled "Team Wallet" might be a multisig that the team does not actively control anymore.

Finally, Dune shows you what the chain shows you. If a protocol runs activity on a layer 2, a sidechain, or its own app-chain, the mainnet Ethereum or Solana tables may miss parts of it. Reading one Dune dashboard in isolation gives you a slice of reality, not reality.

How queries and dashboards fit together

At the base layer, a Dune query is a SQL script. SQL, which stands for Structured Query Language, is the standard way to ask a database questions like "sum the value of every transaction sent to address X in the last 30 days, grouped by day." Dune ships with hundreds of pre-built tables: raw transaction tables, decoded event tables for popular contracts, ERC-20 token transfer tables, NFT trade tables, and labeled address tables.

A dashboard is simply a collection of charts that come from one or more queries. When you open a dashboard, you are looking at saved query results with a visualization on top. If you click a chart, Dune shows you the underlying SQL, which is one of the most useful features on the platform because it lets you see exactly how the chart was built.

This matters because reading a Dune chart without reading the query is a bit like reading a journalist's article without checking the source links. The skill is not writing SQL from scratch, it is scanning a query to check whether it joins the right tables, filters on the right columns, and uses the right label set. You can learn to do that in an afternoon, even if you have never written a database query before.

Finding and bookmarking reputable dashboards

The fastest way into Dune is not writing anything. It is finding a small set of trusted dashboards and checking them regularly. There are several practical ways to build that shortlist.

• Start with protocol teams. Major DeFi protocols such as Uniswap, Aave, and Lido publish official Dune dashboards from verified team accounts. These are usually the cleanest, best-labeled sources of protocol activity.
• Look for known analysts. A handful of pseudonymous Dune users have built reputations over multiple cycles for accurate, well-commented queries. Their dashboards tend to have "star" counts in the thousands and explicit notes about methodology.
• Check the query, not just the chart. Open a chart, click into the SQL, and skim it. If the query is short, well-commented, and references clearly named tables like erc20_evt_Transfer, it is probably in better shape than a 400-line script with no comments.
• Watch for versioning. Good dashboards are versioned: the author publishes a new query when they change methodology and notes what changed. A dashboard that has not been touched in 18 months may reflect a protocol that has since changed its contracts.
• Cross-check with on-chain explorers. If a dashboard claims a wallet holds X tokens, paste the address into a block explorer like Etherscan or Solscan and confirm. Two minutes of cross-checking catches most labeling errors.

Once you have a shortlist of 10 to 20 dashboards, bookmark them. Dune lets you follow creators and dashboards, and the platform will surface new versions of dashboards you have starred. Treat the bookmark list like an RSS feed of curated on-chain research, not like a market terminal.

Reading wallet cohorts and token-flow dashboards

Two dashboard patterns show up again and again in serious on-chain research: wallet cohort analysis and token flow analysis. Both are worth being able to read at a glance.

Wallet cohort analysis groups addresses by behavior rather than by balance. A common cohort is "wallets that first received ETH in the last 30 days," which approximates new entrants. Another is "wallets that bridged in from a specific chain," which approximates users of a cross-chain feature. When you see a chart titled something like "Active Addresses by Cohort," you are looking at how a group of wallets behaves over time. Rising cohort activity is often read as growing adoption, but the right question is always: which wallets are in this cohort, and could the cohort be inflated by airdrop farmers or Sybil attackers splitting funds across many addresses?

Token flow analysis tracks the movement of a specific token between categories of wallets: exchange wallets, smart contracts, treasury wallets, and "other." A chart showing "USDC flowing into centralized exchanges" is often interpreted as selling pressure, since users typically deposit stablecoins to an exchange when they want to buy other assets or to cash out. That interpretation is reasonable, but it is not automatic. The same inflow could reflect an exchange rebalancing its own treasury.

The reading skill you are building is calibration. After you have looked at a dashboard for a few weeks and compared its claims with what actually happened in the market, you start to develop intuition for which Dune charts lead and which lag, and which are noise dressed up as signal.

Writing a basic query without being a SQL expert

If you want to go from reader to writer, the realistic path is small: copy, tweak, run. Dune ships with a large library of "spellbook" queries that are pre-written by the community for common questions. Pick one that is close to what you want, read it line by line, change one thing at a time, and rerun.

A starter query might count the number of daily transactions on a given chain. The SQL is roughly: select the day, count the transactions, group by the day, order by the day. Once that works, add a filter for a specific contract address to count only the transactions that touched a single protocol. Once that works, add a filter for transactions above a certain size to focus on "whale" activity. Each step is small.

You do not need to learn SQL in depth to do this. You need to learn four ideas: select (which columns to return), from (which table), where (which rows to keep), and group by (how to bucket them). Everything else is decoration. There are also AI-assisted features inside Dune that can generate or modify a query from a plain-English prompt, which lowers the barrier further, though you should still read the generated SQL before trusting its output.

The point of writing a first query is not to publish a viral dashboard. It is to make yourself a more skeptical reader. Once you have struggled with a missing join that doubled your result, you will never look at a slick chart the same way again.

Follow on-chain data the smart way

On-chain data moves fast, and the dashboards that matter most often change their methodology right when the chart becomes interesting. Tracking the right Dune creators, refreshing the right dashboards, and noticing when a query has been updated is a real job if you do it by hand. Zippfeed surfaces Dune and on-chain headlines alongside sentiment scoring, so you can see at a glance whether the latest chart about ETH or SOL flows is being treated as bullish, neutral, or bearish by the wider crypto press, and how important editors actually rate the story.