MASTR · CRYPTO & WEB3
What the dashboard counts: addresses, transactions, volume and people
A chain can be busy without every transaction representing a new person, a successful trade or an economic gain.
Les chapitres et schémas techniques sont en anglais. La navigation existe en sept langues.
What is the denominator behind the headline?
Start with the unit of measurement
A transaction count, active-address count and user count are not interchangeable. One person can use several addresses; an exchange can represent many people behind a small set of addresses. An automated system can submit repeated transactions. Some transactions fail while still consuming resources. Before comparing chains or applications, define exactly what was counted and what was excluded.
Volume is turnover, not new money
The same assets can trade repeatedly during a period. Gross trading volume can therefore be much larger than net new capital. Routing through several pools can also create multiple legs for one user action. Whether a dataset counts user intent, route legs, transfers or venue fills changes the result. A chart needs its methodology beside it, especially when used to infer demand or economic value.
Success is not profitability
A transaction that executes successfully may realise a loss after fees or be part of a larger strategy. Failed transactions can still incur costs under the relevant chain rules. Revenue collected by a terminal or infrastructure provider is not the same quantity as its users’ profit. The Solana bot study is useful partly because it separates implementation categories from observed onchain behaviour instead of equating activity with success.
Selection can change the story
A study of successful migrations excludes launches that never migrated. A leaderboard excludes inactive or failed accounts under its rules. An average among survivors may not describe the typical initial participant. Preserve the observation window and sample construction, and avoid extrapolating a measured rate to a much broader population. A carefully bounded result is more valuable than an impressive but unsupported universal percentage.
Build a reproducible measurement note
For a chart, state the chain, programme or contract set, start and end times, included transaction types, treatment of failures, unit conversions and deduplication rules. Report the source and retrieval time. Where the dataset cannot identify people or beneficial owners, use address or account rather than user. This small discipline prevents a large fraction of misleading growth claims.
Exemple expliqué
One bot submitting 1,000 transactions from ten addresses does not demonstrate 1,000 users or ten people. A service batching payments for 1,000 customers into a small number of transactions shows the opposite problem. The ledger action and the human activity require different evidence.
Questions à retenir
- State the counted unit.
- Disclose exclusions and failed transactions.
- Do not rename addresses as people.