Account Intelligence for Crypto: From Wallets to Accounts
Key takeaways
Account intelligence for crypto groups wallets into the person, team, DAO, or company that controls them, then enriches, scores, and segments those accounts. The unit is the account, not the address.
62% of crypto users manage two or more wallets (Reown and Nansen, 2025), so wallet-level dashboards inflate active users, understate retention, and split attribution.
An account can include EOAs, embedded wallets, ERC-4337 smart accounts, EIP-7702 delegated EOAs, and Safe multisigs. Exchange and infrastructure addresses must never be merged into a user account.
Link wallets with evidence and record the confidence: shared user IDs and signed links are deterministic, shared sessions are strong, and onchain heuristics should only suggest links.
Score accounts on three separate axes (fit, engagement, and risk) and turn scores into a few segments, each with an owner and a next action.
For B2B crypto, treat a Safe as the account and its signers as the buying committee, and report sourced and influenced pipeline separately.
Account intelligence for crypto is the practice of grouping wallets into the people, teams, and organizations that control them, then enriching, scoring, and segmenting those accounts so growth, sales, and product teams know who to prioritize. The unit of analysis is the account, not the address.
That distinction matters more every year. In a 2025 YouGov survey for Reown and Nansen, 62% of crypto users said they manage two or more wallets. A user who connects a hardware wallet, an embedded wallet, and a Safe to your app looks like three users in a wallet-level dashboard. Your CAC is then too low, your retention looks worse than it is, and your best accounts are split into fragments that nobody prioritizes.
This guide explains what an account is in crypto, how B2B account intelligence maps to onchain data, how to resolve wallets into accounts with honest confidence levels, and how to score, segment, and act on the result.
What is account intelligence for crypto?
Account intelligence for crypto is the process of resolving wallets, sessions, and user IDs into accounts (a person, team, DAO, or company), enriching each account with onchain and offchain data, and scoring it for fit and engagement. It adapts B2B account intelligence, which relies on firmographics and intent data, to wallet-native signals.
In mainstream B2B software, the definition is well established. Demandbase describes account intelligence as connecting "static company information with third-party data and behavioral insights" to build account-based marketing campaigns. ZoomInfo describes software that aggregates "firmographic, technographic, contact, and intent data" to show which companies match an ideal customer profile and are actively researching solutions.
Every step in that model has an onchain equivalent. The difference is the starting point: B2B tools start from a company domain, while crypto apps start from a wallet address that says nothing about who is behind it.
Account intelligence vs wallet intelligence
Wallet intelligence describes a single address. Account intelligence describes the entity behind one or more addresses. You need wallet intelligence to build account intelligence, but the two answer different questions.
Dimension | Wallet intelligence | Account intelligence |
|---|---|---|
Unit of analysis | One address on one or more chains | A person, team, DAO, or company with many addresses |
Core question | What does this wallet hold and do? | Who is this, how valuable are they, and what should we do next? |
Typical data | Balances, DeFi positions, transaction history, labels | Linked wallets, combined value, app engagement, campaign touchpoints, identity, score |
Main risk | Treating one person's wallets as separate users | Merging wallets that do not belong together |
Primary users | Analysts, risk teams, researchers | Growth, BD, sales, lifecycle marketing, product |
For a deeper look at the address layer, see What Is Wallet Intelligence? and the wallet intelligence glossary entry.
What counts as an account in crypto?
The word "account" has two meanings in crypto. At the protocol level, an account is one address. At the go-to-market level, an account is whoever controls a set of addresses. Account intelligence works at the second level, so it has to understand every type of address that can belong to one entity.
Ethereum's documentation defines two protocol account types: externally owned accounts (EOAs), "controlled by anyone with the private keys," and contract accounts, "controlled by code." Account abstraction has blurred the line between them.
