Entity resolution is the process of stitching together multiple records relating to the same individual or business in order to get a more concise view of risk on a given person or company. For entity resolution to work, unique entity records are standardized and displayed together in a single cluster instead of multiple, independent records.
Without proper consolidation, risk assessments become cumbersome, with analysts executing time-intensive processes to review red flags for onboarding or investigations. In McKinsey research, banks said that as much as 85 percent of financial crime compliance and AML activities remain administrative or non-analytical in nature. This includes the manual collection of data from some systems to import into others.
The problem has grown since. Screening alert volumes keep rising, with 83% of mid- and large-sized institutions and 87% of smaller ones in the US and Canada reporting more alerts. And as institutions adopt AI, data quality has become the top challenge cited by 48% of banks in the Wolters Kluwer Q1 2026 Banking Compliance AI Trend Report. Entity resolution sits at the center of both problems.
Key takeaways
- Entity resolution links records that refer to the same person or company across sources, so analysts review one consolidated profile instead of many fragments.
- The main benefits for compliance teams are fewer duplicate records and false positives, faster reviews, more accurate risk scores, better detection of hidden ownership and network risk, and a cleaner audit trail.
- Strong entity resolution combines several identifiers, such as registration numbers, names, addresses, and dates, using deterministic, probabilistic, and machine learning matching.
- Resolution should be tuned to your risk appetite: stricter linking produces fewer, more certain matches, while broader linking catches more potential connections.
- Entity resolution is the foundation for AI in compliance. Agents and models are only as reliable as the entity data they work from.
- Recent sanctions evasion cases, including companies reincorporated under new names with new nominal owners, show why resolving entities across names and records matters.
What is entity resolution?
Entity resolution determines whether two or more records, from the same source or from different ones, describe the same real-world person or organization, and links them into a single profile. It also does the opposite: it keeps apart records that look similar but refer to different people or companies.
In financial crime compliance, the records being resolved come from sanctions lists, PEP databases, adverse media, corporate registries, ownership data, and an institution’s own customer files. A single company might appear as “ABC Trading Ltd”, “A.B.C. Trading Limited”, and a transliterated name in a foreign registry, each with a different address format. Entity resolution recognizes these as one entity.
Entity resolution vs related terms
| Term | What it does | How it relates |
|---|---|---|
| Deduplication | Removes duplicate records within one dataset | A narrow part of entity resolution |
| Record linkage | Links records across datasets | The matching step within entity resolution |
| Identity resolution | Links identifiers to a known individual, often for marketing or fraud | A type of entity resolution focused on people |
| Name screening | Compares customer names against watchlists | Relies on entity resolution to separate true matches from false positives |
| Network analysis | Maps relationships between resolved entities | Builds on entity resolution to find indirect risk |
How entity resolution works
Most entity resolution systems follow the same stages:
- Ingest: collect records from internal systems and external risk data sources.
- Standardize: clean and normalize names, addresses, dates, and identifiers, including transliteration across scripts.
- Block: group records that are plausible candidates for a match, so the system compares likely pairs instead of every record against every other.
- Match: score each candidate pair using several identifiers.
- Cluster: group matched records into a single resolved entity.
- Review and learn: route uncertain matches to analysts and feed their decisions back to improve accuracy.
Matching approaches
| Approach | How it works | Strengths | Limitations |
|---|---|---|---|
| Deterministic (rules-based) | Links records that match exactly on defined fields, such as a registration number | Transparent and precise | Misses variations, typos, and missing data |
| Probabilistic (fuzzy) | Weighs similarity across several fields to calculate a match probability | Handles spelling variations and incomplete records | Needs careful threshold tuning |
| Machine learning | Learns matching patterns from labeled examples and analyst decisions | Adapts to complex data at scale | Needs governance, explainability, and monitoring |
In practice, strong systems combine all three, using exact identifiers where they exist and probabilistic and machine learning methods where they do not, while keeping each match decision explainable.
How entity resolution increases team output
Whatever the risk appetite, entity resolution uses multiple data points to group records together. This results in two advantages:
- Analysts see fewer results when performing searches, and in many cases just one for the entity they searched, thanks to the AI-driven resolution of key identifiers.
