Entity Resolution

Learn how entity resolution matches and links business records across data sources to create unified, verified business identities.

3 min read

Entity resolution is the process of matching, linking, and deduplicating records that refer to the same real-world business across different data sources. It’s the technical foundation that transforms fragmented business data into verified identities.

The Problem Entity Resolution Solves

Business information is scattered across thousands of sources—state registries, credit bureaus, web data, transaction records, and more. The same business appears differently in each:

State filing: “GTL Services LLC”
Website: “Green Thumb Landscaping”
Credit card processor: “GREEN THUMB LANDSCAPE”
Google listing: “Green Thumb Landscaping & Lawn Care”

Without entity resolution, these look like four different businesses. With it, they’re recognized as one.

Why Entity Resolution Is Hard

Name Variation

Businesses operate under multiple names:

Legal names vs. trade names (DBAs)
Abbreviations and acronyms
Spelling variations and typos
Name changes over time

Address Complexity

Registered addresses differ from operating locations
Suite numbers, formatting inconsistencies
Businesses relocate
Multiple locations under one entity

Ownership Structures

Parent companies and subsidiaries
Franchises with shared branding but separate entities
Complex corporate hierarchies

Data Quality

Outdated records
Incomplete information
Conflicting data across sources
Intentional obfuscation

Entity Resolution Techniques

Deterministic Matching

Exact matches on unique identifiers:

EIN (Employer Identification Number)
State registration numbers
DUNS numbers

Pros: High precision, fast Cons: Requires exact match; many records lack identifiers

Probabilistic Matching

Statistical comparison of multiple attributes:

Name similarity algorithms (edit distance, phonetic matching)
Address standardization and comparison
Weighted scoring across attributes

Pros: Handles variation and partial matches Cons: Requires tuning; can produce false positives

Graph-Based Resolution

Connecting records through relationships:

Shared addresses link entities
Common officers or owners
Business relationships and transactions

Pros: Captures complex structures Cons: Computationally intensive; requires relationship data

Entity Resolution in KYB

Entity resolution is essential for effective KYB:

Verification accuracy: Correctly matching a business application to its official registration—even when names differ—enables accurate verification.

Ownership tracing: Linking entities through ownership chains to identify ultimate beneficial owners.

Risk detection: Recognizing when multiple applications share suspicious patterns (same registered agent, same formation date, same address).

Deduplication: Ensuring the same business isn’t onboarded multiple times under different names.

The Quality Spectrum

Entity resolution exists on a spectrum from basic to comprehensive:

Level	Approach	Result
None	Exact name match only	Misses most legitimate matches
Basic	Simple fuzzy matching	High false positive rate
Intermediate	Multi-attribute probabilistic	Reasonable accuracy
Advanced	Graph-based with multiple sources	High accuracy, reveals structure

The right level depends on risk tolerance and use case. High-stakes decisions (lending, compliance) demand advanced resolution.

Key Takeaways

Entity resolution connects fragmented records into unified business identities
Name variation is the core challenge—businesses appear differently across sources
Multiple techniques exist—deterministic, probabilistic, and graph-based
Resolution quality directly impacts KYB accuracy—poor resolution means missed matches or false positives
Advanced resolution reveals structure—ownership chains, related entities, and risk patterns