Daily Transaction Count

Daily Transaction Count is a node-derived number: for each day it sums the confirmed transactions across every block mined in that 24-hour window. Nothing is estimated and nothing is inferred from an exchange order book — this is what the chain actually recorded, change outputs and consolidation sweeps included. The series runs from 17 Jan 2009 to 14 Jun 2026, with the 30-day simple moving average as the primary trend line and the raw daily count beneath it at lower opacity, where the event spikes live. Price runs muted on the log-scale axis so you can read throughput against the cycle without conflating the two.

Today’s 30-day average is 619,425, the year-over-year change is +60.5%, and the regime bucket reads Expansion. The shape of the whole series is governed by one hard bound: block weight is capped at four million weight units, so with 144 blocks a day the network can only confirm a few hundred thousand ordinary transactions before the mempool backs up and the contest moves to fees rather than count. Everything interesting about this chart happens at, above, or below that ceiling.

The single most important fact this chart records is that base-layer transaction count and price have decoupled. The 2013 and 2017 cycle tops printed record transaction counts alongside record prices — the count was a reasonable adoption proxy. The 2021 top did not. The 2024–26 cycle has not. The chart now answers a narrower question than it once did: how busy is the base layer specifically, with what kind of activity?

Walk the four major tops. In December 2013 the count printed in the tens of thousands as price reached its first $1,000+ region. In December 2017 the count peaked near 405,709 a few weeks before the price top, with the top day itself printing 391,910 transactions at $19,424. In November 2021 the cycle-top day printed roughly 301,025 transactions — below the 2017 top — at $67,145, a price 3.5× higher. And in April 2024, a few days after the halving and the Runes launch the count printed roughly 927,010 transactions, the all-time high, but in a regime where the extra volume was almost entirely small inscription activity rather than wallet payments.

Two structural shifts explain the 2017→2021 decline despite price growth. First, the Lightning Network mainnet launched in March 2018; over the following years it absorbed a meaningful share of small-value payment activity off the base layer. Second, exchange-internal transfers and large-custody batching consolidated what used to be many user-visible movements into a single multi-output transaction. The same payment activity, fewer rows in the count.

Then 2023 and 2024 reversed direction for a different reason. Ordinals in early 2023, the BRC-20 token wave from March 2023, and the Runes launch at the April 2024 halving each pushed the count above its prior payment-only ceiling — but the extra transactions are protocol metadata and small token mints, not user payments. The chart no longer reads as a single coherent “adoption” gauge. It reads as a composite of payments + inscriptions + batching effects, which is more useful but harder to interpret.

For each day d:

count(d) = number of confirmed transactions in blocks mined that day
sma30(d) = mean(count(d-29) … count(d))
yoy(d) = sma30(d) / sma30(d-365) − 1

The headline regime bucket comes from yoy on the 30-day SMA: Expansion above +30%, Contracting below −10%, Stable in between. The 30-day window damps the day-to-day noise from inscription waves; the YoY frame puts each reading in cycle context. Methodology and the source reconciliation are documented on the methodology page.

Three useful lenses, in increasing order of analytic value. Level reading. Today’s 30-day average against the 250–500k payment-era band: above the band, inscription activity is alive; below it, either bear-market quiet or off-chain routing has won. YoY reading. The headline regime bucket; useful as a momentum filter but noisy near the 30-percent threshold. Cycle-aligned reading. The cycle-anchor table below puts every reading in context against price at that anchor — this is where the decoupling shows up unambiguously.

The raw daily line is essential context for the SMA. Inscription days print multi-hundred-thousand-transaction one-day spikes that the 30-day average smooths into a small step rather than a needle. Reading the SMA without the raw line risks treating a one-week token mint as a regime shift.

The cycle-anchor table makes the decoupling explicit. Each row reads the raw count and the 30-day SMA at a canonical date alongside the closing price. The 2017 top, the 2021 top, and the 2024 inscription peak are the three readings most worth lingering on — together they describe the chart’s shifting meaning across three cycles: a clean adoption proxy, then a proxy hollowed out by off-chain routing, then a count inflated by activity that has nothing to do with payments.

Three protocol events account for most of the 2023–26 spikes above the historical payment-only ceiling. Ordinals — inscribing arbitrary data into the witness portion of a SegWit transaction — began in early 2023 and produced the first sustained counts above 500,000 since 2017. BRC-20, a fungible-token experiment built on Ordinals, launched in March 2023 and pushed counts above 600,000 by mid-year. Runes, a more compact token protocol, launched at the April 2024 halving and produced the single-day all-time high in count.

Each of these produces small base-layer transactions in volume: mints, transfers, and metadata commits. They are real consensus-layer activity — the network genuinely processes every one — but they are not the wallet-payment use case the chart historically tracked. Treating an inscription spike as “adoption growth” is a category error; treating it as zero is also wrong, because miners are paid for it and it consumes the same blockspace as payments. The honest reading is that the chart now measures blockspace consumption, full stop, and the analyst has to decide which of its components matter to the question being asked.

Base-layer count is a high-integrity number that answers a question most readers do not actually have. Here is precisely where, and when, it stops meaning what people assume it means.

March 2018 — Lightning made the count blind to small payments. The Lightning Network mainnet launched in March 2018. Channel opens and closes still hit the base layer, but the routed payments between them do not. From 2018 onward, a rising share of genuine spending simply stopped showing up here — which is the largest single reason the 2021 top printed 301,025 transactions against the 2017 top’s 391,910.

August 2017 onward — batching collapsed many actions into one row. After the late-2017 fee crisis, exchanges adopted output batching as standard practice: one transaction with hundreds of outputs replaces hundreds of separate sends. Coinbase, Binance, and others now sweep withdrawals in batches, so the count systematically undercounts user-facing activity in proportion to how much of it flows through custodians — a share that has only grown.

Early 2023 to April 2024 — inscriptions inverted the signal. Ordinals (early 2023), the BRC-20 wave (the 2023-04-23 break-out in the table above), and Runes (launched at the 2024-04-19 halving, peaking 2024-04-23) each generated huge volumes of small data-carrying transactions. The all-time high in count is a token-launch artifact, not a demand peak. Anyone reading the 2023–24 spikes as a wave of new Bitcoin users read them backwards.

December 2017 — the saturation ceiling caps the number entirely. When the mempool is full, the count flatlines near 400–500k no matter how much demand arrives, because the contest moves to the fee market. The flat count through the 2017 blow-off was not flat demand — it was a binding capacity constraint, with the real signal showing up on the fees chart instead.

Before 2020 — the data itself is coarser. The blockchain.info feed is published on a roughly four-day cadence for the early years, so pre-2020 SMA windows are effectively wider and daily-precision readings are not available. Treat that era as a coarse long-run trend, not a day-by-day record.

If you treat on-chain activity as an adoption gauge, stop using this number for that. The 2017→2021 trough below price reflects structural off-chain migration, not declining usage; the 2023–24 spikes above the ceiling reflect inscription mints, not payments. For the adoption question, reach for active addresses for unique participants and ETF flows for institutional demand. Use this chart for blockspace, not adoption.

If you read the network operationally, this is the cleanest congestion signal there is. Sustained counts near 400–500k mean the next increment of demand spills into fees and that batching or Lightning offload becomes economically attractive. Read it against daily fees: rising fees with rising count is demand-bound; rising fees with flat count is capacity-bound; flat fees and flat count despite obvious user-side activity is routing-displacement. The component decomposition is where the analytical value lives — the headline number is only the starting point.