- Risk events related to an entity are also resolved, so the same information from multiple sources is grouped together, which simplifies reviews.
How entity data is compacted for consumption is extremely important. An organization’s risk appetite will inform fine tuning, whether the goal is to return focused or broad results. Organizations with higher risk appetites may resolve conservatively, linking only exact matches as part of due diligence processes. On the opposite end of the spectrum, highly scrutinized organizations often lean towards broader linking, including fuzzy matches when resolving entities.
| Setting | What it does | Best suited to | Trade-off |
|---|---|---|---|
| Conservative (exact matches) | Links records only on strong, exact identifiers | Lower-risk portfolios and high-volume onboarding | Fewer alerts, but some true links may be missed |
| Balanced | Combines exact identifiers with high-confidence fuzzy matches | Most banks and fintechs | Moderate review load with good coverage |
| Broad (fuzzy and network) | Links on partial matches and relationships | High-risk segments, EDD, and investigations | Maximum coverage, more records to review |
Benefits of entity resolution
Here are the ways entity resolution makes compliance work more efficient and more effective:
- Reduce duplicate records: duplicate records for the same entity can proliferate within databases and systems, especially those that pull information from multiple external sources. This can lead to confusion about which record is the accurate one.
- Reduce false positives: by confirming whether a screening hit really refers to your customer, entity resolution clears name-only matches before they reach an analyst. Our AI Investigator Agent uses AI and entity resolution models to clear easily identifiable false positives, and we report up to a 90% reduction in manual match review.
- Speed up reviews and investigations: analysts work from one consolidated profile with all related risk events attached, instead of reconciling records across systems.
- Avoid inaccurate analytics: without entity resolution, data analytics and reporting may be compromised. Analysis based on incomplete or inaccurate entity data can lead to flawed insights and poor decision-making.
- Improve risk scoring: a customer risk score is only as accurate as the data behind it. Resolved entities give a complete picture for customer risk assessment.
- Reveal hidden ownership and network risk: once entities are resolved, relationships between them become visible, including shared directors, shared addresses, and indirect links to sanctioned parties.
- Manage regulatory compliance risks: failing to resolve entities accurately can lead to non-compliance and regulatory fines. Accurate customer identification is essential for know your customer (KYC) and anti-money laundering (AML) regulations.
- Stronger fraud detection and prevention: incomplete or inaccurate entity data can hinder fraud detection efforts. Fraudsters can exploit gaps in entity identification to engage in identity theft or financial fraud.
- Prepare data for AI: AI agents and models inherit the quality of the data they use. Resolved, standardized entity data makes AI-assisted review more accurate and easier to explain.
- Strengthen the audit trail: a single resolved profile, with the sources behind each link, makes it easier to show examiners how a decision was reached.
- Address mergers and acquisitions challenges: when organizations undergo mergers or acquisitions, the lack of entity resolution can complicate the process of integrating customer, supplier, and partner data.
Automating data consolidation processes with technology frees risk teams to focus on analysis and escalation tasks that actually require human intervention.
In a proof of concept with a global payments firm, our AI Investigator Agent, built on entity resolution and AI, cut 93.3% of false positives, with projected savings of up to $1 million a year. Read the full global payments case study, and see our guide on how to reduce false positives in AML screening.
Identifiers used in entity resolution
There are many ways to resolve entity data. Each technology you consider may have a different mix of identifiers contributing to the resolution algorithm. A strong system will often combine multiple identifiers to ensure accuracy.
Some identifiers that can be used for entity resolution include:
- Government identifiers: for individuals, this might include Social Security numbers or government-issued ID numbers. For companies, it could be registration numbers, tax identification numbers, or legal entity identifiers (LEIs). Exact matching on these identifiers is a strong basis for linking and resolving entities.
- Name variations: names of individuals and organizations can vary significantly due to misspellings, abbreviations, transliteration, or variations in formatting. Algorithms for fuzzy or phonetic matching are used to link records with similar but not identical names.
