Key Takeaways
- Four series, not one: raw native units, entity-adjusted native units, raw dollars, and adjusted dollars are four separate measurements, and a headline number rarely states which one it is.
- Change outputs inflate the raw figure: a UTXO transaction that spends a large output to make a small payment returns the remainder to its own owner, and a naive sum of outputs counts that remainder as value transferred.
- Adjustment is a list of subtractions: adjusted volume = raw volume, less identified change outputs, less transfers between addresses attributed to the same entity, less other movement defined as non-economic.
- The dollar conversion multiplies a quantity by a volatile price: dollar volume is native volume times price, so it moves when either factor moves and the chart cannot tell you which.
- Practical rule: establish which variant you are reading before drawing any conclusion, and if it is denominated in dollars, split the move into a quantity change and a price change before calling it demand.
Who This Guide Is For
Read this page if you have seen a headline claiming that billions of dollars moved on a blockchain in a day and wondered what that figure counted. It counted outputs, most likely, and a large share of those outputs went back to the people who created them.
Educational content. Not individualized financial advice.
What Does On-Chain Transfer Volume Measure?
Transfer volume measures the sum of value appearing in output or recipient positions during a period. That is not the same quantity as the value that changed hands, and on a UTXO chain the difference is usually the larger part of the total.
A Bitcoin-style transaction consumes whole prior outputs and creates new ones. To pay 0.4 units from an output holding 10 units, the spender must consume all 10, send 0.4 to the recipient, and send the remainder back to an address it controls. That returned remainder is a change output. It is a real ledger record and no value left the owner's control when it was created.
A second inflation source is internal movement. Exchanges rebalance between hot wallets, cold storage, and deposit sweeps constantly, and each of those is a valid transfer between two addresses with one beneficial owner.
Plain-language definition
Transfer volume is the total value written into recipient positions on the ledger over a period. Adjusted transfer volume is that total after removing the parts that came straight back to the sender or moved between two accounts the same party controls.
Technical definition
For a UTXO chain, raw native volume is the sum of the values of all transaction outputs created in blocks within the interval. For an account chain, it is the sum of the value fields of all successful value-bearing transfers, optionally including internal transfers produced by contract execution.
How Is On-Chain Transfer Volume Constructed?
Stage one: extract the movements
On a UTXO chain the raw material is the output set of every transaction, which the Bitcoin Developer Guide: Transactions documents alongside the spending conditions attached to each output. On an account chain it is the value field of each transaction plus, if included, value moved by internal calls during contract execution. Both models are covered in the Swoopr Investment guide to UTXO and account models.
Stage two: identify change outputs
No protocol field marks an output as change. The distinction is inferred, and the heuristics are pattern matching rather than proof: an output whose script type matches the inputs while its sibling does not, an unrounded value beside a round one, a never-before-seen address later spent by the same wallet, the position of the output within the transaction. Each can be wrong on a given transaction, and privacy-preserving wallet software is written specifically to break them.
Stage three: strip internal entity movement
Removing exchange shuffles requires knowing which addresses belong to the same entity, which is a labelling problem rather than a ledger fact. Providers cluster addresses using heuristics such as common-input ownership and then attach entity labels, a process described in the Swoopr Investment guide to address labelling and entity clustering. Better labelling lowers reported adjusted volume, because more movement is classified as internal, and labels get revised, so an adjusted series can change retroactively with no chain data changing.
Stage four: attach a price
A daily close applied to a whole day of volume prices a transaction that settled at 02:00 using a number from 23:59. A volume-weighted reference drawn from several venues gives a different answer from a single-venue close, and for a thinly traded token the price itself is an estimate. None of this is visible in the finished chart.