Address type | What it is | Why it matters for account intelligence |
|---|---|---|
EOA | A key-controlled wallet such as MetaMask or a hardware wallet | Still the most common way users connect. One person often holds several. |
Embedded wallet | A wallet created at signup through email or social login (Privy, Dynamic, Turnkey) | Often tied to an app user ID from day one, which makes deterministic linking easier. |
ERC-4337 smart account | A contract wallet that sends UserOperations through bundlers (EIP-4337) | The onchain sender can differ from the key that signs. Gas may be paid by a paymaster. |
EIP-7702 delegated EOA | An EOA that sets contract code on its own address (EIP-7702) | Same address, new behavior. History stays continuous, but transaction patterns change. |
Safe multisig | A smart account with N owners and an M-of-N approval threshold (Safe docs) | Usually a team, fund, or DAO treasury. The signers are the people you actually talk to. |
Exchange or custodian address | Deposit and hot wallets operated by a centralized service | One entity can control millions of addresses. These must never be merged into a user account. |
These are not edge cases. BundleBear's live dashboard counted more than 68 million ERC-4337 accounts that had sent at least one UserOperation (read October 2, 2026). The Safe Foundation's Q1 2026 report counted 61.11 million Safe accounts securing $35.25 billion in assets. Raw counts overstate real users: many EIP-7702 authorizations, for example, come from automated sweeper contracts rather than people. That is exactly why account-level resolution matters.
Why wallet-level analytics miscounts your users
Wallet-level dashboards assume one wallet equals one user. When that assumption fails, four metrics break at once.
Active users are inflated. A power user who rotates between three wallets is counted three times, while a real increase in new accounts is hidden in the noise.
Retention is understated. A user who moves from a browser wallet to a mobile embedded wallet looks like a churned user plus a new user.
Attribution is split. The campaign that acquired the person is credited on the first wallet, while deposits happen on the second. Neither wallet shows the full journey.
Value is fragmented. An account with $2 million across a Safe and two EOAs can appear as three mid-sized wallets, so it never reaches the top of a whale list.
Academic work shows how common multi-address control is. In Address Clustering Heuristics for Ethereum (Friedhelm Victor, Financial Cryptography 2020), onchain heuristics alone clustered 17.9% of all active EOA addresses into entities that likely control several addresses. App-level signals, such as one session connecting several wallets, find links that onchain heuristics cannot.
The five layers of crypto account intelligence
B2B account intelligence follows a familiar sequence: identify the account, enrich it, read intent, score it, then act. The same sequence works for crypto with wallet-native inputs.
Layer | B2B equivalent | Crypto inputs | Output |
|---|---|---|---|
1. Resolve | Reverse IP, domain matching | Login user IDs, multi-wallet sessions, Safe owners, signed links | An account with linked wallets and a confidence level |
2. Enrich | Firmographics, technographics | Net worth, DeFi positions, chains, wallet age, ENS, socials, attestations | An account profile |
3. Observe intent | Third-party intent data, web visits | Page views, feature use, wallet connects, signatures, transactions, contract events | Engagement and lifecycle stage |
4. Score | ICP fit and predictive scores | Fit, engagement, and risk signals combined into a transparent score | A ranked list of accounts |
5. Activate | ABM campaigns, sales alerts | Segments, alerts, CRM sync, in-app personalization, BD outreach | Actions and measured outcomes |
Teams that skip layer 1 build scores on fragments. Teams that skip layer 5 build dashboards nobody uses. The rest of this guide covers each layer in turn.
How do you link multiple wallets to one account?
Link wallets to an account using evidence, and record how strong the evidence is. Deterministic signals, such as one logged-in user ID across wallets, can be merged automatically. Behavioral signals, such as two wallets used in the same session, are strong but need rules. Onchain heuristics, such as shared funding sources, should suggest links, not merge them.
Chainalysis draws a useful line between clustering (grouping addresses under shared control) and attribution (naming the real-world entity), and notes that attribution is fundamentally an offchain activity. Keep the same separation in your data model.
Signal | Confidence | Recommended handling |
|---|---|---|
Same app user ID (login, email, embedded wallet) across wallets | Deterministic | Merge automatically |
User signs a message to link a second wallet | Deterministic | Merge automatically |
Address is a listed owner of a Safe the user operates | High | Link the signer to the Safe account, keep roles separate |
Several wallets connected in the same browser session | High, with exceptions | Link, but exclude shared devices and demo machines |
Repeated direct transfers between two wallets | Medium | Suggest a link for review |
Shared funding source or similar timing patterns | Low | Use for research only. Never merge. |
Exchange deposit addresses, routers, bridges, bundlers | Not a user link | Exclude from clustering |
The fastest way to raise confidence is to own the deterministic signals. Call an identify method with your internal user ID when a user logs in or connects a wallet, so every wallet they use joins the same profile. The wallet clustering and identity resolution glossary entries cover the terms in more detail.
