The problem it exists to solve
When you paste an address into a hosted explorer, you make a disclosure. You tell that operator that someone at your IP address, with your browser fingerprint, is interested in that specific address. Do it repeatedly with an address you control and the inference is trivial.
This is not hypothetical harm, and it is not solved by a privacy policy. It is solved by not making the request — which means the explorer has to run where you are.
Otterscan takes that seriously enough to have no hosted instance. There is nowhere to visit. You install it, point it at your own Erigon node, and the entire system runs on hardware you control. No operator, no logs, no terms of service, no acquisition risk, no policy change.
Why Erigon specifically
Otterscan is deliberately thin. Rather than building its own indexer, it uses Erigon's — Erigon is an Ethereum client designed around efficient storage and rich tracing, and it already exposes most of what an explorer needs through custom RPC methods.
That design decision is what makes Otterscan small enough to be auditable and fast enough to be pleasant. There is no separate index to build and keep in sync; the node is the index.
It also means the requirement is firm. You need Erigon, and for full history you need it in archive mode. That is a multi-terabyte commitment on fast storage and a sync measured in days. There is no way around it, and the project does not pretend otherwise.
The honest cost
An Ethereum archive node is a serious piece of infrastructure — several terabytes on NVMe, considerable RAM, and days of initial sync. For most individuals that is the wrong trade for private lookups alone. It becomes obviously right if you already run a node for staking, for a wallet back end, or because you would rather not trust an RPC provider either.
Where it is genuinely better, not just more private
Privacy is the headline, but Otterscan's transaction tracing is independently excellent, and that is worth saying because it means running it is not purely a sacrifice.
Erigon's tracing support is thorough, and Otterscan surfaces it well: the full internal call tree of a transaction, with values, gas and revert reasons at every level. When a contract call fails several layers deep, this shows you exactly where, which even Etherscan does not always make easy.
It is also simply fast. Local disk beats a remote API by a wide enough margin that the difference changes how you work — you stop batching your questions because each one is free.
What it will not do is enrichment. No token logos, no price conversion, no "this address belongs to a major exchange" labels. Those require external data sources, and external data sources are exactly what Otterscan is designed not to have. The output is the chain, unadorned.
Who should actually run it
Anyone already running Erigon. The marginal cost is close to zero and the benefit is immediate — this is unambiguously the right call.
Anyone whose lookups are sensitive. Researchers, people handling significant holdings, anyone whose interest in an address would itself be revealing. The node cost is real but so is the exposure it removes.
Developers who want unlimited queries. No rate limits and no key means you can loop as hard as your hardware allows, which for analysis work removes an entire class of friction.
Everyone else should not. If you check a transaction once a month, run Blockstream.info or Blockchair over Tor and get most of the privacy benefit for none of the cost. Our self-hosting guide lays the options out side by side, and being honest about who should not self-host is part of that.
What our reference tests showed
Testing Otterscan meant running it, which is itself the finding: there is no way to evaluate this explorer without committing to the node.
Against an Erigon instance, our EVM reference items resolved faster than any hosted explorer we tried, by a margin large enough to change working habits. Queries that we would normally batch to be polite became free, and we stopped thinking about rate limits entirely.
The internal call tree on our failed-contract-call item was the best rendering we saw anywhere, including Etherscan: every level of the call stack with its value, gas and revert reason, in one view rather than behind a tab.
What was absent was equally instructive. No token logos, no price conversion, no address labels. An address belonging to a major exchange looked exactly like any other address. Every one of those omissions is a third-party data source that Otterscan deliberately does not have, and after a few days we stopped noticing.
The honest caveat from testing: the node was the project. The explorer took minutes to set up; getting the node to a usable state took days.