- Address information: addresses, including street addresses, postal codes, and city names, can be used to link records, especially in cases where names may vary. Geospatial data may also be used for this purpose.
- Contact information: email addresses, phone numbers, and other contact information can establish connections between multiple entity records.
- Date of birth or incorporation: for individuals and companies, dates of birth and incorporation can be used to establish links, especially when other information is incomplete or inconsistent.
- Social network and relationship data: information about relationships between entities, such as family connections, directorships, or shareholdings, can be used to create links between related records.
- Transaction data: data related to transactions, such as dates, amounts, and transaction types, can be used to identify connections between entities involved in financial activities.
Entity resolution use cases in compliance
| Use case | How entity resolution helps |
|---|---|
| Sanctions and watchlist screening | Confirms or clears matches and links entities owned by sanctioned persons |
| Adverse media screening | Maps each article to the right entity and groups repeat coverage of the same event |
| KYC and KYB onboarding | Builds one verified profile from customer data, registries, and ownership records |
| Enhanced due diligence | Connects owners, directors, and associates to reveal ultimate beneficial ownership |
| Perpetual KYC | Attaches new risk events to the right existing customer as they occur |
| Third-party risk management | Unifies supplier and partner records across business units |
| Investigations | Links scattered records into one case view with sources |
The challenges of unresolved customer risk data
Below are four main issues commonly encountered by compliance teams that don’t use entity resolution to streamline operations.
1. Fragmented data streams
Multiple records for the same entity from various sources, each with their own format and structure, can make it practically impossible for banks and mid-market corporations to accurately assess customer-associated risks at scale. We’ve seen organizations grow compliance and risk teams from 5 to 50 and more, and still face the same issues with data management.
2. Increased manual efforts
Consolidating multiple risk sources can mean a lot of noise in search results. The time saved from searching disparate sources is often reallocated to combing through long lists of results.
Take for example a single search relating to a current customer: you’ll see unique records for sanctions, PEPs, adverse news, and more from multiple databases. Your team may be looking through different records pointing to the exact same entity, or even the exact same risk, which can be cumbersome in reviews.
3. Gaps in internal data
Challenging scenarios can arise when internal data is incomplete or “dirty.” For example, let’s say a client returns multiple Social Security numbers, what do you do? Imagine these types of edge cases piling up, at scale.
Without resolving external risk data to provide deeper insights, gaps in internal data add to the manual workload of risk and compliance teams. Even with automated processes, the volume of alerts generated can lead to backlogs, while a lack of complete data coverage can result in gaps in surveillance.
4. Hidden networks and ownership
The most serious risk often sits in records that look unrelated. Without entity resolution, a customer’s link to a sanctioned party through a shared director, a common address, or a renamed company stays invisible, because no single record shows it.
Why entity resolution matters now: examples from 2026
Recent enforcement shows how sanctions evasion and illicit finance exploit unresolved data:
- Renamed companies and new nominal owners. OFAC’s March 2026 advisory on sham transactions describes a company sanctioned for narcotics trafficking that was reincorporated under a different name with new nominal owners while continuing the same operations, and oligarchs who moved assets through layers of shell companies and trusts. Linking the old and new entities depends on resolving people, addresses, and relationships, not just names.
- Look-alike front companies. FinCEN’s May 2026 alert on Iranian front companies found that likely shell companies moved $5 billion in 2024, and identified them through indicators that included shared addresses, shared counterparties, and name similarities with OFAC-designated companies.
- Cartel networks behind ordinary businesses. Our analysis of Treasury’s Sinaloa Cartel action shows how cartel finance runs through companies with no visible link to any listed person until ownership and relationships are mapped.
Research supports the approach. A peer-reviewed study in the Journal of Economic Criminology found that combining graph analytics with machine learning on public data detected shell companies used for money laundering with accuracy between 88% and 98%.
How to evaluate entity resolution in compliance software
Most screening and KYC vendors claim entity resolution. These questions help separate strong implementations from basic name matching:
- Which identifiers does the matching use, and how are they weighted?
- How are names handled across scripts, transliterations, and aliases?
- Can we tune match thresholds to our risk appetite without engineering support?
- Can analysts see why two records were linked or kept apart?