What data belongs in a crypto account profile?
A useful account profile combines what the account holds, what it does in your app, how it arrived, and who it is, with each field labeled as observed, user-provided, or inferred.
Signal group | Examples | Source | Type |
|---|---|---|---|
Value | Combined net worth, per-chain split, stablecoin share | Onchain | Observed |
Ecosystem behavior | Apps used (Aave, Uniswap, Hyperliquid), open positions, wallet age, transaction count | Onchain | Observed |
In-app engagement | Sessions, features used, funnel stage, lifecycle stage, volume and revenue in your app | Your app and contracts | Observed |
Acquisition | First and last touch UTM, referrer, referral code, ad click ID | Your app | Observed |
Identity | ENS or Basename, X, GitHub, LinkedIn, website | Public name services and profiles | User-provided |
Trust | Coinbase Verified, Human Passport score, sybil list membership, sanctions status | Attestations and public lists | Observed |
Organization | Company, fund, or DAO name, role of each signer | CRM, BD research, self-reported forms | Inferred or user-provided |
A wallet balance does not prove company size, budget, or buying intent. Treat organization-level fields as research that needs a source and a date, and keep them separate from observed onchain facts.
How do you score and segment crypto accounts?
Score accounts on three independent axes: fit (does this account match who you want), engagement (is it active in your product now), and risk (is it a bot, sybil, or sanctioned address). Keep the axes separate so a whale with no engagement and an engaged small account are both visible for different reasons.
A transparent starting model for a DeFi lending app might look like this:
Axis | Signal | Example rule |
|---|---|---|
Fit | Combined net worth | Above $50,000 across linked wallets |
Fit | Relevant ecosystem activity | Open lending or borrowing position on another protocol |
Engagement | Lifecycle stage | New, Returning, or Power user (not Churned) |
Engagement | Product depth | Completed a deposit, not only a wallet connect |
Risk | Sybil and trust labels | Exclude listed sybils, require a minimum humanity score for rewards |
Turn the scores into a small number of segments that each have an owner and a next action:
High fit, high engagement: your best accounts. Give them priority support, early access, and a direct line to the team.
High fit, low engagement: connected but did not deposit. Investigate the drop-off and send a tailored onboarding path.
Low fit, high engagement: loyal smaller accounts. Good candidates for referral and ambassador programs.
Previously high engagement, now At Risk: trigger a retention play before they churn.
Validate the model against outcomes every quarter. If high-score accounts do not retain or generate more revenue than low-score accounts, the weights are wrong. See onchain user segmentation and how to identify at-risk DeFi users for segment examples.
Account intelligence for B2B crypto: Safes, DAOs, and funds
For teams that sell to other organizations, such as infrastructure providers, institutional DeFi products, and tooling companies, account intelligence looks closer to classic account-based marketing. According to the 6sense 2024 ABM Benchmark, 64% of B2B marketers say their teams run an account-based or target-account approach. Crypto adds a few twists.
The Safe is the account, the signers are the buying committee. A fund's treasury Safe holds the value, but its owners are the people who approve a new integration or deposit. Map signers to people and roles in your CRM.
A wallet is a signal, not proof of a company. Connect a wallet to an organization only when a signup, a verified business relationship, or an explicit product action supports the link.
Separate sourced and influenced pipeline. Sourced pipeline credits the campaign that created an opportunity under a stated rule. Influenced pipeline records campaigns that engaged an existing opportunity. Report them separately.
Keep records distinct. CRM companies and opportunities, app users, and wallet addresses should be separate objects with explicit relationships, so one bad link does not corrupt the others.
Example: building account intelligence for a lending protocol
A lending protocol wants to find and retain its most valuable depositors. Here is a practical workflow.
Capture identity at the source. Install an analytics SDK that records page views, wallet connects, and transactions, and call identify with the internal user ID after login.
Ingest contract events. Track Deposit, Withdraw, and Borrow events from the protocol's contracts so value is measured from the chain, not the frontend.
