The three numbers people misread
Bitcoin explorers are accurate. The confusion comes from three fields that mean something different from what they appear to mean, and every one of them has been used to mislead someone.
The amount
Bitcoin has no accounts and no balances. It has unspent transaction outputs, and a transaction must spend them whole. If you hold a single 2 BTC output and want to send 0.01 BTC, the transaction consumes the whole 2 BTC and creates two new outputs: 0.01 to the recipient and roughly 1.99 back to an address you control.
An explorer reporting "total output" therefore shows about 2 BTC. That figure is correct and it is not what you paid. Screenshots showing enormous apparent transfers are frequently just ordinary payments from well-funded wallets, and this is worth internalising before someone shows you one as evidence of something.
The fee
Miners select transactions by fee per unit of block space, not by total fee. A large transaction paying a high total can be less attractive than a small one paying less. The unit that matters is satoshis per virtual byte, and a wallet that shows you only a total in pounds is concealing the number that determines your position in the queue.
The confirmation count
One confirmation means your transaction is in a block. Each subsequent block makes reversal exponentially more expensive. There is no universal threshold — the number that matters is whatever your exchange or counterparty requires, and they publish it.
Fees, and why mempool.space wins on this chain
Bitcoin's block space is limited and demand for it varies enormously. During quiet periods a transaction paying one satoshi per virtual byte confirms in the next block; during congestion the same transaction may wait days.
This is the single most common reason people open a Bitcoin explorer, and it is where the difference between explorers is starkest. Most of them will tell you a transaction is unconfirmed and stop there — which is precisely the least useful word available at that moment, because the question you have is how long.
mempool.space answers it. The landing page is the queue: projected blocks to the left, each labelled with the fee range it will likely contain, so you can see the boundary your transaction needs to clear. Within two minutes you know whether to wait or to act.
The minimumFee band is the one worth knowing about. Below that floor, nodes will not relay your transaction at all — not "slow" but "may never propagate". Wallets offering a free-text fee field will let you go below it.
If it is stuck
Check the RBF flag. If replace-by-fee was signalled, bump the fee from your wallet. Most modern wallets signal it by default.
If RBF is off
Child-pays-for-parent: spend an output of the stuck transaction with a high enough fee to make the pair attractive together.
What not to do
Do not rebroadcast through a third-party service. It cannot help, and the accelerator products advertising into this moment mostly sell what your wallet already does free.
Our guide on unconfirmed transactions covers the other causes, including the ones a fee bump will not fix.
What you give away by looking
Bitcoin's transparency is deliberate and total. Every transaction since January 2009 is public, permanent and analysable, and it will be analysable with better tools in ten years than exist today.
The pseudonymity people assume they have is not a property of the protocol. It is a temporary consequence of nobody having bothered to look. Once an address is linked to you — through an exchange withdrawal that required identity documents, a published donation address, a purchase with a delivery address, or simply reusing an address — the entire history connected to it becomes linked too, retroactively.
Explorers participate in this in a smaller but real way. Looking up your own address tells the operator that someone at your IP is interested in it. Do it repeatedly and the inference is not subtle.
The practical mitigations, in increasing order of effectiveness: use Tor, and prefer an explorer with an onion service — Blockchair, mempool.space and Blockstream.info all publish one. Use an explorer that works without JavaScript, which removes a class of fingerprinting. Or run your own, which removes the question entirely.
Our guide on what an explorer learns about you covers the threat model, and the self-hosting guide covers what running your own actually costs — which for Bitcoin is comfortably over 700 GB and several days of initial sync.
The keyless APIs, and why they are worth protecting
Bitcoin is unusually well served here, and the situation is better than on any other chain we cover. We tested each of these with live requests from a clean machine rather than reading a pricing page, because the gap between the two is where most API comparisons go wrong.
Blockstream Esplora at blockstream.info/api — no key, no registration, no email address. Mainnet, testnet, signet and Liquid on the same endpoint shape, which makes development straightforward.
mempool.space at mempool.space/api — keyless REST plus a WebSocket for live updates, Esplora-compatible so code is portable between the two.
Blockchair at api.blockchair.com/bitcoin — keyless at modest volume, with the query layer that nothing else offers free.
Blockchain.com's legacy endpoints — the /q/ family returns a single value as plain text, which for a shell script or a status board is about as convenient as an API gets.
All four send Access-Control-Allow-Origin: *, which means they can be called directly from browser code with no server in between. That is a shrinking capability: Etherscan has always required a key, and Blockscout moved behind one on 1 July 2026. Bitcoin's keyless APIs are a public good maintained largely at somebody else's expense, and the polite response is to cache aggressively and to self-host if your usage becomes significant.
Ordinals, Runes and the metaprotocol problem
Since 2023 Bitcoin has carried inscriptions — arbitrary data embedded in witness fields, interpreted by the Ordinals convention as individually identifiable satoshis with attached content. Runes followed as a fungible token protocol built on the same idea.
The critical thing to understand is that Bitcoin itself has no concept of either. These are conventions layered on top, and an explorer showing you an inscription is applying an interpretation rather than reading a protocol feature. Different explorers can and do interpret edge cases differently.
OKLink was early and remains among the most complete, with inscription content rendered and collections grouped. mempool.space added support later and handles the fee implications well — inscriptions are large, so they compete for block space in a way that affects everyone.
If you hold anything of value in either protocol, cross-check between two explorers before acting. A disagreement about a metaprotocol is not the same kind of event as a disagreement about a transaction, and it is considerably more likely.