Transaction Graph

The transaction graph is the network of connections that Bitcoin transactions create between addresses and outputs over time. Because every spent output becomes an input somewhere else, the entire ledger forms one enormous directed graph in which coins flow from output to output, block after block, forever. Chain-analysis firms study this graph to cluster addresses into wallets and to follow funds from source to destination. Understanding the transaction graph is the first step to understanding both blockchain surveillance and every privacy tool built to resist it.

It helps to appreciate the scale and the asymmetry involved. The graph now spans over a billion transactions accumulated across more than fifteen years, and it grows with every block — yet analyzing it is embarrassingly parallel work that commercial firms, academics, and hobbyists all perform routinely on ordinary hardware. The asymmetry is temporal: you make privacy decisions in the present, but analysis happens in the future, with better heuristics, more identified endpoints, and cheaper compute than exist today. Coins you moved carelessly in 2020 can be re-examined in 2030 under techniques not yet invented. That is why graph hygiene is best treated as a habit practiced by default rather than a remedy applied after something becomes sensitive.

What the graph is made of

Formally, the nodes of the graph are transactions and the edges are the outputs that connect them: each UTXO created by one transaction and consumed by another is a directed edge between the two. Layered on top are the addresses that outputs pay to, which let an observer group activity. The graph is append-only and universally replicated — every full node holds a complete copy — so analysis requires no special access, no subpoena, and no cooperation from anyone. Anyone with a database and patience can build it from public data.

Why the graph leaks privacy

Individual addresses are pseudonymous — just numbers — but the linkages between them are anything but random. Analysts apply heuristics to collapse the graph into wallets and identities. The common-input-ownership heuristic assumes all inputs to a transaction belong to one party, since they all had to be signed together. Change detection guesses which output returns leftover funds to the sender, based on round amounts, script types, and spending patterns. Address reuse is the gift that keeps giving: every payment to a reused address collapses into one obvious cluster. Once a single endpoint is identified — typically a KYC exchange deposit — those heuristics let an analyst walk the graph outward and reconstruct a user's broader financial life. And the graph never forgets: every edge recorded in a block is permanent, so a privacy mistake made today can retroactively expose transactions from years ago.

Reshaping what the graph shows

Privacy techniques work not by erasing the graph — that is impossible — but by making the conclusions an analyst draws from it unreliable. CoinJoin builds transactions with many equal-value outputs so that the mapping from inputs to outputs becomes ambiguous, enlarging the anonymity set. PayJoin breaks the common-input heuristic by having sender and receiver both contribute inputs to an ordinary-looking payment. Disciplined coin control and a fresh address for every receipt prevent unrelated payments from merging into one cluster in the first place. None of these hide the edges; they poison the heuristics that give the edges meaning.

Why miners should care too

Mining rewards enter the graph with a distinctive birthmark: coinbase outputs are publicly labeled as newly minted coins, and pool payout patterns are well understood by analysts. A home miner who consolidates months of payouts into one transaction, then spends into a KYC service, has drawn a clean line from hashrate to identity. Treating payout management with the same graph-awareness as any other coins — separating them, avoiding careless consolidation, using fresh addresses — keeps options open later. The act of tracing edges through this graph is detailed in our output linking entry, and the clustering it enables is covered under wallet clustering.