Resolve accounts. Merge wallets that share a user ID or a session, link Safe owners to their Safes, and exclude routers, bundlers, and exchange addresses.
Enrich. Add combined net worth, positions on other lending markets, wallet age, ENS and socials, and trust labels.
Score and segment. Apply the fit, engagement, and risk model above and save the segments.
Activate. Send At Risk high-value accounts to the retention team through a webhook alert, export high-fit non-depositors for a targeted campaign, and give BD a weekly list of new Safe accounts above a value threshold.
Measure. Compare deposit retention and revenue per account for each segment before and after each play.
Privacy and data handling
Onchain data is public, but linking wallets to each other and to offchain identifiers creates new information about real people. The European Data Protection Board's Guidelines 02/2025 on blockchain (version 2.0, adopted July 2026) state that wallet addresses and public keys are personal data when a person can be identified by means reasonably likely to be used. Apply three rules:
Collect with a purpose. Only link identities when it serves a stated product, security, or support purpose, and disclose it in your privacy policy.
Label the source of each field. Mark every signal as observed, user-provided, or inferred, and store when it was collected.
Do not export more than you need. Ad platforms and partners rarely need wallet addresses or portfolio details. Send the minimum fields required.
How Formo supports account intelligence
Formo is an analytics and wallet intelligence platform for onchain apps. It covers the five layers in one data model:
Resolve: Formo groups wallets that share a session or a common user ID into clusters. Each wallet profile lists its linked addresses, and the Clusters view shows every multi-wallet group. The SDK's identify call links wallets to your own user ID.
Enrich: profiles include net worth across EVM chains and Solana, DeFi positions by app, token holdings, wallet age, ENS and social profiles, and labels such as Coinbase Verified, Human Passport scores, and sybil lists.
Observe intent: the web, mobile, and server SDKs capture page views, wallet connects, signatures, and transactions, and contract event ingestion adds decoded onchain actions with first-touch and last-touch attribution.
Score and segment: automatic lifecycle stages (New, Returning, Power user, At Risk, Churned, Resurrected) combine with value, label, and behavior filters in saved segments.
Activate: export segments, query profiles through the Profiles API, trigger webhook alerts, or ask questions in plain language with Formo MCP.
To see how profiles are built, read How to Build Wallet Profiles. To connect accounts to campaigns, see Onchain Attribution: The Complete Guide for DeFi Teams.
Frequently asked questions
What is account intelligence in crypto?
Account intelligence in crypto is the process of grouping wallets into the person, team, or organization that controls them, then enriching, scoring, and segmenting those accounts. It adapts B2B account intelligence to onchain data, so teams can prioritize real accounts instead of individual addresses.
What is the difference between account intelligence and wallet intelligence?
Wallet intelligence describes one address: its holdings, history, and labels. Account intelligence describes the entity behind one or more addresses and adds resolution, combined value, in-app engagement, attribution, and scoring. Wallet intelligence is an input to account intelligence.
Can you link multiple wallets to one person?
Yes, with evidence. A shared login user ID or a signed linking message is deterministic. Wallets connected in the same session are a strong signal. Onchain heuristics such as shared funding sources are weaker and should only suggest a link for review.
Is a Safe multisig one account or many?
Treat a Safe as one account with several related people. The Safe holds the value and executes transactions, while its owners approve them. For B2B work, the owners act like a buying committee, so map each signer to a person and a role.
Does EIP-7702 change how wallets are tracked?
EIP-7702 lets an EOA set contract code on its own address, so the address and its history stay the same. Analytics that rely on addresses keep working, but transaction patterns can change, for example when a sponsor pays gas or several actions are batched.
How do you exclude bots and sybils from account scores?
Use a separate risk axis. Exclude addresses on known sybil lists, require a minimum humanity or verification score for rewards, and remove infrastructure addresses such as routers, bundlers, and exchange deposit wallets from clustering.
Can you do account-based marketing in web3 without emails or cookies?
Partly. Wallets, onchain behavior, and public profiles let you identify and prioritize accounts without third-party cookies. You still need an offchain channel, such as a CRM contact, a community channel, or an in-app message, to reach the people behind the account.