Formula and Measurement Logic
Raw native volume = sum of value in all output or recipient positions
Adjusted native volume = raw native volume, less change outputs, less intra-entity transfers, less other defined non-economic movement
Dollar volume (either variant) = native volume x price
| Series | What it is | Question it can answer | What it cannot do |
|---|---|---|---|
| Raw native volume | Sum of output values in native units | How much value the ledger recorded moving, self-payments included | Cannot indicate economic activity; change and internal shuffles dominate |
| Adjusted native volume | Raw native with change and intra-entity movement removed | How much value plausibly changed owner, free of price | Depends on one provider heuristic set and label set, and is revisable |
| Raw dollar volume | Raw native times price | The headline settlement figure usually quoted | Inherits every raw-series problem and adds price sensitivity |
| Adjusted dollar volume | Adjusted native times price | Economic settlement in a unit comparable across assets | Still a product of two moving series, so not an activity measure |
How Should On-Chain Transfer Volume Be Interpreted?
Pick the variant that matches the question.
- How busy is the ledger? Raw native volume, read alongside transaction count. It includes self-payments, which is acceptable here: a change output still consumed block space and still had to be validated.
- How much value changed hands? Adjusted native volume, with the provider adjustment rules stated. This is the only variant attempting the economic question directly.
- How large is settlement compared with other assets? Adjusted dollar volume, because native units are not comparable across chains. Accept that the series now carries price risk.
Look at the native and dollar series together every time. When they agree in direction, the price contribution is not driving the story. When they disagree, commentary written from the dollar chart alone is describing the market price while claiming to describe network usage.
Step-by-Step Workflow
- Identify which of the four variants the series is. If the source does not say, assume raw dollars, the most commonly published and least informative.
- On a UTXO chain, obtain the raw and adjusted native series together and compute the ratio between them.
- Read the provider adjustment rules: which change heuristics, which entity labels, which movements excluded.
- Check whether the entity label set was revised during your window, which changes adjusted history without any chain change.
- Work in native units for any question about activity over time.
- If a dollar series is required, compute the price change over the same window and separate its contribution from the quantity contribution.
- Check whether stablecoin transfers are included, since a stablecoin series behaves nothing like a volatile-asset series once converted to dollars.
- State the variant, the provider, and the price source alongside any figure you publish.
Worked Hypothetical Scenario
Every figure below is hypothetical and constructed so the arithmetic can be checked by hand.
Part one: reconciling one day of UTXO transactions
Start with a single transaction. A wallet holds one output worth 10.000 units and wants to pay 0.400 units. It must spend the whole output. With a fee of 0.001 units, the change returning to the sender is 10.000 less 0.400 less 0.001, which is 9.599 units. The sum of the outputs is 0.400 plus 9.599, which is 9.999 units. Raw transfer volume records 9.999. The amount that changed owner was 0.400. Raw is roughly 25 times the economic figure on this one transaction.
| Transaction | Value spent from inputs | Output to a counterparty | Output returned to sender or same entity | Fee | Raw volume (all outputs) | Adjusted volume |
|---|---|---|---|---|---|---|
| A: small payment from a large output | 10.000 | 0.400 | 9.599 | 0.001 | 9.999 | 0.400 |
| B: payment consuming most of an output | 2.000 | 1.900 | 0.099 | 0.001 | 1.999 | 1.900 |
| C: exchange consolidating its own funds | 50.000 | none | 49.998 | 0.002 | 49.998 | 0.000 |
| D: payment leaving no change | 5.000 | 4.995 | none | 0.005 | 4.995 | 4.995 |
| Day total | 67.000 | 7.295 | 59.696 | 0.009 | 66.991 | 7.295 |
Raw volume for the day is 66.991 units and adjusted volume is 7.295 units. Raw is 9.18 times adjusted, and the adjusted figure is 10.9 percent of the raw one. Transaction C alone contributes 49.998 units, roughly three quarters of the day headline, while moving nothing between owners.
Notice how differently the four transactions inflate. Transaction B, which spent almost all of its input, added only 0.099 units of change; transaction D added none. The distortion is not a fixed percentage that can be assumed away. It depends on the size distribution of the outputs being spent, which shifts as wallets consolidate or fragment their holdings.
