Combination Similarity Keys: Higher-Precision Matching with Two-Field AI-Powered Keys
Single-field matching has been the standard approach in data quality for years. You match on company name, or on address, or on a person's name, and hope for the best. In many cases it's fine. But anyone who has worked with real-world data knows the challenge: matching on one field alone is difficult with common data. "James Kelly" could appear dozens of times in a CRM. "IBM" shows up at hundreds of addresses. "100 Main Street" could be anywhere.
Interzoid's Combination Similarity Key APIs solve this by generating a single AI-powered hashed similarity key from two data fields simultaneously. The resulting key only matches when both fields are similar, giving you dramatically higher precision for deduplication, record linkage, and fuzzy matching across messy, inconsistent datasets.
Company + Address | Company + Full Name | Address + Full Name
Register for an API key to start using them immediately, or try them interactively with no code required.
The Challenge with Single-Field Matching
Consider a common scenario: you're deduplicating a vendor database and you match on company name alone. "IBM" at your New York office and "International Business Machines" at your Armonk headquarters both generate the same similarity key, even though they are distinct business locations that you need to keep separate. Match on address alone, and "100 Main Street" links records for completely unrelated companies that happen to share the same street address.
The core issue is that individual data fields in many cases do not carry enough information to uniquely identify a record by themselves. Names are common. Addresses are shared. Company names are abbreviated, misspelled, and formatted inconsistently. You need the combination of multiple fields to achieve the precision that real-world data quality demands.
How Combination Similarity Keys Work
Each of the three Combination Similarity Key APIs takes two input fields and returns a single hashed similarity key. Under the hood, each field is processed by its own specialized AI model—company names are standardized using Interzoid's model-v4-wide algorithm (handling abbreviations, acronyms, and international names), addresses use the model-v3-wide algorithm (normalizing street types, directionals, and unit numbers), and individual names use AI-powered phonetic matching with nickname resolution and international name handling.
The two processed fields are then combined into a single composite key. Records generate the same combined key only when both fields are similar. If either field differs significantly, the key will be different—even if the other field is a perfect match. This is what eliminates the false positives that plague single-field approaches.
1. Company + Address Similarity Key API
Deduplicate Business Locations
The Company and Address Match API generates a combined similarity key from a company/organization name and a street address. This is ideal for deduplicating vendor lists, supplier databases, and account records where the same business location may appear with inconsistent company name or address formatting.
Endpoint:
| Company | Address | Match? |
|---|---|---|
| IBM | 1 New Orchard Road | ✓ Same key |
| Intl Business Machines | 1 New Orchard Rd | ✓ Same key |
| International Business Machines Corp. | 1 New Orchard Rd. | ✓ Same key |
| IBM | 500 Oak Drive | ✗ Different key |
Notice that "IBM," "Intl Business Machines," and "International Business Machines Corp." all match at the same address—but "IBM" at a different address correctly produces a different key. The API handles abbreviations like "Rd" vs. "Road," acronyms like "IBM" vs. the full name, misspellings like "microsot," and international company names like "PetroVietnam" vs. "Vietnam Oil and Gas Group."
2. Company + Full Name Similarity Key API
Deduplicate Contacts Within Organizations
The Company and Full Name Match API generates a combined similarity key from a company/organization name and an individual's full name. This is the go-to API for deduplicating contacts across CRM systems, marketing databases, and sales pipelines where the same person at the same company appears with inconsistent formatting.
Endpoint:
| Company | Full Name | Match? |
|---|---|---|
| IBM | James Kelly | ✓ Same key |
| Intl Business Machines | Jim Kelley | ✓ Same key |
| International Business Machines Corp. | Mr. Jim H. Kellie | ✓ Same key |
| IBM | Sally Anderson | ✗ Different key |
The API correctly identifies that "James Kelly" at "IBM" is the same contact as "Jim Kelley" at "Intl Business Machines"—resolving nicknames, phonetic spelling variations, and company name abbreviations simultaneously. But "Sally Anderson" at "IBM" correctly gets a different key because the person is different.
3. Address + Full Name Similarity Key API
Deduplicate People at Locations
The Address and Full Name Match API generates a combined similarity key from a street address and an individual's full name. This is especially valuable for identity verification, household deduplication, and record linkage scenarios where you need to confirm that the same person appears at the same location across different data sources.
Endpoint:
| Full Name | Address | Match? |
|---|---|---|
| James Kelly | 100 Main St | ✓ Same key |
| Jim Kelley | 100 Main Street | ✓ Same key |
| Mr. Jim H. Kellie | 100 Main St. | ✓ Same key |
| James Kelly | 500 Oak Drive | ✗ Different key |
"James Kelly" at "100 Main St" is recognized as the same person as "Jim Kelley" at "100 Main Street" and "Mr. Jim H. Kellie" at "100 Main St."—but "James Kelly" at a different address correctly produces a different key. The API handles international names, honorifics, suffixes, and address formatting variations seamlessly.
Use Cases for Combination Similarity Keys
Two-field similarity keys open up matching scenarios that are simply not practical with single-field APIs:
- CRM deduplication – The Company + Full Name API catches duplicate contacts where both the company name and person's name have formatting inconsistencies, spelling variations, or abbreviations. This is the most common deduplication challenge in sales and marketing databases.
- Vendor and supplier consolidation – The Company + Address API identifies duplicate business locations across vendor lists and AP systems, even when company names are abbreviated differently or addresses use varying formats.
- Identity verification and KYC – The Address + Full Name API confirms whether the same individual appears at the same location across different data sources—critical for compliance, fraud detection, and customer onboarding.
- Household deduplication – Direct mail and marketing teams can use Address + Full Name matching to consolidate records for the same person at the same address, reducing duplicate mailings and improving campaign efficiency.
- Cross-system record linkage – When merging datasets from different systems (ERP, CRM, billing, etc.), combination keys enable high-confidence matching without requiring exact field agreement.
Integration and Usage Options
All three Combination Similarity Key APIs are standard REST endpoints that return clean JSON responses. They can be integrated using any approach that works with Interzoid's platform:
- Direct API calls from Python, Node.js, Go, Java, TypeScript, and more. Code examples are available on GitHub.
- Interactive testing via the Try APIs app—test any combination API with a UI, no code required.
- Batch processing via the Batch API tool—run against entire CSV/TSV files in Interzoid's cloud platform for high-volume matching.
- Database queries—use combination keys directly within SQL queries on Snowflake, Postgres, MySQL, and Azure SQL for in-database deduplication.
- MCP Server—use Interzoid's MCP Server for integration with AI assistants and agent workflows.
Combination Keys vs. Single-Field Keys
Interzoid's combination APIs complement the existing single-field matching APIs—they don't replace them. The right choice depends on your use case:
- Use single-field APIs (Company Name, Individual Name, Street Address) when you want broad matching on a single dimension, or when your data only has one reliable field to match on.
- Use combination APIs when you need higher precision and can provide two fields. The combined key approach significantly reduces false positives by requiring similarity on both dimensions simultaneously.
- Use both together for tiered matching strategies—start with combination keys for high-confidence matches, then fall back to single-field keys to catch additional potential matches at a broader threshold.
Real-world data is messy, and matching on a single field rarely provides the precision that serious data quality work demands. Interzoid's Combination Similarity Key APIs bring two-field matching into a single API call—combining AI-powered standardization of company names, addresses, and individual names to deliver dramatically fewer false positives and higher-confidence record linkage.
Whether you're deduplicating contacts in a CRM, consolidating vendor records, verifying identities across data sources, or linking records across systems, these three APIs give you the precision you need through simple REST calls with structured JSON responses.