- Are risk events, such as news articles and list entries, resolved to entities too, not just names?
- Does resolution extend to relationships, so we can see owners, directors, and associates?
- How are analyst decisions fed back to improve matching over time?
- How is the matching model validated, monitored, and governed?
The last question matters more as AI takes on more review work. US banking agencies replaced their model risk guidance with SR 26-2 in April 2026, and examiners expect machine learning used in compliance to be explainable and well governed. Our guide to AI in AML compliance covers what that looks like in practice.
Final thoughts on entity resolution
In a landscape where even the largest enterprises with extensive compliance resources are often penalized for violating regulatory requirements, no organization can afford to look the other way when it comes to taking a proactive and complete approach to risk management.
Regulatory bodies and other stakeholders expect organizations to have the processes needed to identify financial crime quickly amid frequently expanded and updated regulations, and that requires additional efficiencies.
Ensuring strong entity resolution in your current and future risk software is one way to stay ahead of regulatory requirements without overwhelming your team.
Support your organization’s compliance transformation
We give compliance teams instant access to a vetted, ready-to-use risk data library covering KYC, sanctions, PEPs, adverse media, and corporate data. Our risk intelligence platform draws on more than 100 billion data points and 150+ corporate registries, and can ingest and optimize any structured and unstructured risk-related data, including your internal customer data, so that you can access a unified view of risk on any person, company, or entity in a single dashboard.
Entity resolution runs through every part of the platform:
- Screening: resolves sanctions, PEP, and watchlist hits to the right entity, so obvious false positives are cleared before they reach an analyst.
- Adverse media: maps every article to named entities and consolidates related coverage into one story with our Adverse Media Summary.
- Network intelligence: derived intelligence such as shared addresses, shared directors, and nominee relationships links resolved entities to reveal ultimate beneficial ownership and indirect risk.
- AI-assisted review: our AI360 features, including the AI Investigator Agent and Entity Summary, build on resolved entity data to automate first-level review and compile standardized risk profiles in seconds.
- Configurable controls: compliance teams can tune thresholds and filters to their risk appetite without engineering support, and every decision is logged with its sources.
Discover why global companies choose Sigma360 to simplify risk management for onboarding, periodic reviews, and ongoing monitoring workflows. Request a demo.
Frequently asked questions
What is entity resolution?
Entity resolution is the process of identifying records that refer to the same real-world person or organization, across one or many data sources, and linking them into a single profile. It also keeps apart records that look similar but describe different entities.
What are the benefits of entity resolution?
The main benefits are fewer duplicate records and false positives, faster reviews and investigations, more accurate analytics and risk scoring, visibility into hidden ownership and network risk, stronger fraud detection, better data for AI, and a clearer audit trail for regulators.
How does entity resolution reduce false positives in AML screening?
It checks whether a watchlist or adverse media hit actually refers to your customer by comparing several identifiers, such as dates of birth, registration numbers, addresses, and nationality, not just names. Hits that clearly refer to a different person or company can be cleared automatically, so analysts review only plausible matches.
What is the difference between entity resolution and deduplication?
Deduplication removes duplicate records within a single dataset. Entity resolution is broader: it links records across many sources, handles name variations and missing data, and resolves related risk events, so the result is one complete profile for each person or company.
Which identifiers are used in entity resolution?
Common identifiers include government and company registration numbers, legal entity identifiers, names and their variations, addresses, contact details, dates of birth or incorporation, relationship data such as directorships and family links, and transaction data.
How should entity resolution be tuned to risk appetite?
Institutions with a higher risk appetite can resolve conservatively, linking only on strong, exact identifiers to keep alert volumes low. Highly scrutinized institutions, or high-risk segments, often use broader linking that includes fuzzy and relationship-based matches to maximize coverage. The right setting balances review workload against the risk of missing a true link.
Why is entity resolution important for AI in compliance?
AI agents and models make decisions based on the entity data they are given. If records are fragmented or duplicated, AI outputs will be incomplete or inconsistent. Resolved, standardized entity data makes AI-assisted review more accurate, and makes each decision easier to explain to examiners.