Part two: the same week read four ways
| Series | Week 1 | Week 2 | Change |
|---|---|---|---|
| Raw native volume (units) | 800,000 | 900,000 | +12.5% |
| Adjusted native volume (units) | 200,000 | 190,000 | -5.0% |
| Price per unit | $30 | $45 | +50% |
| Raw dollar volume | $24,000,000 | $40,500,000 | +68.75% |
| Adjusted dollar volume | $6,000,000 | $8,550,000 | +42.5% |
Four answers from one week on one chain. Raw dollar volume rose 68.75 percent, the figure a headline would use. Adjusted native volume fell 5 percent, and that is the only one of the four attempting to measure whether more value changed owner.
The dollar figures are the native figures multiplied by 1.5, the price ratio. For raw dollars, 1.125 times 1.5 gives 1.6875. For adjusted dollars, 0.95 times 1.5 gives 1.425. In both cases the price factor is larger than the quantity factor. The gap between raw native at plus 12.5 percent and adjusted native at minus 5.0 percent is a separate finding: the ledger got busier while economic transfer shrank, which is the signature of rising self-churn or internal exchange movement rather than rising settlement demand.
What Can Make the Interpretation Wrong?
- Reading a raw UTXO series as economic activity. Change outputs are counted as transfers, and the inflation factor is not constant.
- Reading a dollar series as an activity series. It is a product of quantity and price. A 50 percent price move produces a 50 percent dollar-volume move with no change in behaviour.
- Assuming adjusted means correct. Adjustment is a set of heuristics plus a label set, both fallible and both revised over time.
- Comparing one provider adjusted series with another. Different clustering produces different exclusions, so the gap is a methodology difference, not a data error.
- Treating a large single transfer as a market event. Custody migration, collateral posting, bridge escrow, and internal rebalancing all look identical on chain to a sale.
- Double counting bridged value. Locking on one chain and minting on another produces two transfers for one movement.
- Mixing stablecoin and volatile-asset volume. A combined dollar series contains a price-sensitive part and a price-insensitive part with no way to separate them afterwards.
- Ignoring token decimals. A token amount read without its decimal exponent is wrong by a power of ten, and 10^18 is a common one.
Cross-Network and Provider Comparison
On a UTXO chain the dominant correction is change outputs, and it is large by construction: every payment that does not exactly consume its inputs creates one. On an account chain change outputs do not exist, because a transfer moves a stated amount from one balance to another and the sender remaining balance is never written into an output. The adjustment there is almost entirely intra-entity movement, contract-internal transfers, and wrapped-token round trips, which makes it a labelling problem rather than a protocol artifact. The two raw series are therefore not comparable quantities even before any adjustment is applied.
Token transfers add a further split. On an account chain the native asset moves in the transaction value field while tokens move through contract events, so a provider must decide whether the published series covers the native asset only, a defined set of tokens, or everything emitted. A native-asset-only series substantially understates settlement on a chain carrying heavy stablecoin traffic.
Provider differences on one chain come mainly from the entity label set. Better exchange coverage means more movement recognised as internal and therefore lower adjusted volume, which makes a level comparison between providers meaningless while a shape comparison is often informative. For reconciling provider disagreements, see on-chain data quality methodology.
Advanced Analytical Methods
Decomposing a dollar move into quantity and price
Because dollar volume is a product, taking logarithms turns it into a sum: the log change in dollar volume equals the log change in native volume plus the log change in price.
Volume by transfer-size band
Bucketing transfers by size distinguishes a handful of very large movements from broad-based activity. A total that rose because one custody migration occurred looks nothing like one that rose because mid-sized transfers grew.
Stablecoin-only transfer volume
Restricting the series to stablecoins removes the price factor almost completely, since the conversion to dollars is close to an identity. What remains is a much cleaner quantity series, and on a chain carrying significant stablecoin traffic it is often the best available proxy for settlement demand.
The raw-to-adjusted ratio as its own indicator
Publish the ratio of raw to adjusted volume in its own right rather than discarding the raw series. A rising ratio means a growing share of ledger activity is self-payment, consolidation, or internal exchange movement, which is a real behavioural signal invisible in either series alone. Dividing adjusted volume by transaction count is a useful companion: a volume rise with a flat mean means more activity, while a rise with a climbing mean means a few larger movements.
Practical Checklist
- I know whether the series is raw or adjusted.
