On-Chain Attribution Can Find the Wallet. Off-Chain Data Is What Makes It Count.
A blockchain can tell you where the money went. On its own, it cannot tell you everything you need to know about the person, entity or activity behind it.
Ruta Kazakiene
CCAS, Compliance Officer
5 min read
A blockchain can tell you where the money went. On its own, it cannot tell you everything you need to know about the person, entity or activity behind it.
Blockchain analysis played a central role in tracing two of crypto's largest DeFi exploits linked to North Korea's Lazarus Group: the roughly $625 million theft from Axie Infinity's Ronin Network in 2022, and the roughly $100 million theft from Harmony's Horizon Bridge two months later, both later attributed to the group by US authorities. Blockchain analytics allowed investigators to trace stolen funds, identify relationships between wallets and follow their movement across the ecosystem.
That illustrates something an identity check alone structurally cannot do: show what happens to funds after onboarding.
But the limitation cuts both ways. A compliance programme that treats a clean KYC file as the finish line is relying on a snapshot taken at account opening to assess a relationship that continues to evolve. On-chain monitoring has the opposite gap: it can identify transaction patterns, wallet relationships and risk exposure, but those signals do not provide the full customer context needed to understand why activity is taking place or whether it is consistent with what is known about the customer.
Neither side finishes the job the other one started.

A fan-out through mule accounts. On-chain, each transfer looks unrelated; off-chain data (shared device, IP, or case history) is what reveals they belong to the same operator.
An On-Chain Signal Has Confidence. Off-Chain Data Adds Context.
Blockchain analytics can identify relationships between addresses, trace transaction flows and assess exposure to known services or risk indicators. But attribution itself is not always absolute, and blockchain data does not contain the full customer context behind an account.
That context comes from information collected and developed throughout the customer relationship: identity and KYC information, expected activity, occupation or business, source of funds and wealth, account behaviour, previous reviews and, where necessary, explanations or supporting evidence provided by the customer.
When blockchain intelligence raises a risk signal that appears inconsistent with what is known about a customer, the harder question is not simply which signal is right. It is how the programme brings those different pieces of information together, determines whether further investigation is needed and decides what evidence is sufficient to resolve the concern.
New On-Chain Intelligence Can Reach Backward. Off-Chain Records Rarely Do.
Crypto transaction monitoring is not necessarily finished at the moment funds move. When new intelligence links a wallet or service to sanctions exposure, a darknet market, ransomware or other illicit activity, that intelligence can be applied retrospectively to historical transactions — potentially surfacing exposure that was not known when the transaction originally occurred.
The customer information behind those transactions may tell a different story. Unless it has been refreshed, a profile may still reflect information collected months or years earlier.
When the on-chain side revises its understanding of the past, how far back should the off-chain side revisit its understanding of the customer?
Some Transactions Have Limited Off-Chain Context to Check Against.
FATF continues to identify peer-to-peer transactions through unhosted wallets as an area presenting particular money-laundering and terrorist-financing challenges. The reason is partly structural: unlike a transfer between two regulated service providers, there may be no obliged entity on the other side holding verified customer information.
The blockchain may provide visibility into the transaction, while corresponding off-chain identity information about the counterparty may simply not exist.
That does not automatically make the transaction suspicious. It does mean that compliance teams may sometimes have to assess blockchain risk signals with less counterparty context than they would have in a fully intermediated transaction. How that absence of information should influence the overall risk assessment remains a fundamentally risk-based question.
Network Behaviour Is Also a Form of Off-Chain Data.
The same idea holds well outside crypto. A joint study by the Bank of England and the BIS Innovation Hub, Project Hertha, tested whether network-wide analysis of a retail payment system could help banks and payment service providers spot financial crime they would otherwise miss. Used alone, the network-wide view actually found fewer illicit accounts than banks and PSPs monitoring their own customers directly (39% versus 44%). Combining the two lifted detection by around 12% overall, and by around 26% for the coordinated, multi-institution schemes that no single participant could see in isolation, in some cases roughly doubling detection accuracy. The models also performed markedly better once trained on confirmed past cases than without them. None of that is on-chain data in the crypto sense, but it plays the same role: relational and behavioural information that only becomes visible once it is assembled from outside the transaction itself, and that a purely transactional signal cannot supply on its own.

Institution-level view versus network-wide view of the same accounts. A coordinated ring across two banks is invisible to either bank alone, but visible once the data is assembled network-wide — the same principle that the Bank for International Settlement's Project Hertha found.
Looking Ahead
On-chain attribution keeps getting faster and more sophisticated, and cases involving actors such as the Lazarus Group demonstrate how powerful blockchain analysis can be in tracing illicit funds. But greater visibility on-chain does not remove the need for customer and behavioural context off-chain, and Project Hertha suggests this is not a crypto-specific problem so much as a structural one.
The challenge for crypto transaction monitoring is therefore not simply collecting more blockchain data or generating more alerts. It is connecting what is happening on-chain with what is known about the customer, their expected activity and their behaviour — and knowing when new information on either side should change the assessment made on the other.
The chain will keep talking. Whether a compliance programme can keep listening on both channels at once is the actual test.




