On-Chain Metrics

Mempool and Transaction Fee Spikes

Why unconfirmed transactions pile up in the mempool, how limited block space turns waiting into a fee auction, and what a spike does and does not mean.

On this page
  1. What the mempool is
  2. Why a queue forms
  3. How a queue becomes a fee auction
  4. What triggers a spike
  5. Stuck and replaced transactions
  6. Reading the data
  7. Practical caution

What the mempool is

When you send a Bitcoin transaction, it does not enter the blockchain instantly. It is first broadcast to the network’s nodes, which check it against the rules and hold it in a waiting area, usually called the mempool (short for memory pool). Miners pick transactions from this area when they build the next block.

There is no single shared mempool. Each node keeps its own, shaped by its settings and by which transactions it has heard about. Websites that show “the mempool” are displaying one node’s view or an aggregate, so figures differ slightly between sources.

Why a queue forms

Block space is limited. Bitcoin blocks arrive roughly every ten minutes on average, and each can hold only a limited amount of transaction data. Block production does not speed up when demand rises, because the difficulty adjustment is designed to keep the pace steady over time. If more transactions are broadcast than fit into the next few blocks, the surplus waits.

Arrival of blocks is also random. Sometimes two blocks come within a minute and sometimes there is a gap of half an hour. During a gap, the mempool keeps growing, and it shrinks again as blocks clear it.

How a queue becomes a fee auction

Miners are free to choose which waiting transactions to include, and they earn the fees attached. A rational miner therefore prioritises transactions paying the most per unit of space. The relevant measure is the fee rate, typically quoted in satoshis per virtual byte, not the total fee in dollars. A large transaction needs a higher total to achieve the same fee rate as a small one. For background on small units, see the article on satoshi units.

Here is a purely illustrative example. Suppose a block has room for 1,000 units of space, and waiting transactions each need 100 units, ten of them at a fee rate of 20, ten at 10 and ten at 5. The block fits ten transactions, so the ones at rate 20 get in first. Those at 10 wait for the next block, and those at 5 wait longer. If new transactions keep arriving at rate 20 or higher, the lower bidders may be left waiting for many blocks. A sender who wants quick confirmation has to outbid the competition, so the going rate rises. The numbers are invented and only show the logic.

What triggers a spike

Fee spikes usually come from a sudden increase in demand for block space against a fixed supply. Common triggers include:

  • Price swings and heavy trading. When markets move sharply, many people shift coins between wallets and exchanges at once.
  • Popular launches or new token formats on the chain. Activity that creates large numbers of transactions can crowd out ordinary payments.
  • Batch events. Exchanges or services may send out many withdrawals together.
  • Low supply of space. A run of slow blocks can let the queue build.

On Ethereum the same dynamic exists, though the fee mechanism differs, as covered in the article on gas fees.

Stuck and replaced transactions

If a transaction pays too little, it may wait for a long time or be dropped from nodes’ mempools when they reach their size limits. Many wallets let the sender raise the fee on an unconfirmed transaction, often called replace-by-fee, or let a recipient or another transaction help it along. Waiting times are discussed from a user’s perspective in the piece on confirmation times.

Because low-fee transactions can be evicted, a sudden drop in mempool size is not always a sign that the queue has cleared. Some of it may simply have been discarded.

Reading the data

Mempool and fee charts are useful for what they directly show: how many transactions are waiting, how much space they need and what fee rates it takes for confirmation within a few blocks. They are less reliable as a source of broader stories.

  • High fees do not by themselves mean strong adoption. One burst of automated or speculative activity can be enough.
  • Low fees do not mean the network is unused. They may mean demand is easily absorbed or that activity has moved to layer two systems or to custodians, where transfers are recorded internally.
  • Fees are lumpy. Short bursts are common and often fade within hours or days.

Fees also matter to the long-run design of the system, because fee revenue is expected to become a larger share of miner income as the subsidy halves over time.

Practical caution

Anyone sending a transaction during a spike should expect to pay more or wait longer, and should check the fee estimate in their wallet before confirming. Spikes also attract scammers who exploit urgency, so be wary of unsolicited messages offering to “speed up” a payment.

Cryptocurrency prices are volatile, and nothing about mempool conditions reliably predicts them. This article is educational and is not financial advice.

The complete guide to on-chain metrics

An Introduction to On-Chain Analysis
Read the guide

Keep exploring

Guides from other topics

All topics →