- I read the provider change-output and entity-exclusion rules.
- I checked for bridge double counting before summing across chains.
- I stated the variant, the provider, and the price source alongside the figure.
Frequently Asked Questions
How is adjusted transfer volume calculated?
Adjusted volume is raw volume less identified change outputs, less transfers between addresses attributed to the same entity, less any other movement the provider defines as non-economic. Nothing in the protocol marks these categories, so both steps rely on heuristics and on an address label set that is revised over time.
Why do two providers publish different adjusted volume for the same chain?
Mainly because their entity label sets differ. A provider with broader exchange coverage recognises more internal movement and therefore reports lower adjusted volume. Change-output heuristics vary too. A level comparison between providers is not meaningful, but comparing the shape of the two series over time usually is.
Why can dollar transfer volume rise while native volume falls?
Dollar volume is native volume multiplied by price, so it moves when either factor moves. If native volume falls 5 percent while the price rises 50 percent, dollar volume rises 42.5 percent, because 0.95 multiplied by 1.5 is 1.425. Reading the dollar chart alone would report growing activity in a week when activity shrank.
Which variant should I use to track network activity over time?
Adjusted volume in native units. It removes the self-payment and internal-movement inflation and carries no price factor, so a change reflects a change in behaviour rather than a change in the market. Use a dollar variant only when comparing across assets, and state that it carries price risk.
Does a very large single transfer mean someone is selling?
Not on the evidence of the transfer alone. Custody migration, collateral posting, bridge escrow, exchange rebalancing, and corporate treasury movement all produce the same ledger record as a sale. Attributing intent requires labelling evidence about the destination, and even then the label may be stale.
Why is median transfer size more informative than the total?
Because transfer size distributions are extremely skewed, so the total and the mean are set by a handful of very large movements while the median describes what a typical transfer looks like. The two tell different stories and frequently move in opposite directions: institutional settlement can lift the total while ordinary usage shrinks. Reporting the median, and ideally several percentiles, alongside the total is what distinguishes a change in the character of activity from a change in one large participant's behaviour.
How do token transfers distort a chain's native transfer volume?
They do not appear in it at all unless deliberately included. Native transfer volume measures movement of the chain's own asset, while stablecoins and other tokens move as contract state changes recorded in event logs. On networks where a large share of economic activity is denominated in stablecoins, native transfer volume can be flat or falling while the value actually settling on the chain grows. Series that combine the two have to state the conversion and the token set they cover.
Does transfer volume include value moved inside contract execution?
Only if the dataset was built from execution traces rather than from transaction records alone. Value moved by a contract during execution is not a top-level transfer and does not appear in the transaction's own fields, so a volume series built the simple way omits it. On chains where a large share of value moves through contracts, that omission is substantial. Whether traces are included is a defining property of a transfer volume series and is rarely stated on the chart.
How does on-chain transfer volume relate to exchange trading volume?
They measure different systems and are not comparable in size or meaning. Trading volume counts matched orders inside venues, most of which never produce a chain transaction, and the same coins can be traded repeatedly without moving. On-chain transfer volume counts settlement between addresses, including deposits, withdrawals, internal custodial movement and payments that have nothing to do with trading. Neither is a subset of the other, and a ratio between them describes the balance of on-chain and off-chain activity rather than a discrepancy.
References
These sources should be reviewed during editorial verification. They support data structures and methods, not the hypothetical conclusion. Provider formulas, chain rules, and APIs can change. Confirm current documentation before publication.
- Bitcoin Developer Guide: Transactions: How outputs are consumed and created, which is where change outputs come from.
- Ethereum.org: Transactions: The account-model value field, which has no change-output equivalent.
- Coin Metrics: Transaction Metrics: Published definitions of transfer-value series, including adjusted variants.
- Coin Metrics: Network Data Glossary: Cross-network definitions distinguishing UTXO output volume from account transfer volume.
- Dune: Address Labels: The entity labelling that intra-entity exclusion depends on.
- Glassnode: Metric Catalog: A second published catalogue for comparing how each variant is named and defined.