On the end of free blockchain analytics
Dune announced this week that its free plan is going read-only. No new queries, dashboards, forks. If you want to write SQL against on-chain data, you need to pay $65/month or more.
A lot of folks are disappointed by this, and reasonably so. With Flipside winding down, this really is the end of an era.
Dune made onchain data legible to a whole generation of analysts and developers, and the community dashboards have become a searchable knowledge base created by hundreds of contributors. It's where I first started playing with on-chain data and finally ended up working in this industry full-time.
The paid plans: for the hours they save on even mild analytics work, is a bargain. But the free lunch is over.
Why the numbers don't work
Say I deploy a contract on Base and want to know how many MAU's I have. Simple question. A few rows in a normal database. But in order to answer that question for me, someone's indexed the complete history of Base. And not just my contract, all of them. And for a company like Dune that means storing all that data in servers around the world every single month, even if you complete your analysis on your contract or move on to other projects.
This mismatch means that the cost of keeping a hosted blockchain analytics platform online and serving queries against it is significantly more expensive than the value that many users get from occasionally using it. Those users aren't doing anything wrong, it's just the economics don't work.
In the announcement of this news from Dune, I expect some companies to try to capture mindshare and market share from analysts, Pitching that their hosted analytics platform is more cost efficient or has some other structural advantage, etc. But ultimately:
Dune has raised some serious amounts of money
The team is smart and has been working on this a long time
They optimized deeply for blockchain data, incl. their own query engine
After all this time, they have decided a free tier doesn't make sense
I'm skeptical that another team will be able to make it make sense on a long term basis. Which means inevitably other hosted platforms will need to charge / monetize somehow too. We've seen this category continuously shrinking between Transpose and Flipside over the past few years.
The fix is a different mental model
The implicit assumption is that we need access to everything, and we need it to be cloud hosted. This warehouse model works well for large team with budgets but less so for basically anyone else; what I think makes the most sense is to invert the thinking here.
Don't start from a warehouse of everything, only the data you need. Don't rent your storage and compute, reuse the one you already have available (ie. your laptop). It's less convenient but AI models can you 95% of the work for us now.
Streamling Blockchain Studio
I built SBS as a local-first analytics workbench that follows this philosophy / mental model. The stack is pretty simple, but deliberately fully open source to prevent any kind of vendor lock in or dependency.
Streamling for EVM data pipelines (logs, transactions, blocks, traces soon) and blockchain goodies like factory contracts and ABI fetching built in.
ClickHouse or SQLite for local storage and querying: ClickHouse for anything heavy or analytical at scale, SQLite for simpler transactional workloads.
Evidence for dashboards and reports, with static site generation to easily host anywhere you want (Vercel, Netlify, Github Pages, etc.).
Reiterating: every component is open source. You own and self-host the data. The first time you run it, Streamling hits your RPC endpoint (Goldsky, Alchemy, QuickNode, Infura, a local node, etc.) and begins pulling the events you specified. Historical data takes a few minutes to a few hours depending on block range and event volume. After that, it's incremental; new blocks stream in as they're produced.
Trade-offs
This is not great for some things. You don't have the pre-indexed history that Dune has already built. If you want to go data from block one on Ethereum Mainnet, you'll pull it yourself. And that's going to take a lot of time. You also don't get the Dune community shared queries for dashboards, the ecosystem of public data that others have already assembled. You also need to worry about how you're going to share and expose your local data store. You can sync it, use a cloud DB (but you've reintroduced recurring expenses), etc. Overall, it's a lot more work.
SBS is great for:
ad hoc point-in-time analytics
people who want to fully own their data stack
teams who are violently concerned with vendor continuity risk
For most other folks, Dune is probably still the best place to go. And there are already folks who are working on sponsorship programs and other ways of retaining free access for people doing important work.
Learning opportunities
There is one other reason to consider exploring something like SBS though, which is the learning opportunity.
Dune has done a lot of work to get us to the fun stuff, the insights. But fundamentally being an onchain analyst is about understanding, and a desire to understand is what I think drove a lot of people to the platform/industry in the first place.
What specifically is happening with that hot new perps platform? Who's using it? For me, that curiosity is extending deeper into the stack, not just to the and analytical methodology, but to how the data got there in the first place. And I'm pretty sure that deeper understanding of how the sausage is made makes me a better analyst too.