# AWS Entity Resolution

> Managed AWS service that matches records stored in Amazon S3 with rule-based, ML or data provider workflows, assigns a shared match ID to related records, and charges per 1,000 records processed.

Source: https://toolsfordata.com/tools/aws-entity-resolution/ · Updated 2026-11-20

[Website](https://aws.amazon.com/entity-resolution/) · [Docs](https://docs.aws.amazon.com/entityresolution/latest/userguide/) · [Pricing](https://aws.amazon.com/entity-resolution/pricing/)

Best for: AWS-native teams that want rule-based or ML entity resolution on S3 data without running infrastructure.

## Overview

- Core Strength: Managed entity resolution on S3, priced per record
- Matching: Rule-based, ML or data provider matching
- Golden Records: Match groups with a match ID; no golden records
- Platform Fit: AWS only: S3, Glue, Clean Rooms
- Deployment Model: AWS managed service
- Pricing: $0.25 per 1,000 records
- Notable Caveat: No steward review; incremental runs not for ML matching

## What it's used for

AWS Entity Resolution is used to match customer, product or business records stored in Amazon S3 without running infrastructure: rule-based, ML or data provider workflows group related records and give each group a shared match ID.

## Pricing

AWS Entity Resolution charges $0.25 per 1,000 records processed for rule-based or ML matching and $0.10 per 1,000 for data provider workflows, plus the provider's subscription. Every processed record counts, matched or not, and there is no free tier.

## Deployment

AWS Entity Resolution is a managed AWS service: it reads records from Amazon S3 and AWS Glue tables in your account and writes match output back to your S3 bucket.

## Review

Strengths:
- Nothing to run: a managed service that reads S3 and Glue tables and writes match groups back to your S3 bucket.
- Rule-based, ML or data provider matching (LiveRamp, TransUnion), with a match ID on every record.
- Priced per record, and incremental rule-based runs only charge for new records.

Caveats:
- No golden records, survivorship or steward review: the output is match groups.
- Incremental and near real-time matching work with rules only, not with the ML model.
- AWS only, with no free tier, and every processed record is billed, matched or not.

Recommendation: For a real-time engine with a GraphQL API on AWS, consider Tilores; for ML matching you train yourself inside Databricks or Fabric, Zingg; for golden records and steward review, Reltio.

## Pricing

- Model: per 1,000 records
- Entry team price: $0.25 per 1,000 records (rule-based or ML)
- Source: https://aws.amazon.com/entity-resolution/pricing/ (checked 2026-11-20)

## Features

### Matching

- ML Matching: Yes — "A preconfigured ML model to find matches" with a confidence score. ([source](https://aws.amazon.com/entity-resolution/features/))
- Rule-Based Matching: Yes — "Ready-to-use rule-based matching" you can reorder, delete or create. ([source](https://aws.amazon.com/entity-resolution/features/))
- Real-Time Matching: Yes — Near real-time resolution for rule-based matching. ([source](https://aws.amazon.com/entity-resolution/features/))
- Incremental Matching: Yes — Incremental processing "compares them against existing records" (not for ML matching). ([source](https://aws.amazon.com/entity-resolution/features/))
- Persistent Entity ID: Yes — Generates a match ID for every record, retrievable later via Get Match ID. ([source](https://aws.amazon.com/entity-resolution/faqs/))
- Match Explainability: Yes — Each match group carries the rule number or an ML confidence score. ([source](https://aws.amazon.com/entity-resolution/faqs/))

### Golden Records

- Survivorship Rules: Unknown — A consolidated output table is mentioned; attribute rules aren't. ([source](https://aws.amazon.com/entity-resolution/features/))
- Survivorship on Read: No — Output is stored match groups with a match ID. ([source](https://aws.amazon.com/entity-resolution/features/))
- Stewardship Review: No — The console shows job status and metrics, no review queue. ([source](https://aws.amazon.com/entity-resolution/features/))
- Record History: No — No record or match-group history is described. ([source](https://aws.amazon.com/entity-resolution/features/))
- Hierarchy Management: No — Flat match groups; no hierarchies. ([source](https://aws.amazon.com/entity-resolution/features/))

### Domains & Modeling

- Multi-Domain MDM: No — Workflows run on S3 inputs; no managed per-domain store. ([source](https://aws.amazon.com/entity-resolution/features/))
- Configurable Data Model: Unknown — You build a schema mapping for input fields, not entity types or relationships. ([source](https://aws.amazon.com/entity-resolution/features/))
- Product Information Management: No — Matches SKUs and UPCs but manages no product content. ([source](https://aws.amazon.com/entity-resolution/faqs/))
- Reference Data Management: Unknown — ID namespaces exist; code-list management is not described. ([source](https://docs.aws.amazon.com/entityresolution/latest/userguide/what-is-service.html))
- Master Data Authoring: No — The console configures matching workflows, not master records. ([source](https://aws.amazon.com/entity-resolution/features/))

### Data Quality

- Validation at Entry: No — Only normalization before matching. ([source](https://aws.amazon.com/entity-resolution/features/))
- Standardization: Yes — "Data inputs are also normalized prior to matching". ([source](https://aws.amazon.com/entity-resolution/features/))
- Third-Party Enrichment: Yes — "Enhance your records with datasets and IDs from trusted data service providers" (LiveRamp, TransUnion, UID 2.0). ([source](https://aws.amazon.com/entity-resolution/features/))

### Data Distribution

- REST API: Yes — Generate Match ID and Get Match ID APIs. ([source](https://aws.amazon.com/entity-resolution/features/))
- Change Events: Unknown — No event output from matching jobs is described. ([source](https://aws.amazon.com/entity-resolution/features/))
- Warehouse Write-Back: No — Output goes to your S3 bucket. ([source](https://aws.amazon.com/entity-resolution/features/))
- Data Syndication: Unknown — Translated IDs interoperate with advertising platforms; channel-format publishing is not described. ([source](https://aws.amazon.com/entity-resolution/faqs/))

### AI & Agents

- MCP Server: Unknown — No MCP server is described. ([source](https://aws.amazon.com/entity-resolution/features/))
- AI Stewardship Assistance: No — ML gives a confidence score; no steward-facing AI. ([source](https://aws.amazon.com/entity-resolution/features/))

### Platform Native

- In-Place Matching: Yes — It "reads your records where they already live", from Amazon S3 and AWS Glue tables, and writes match groups to your S3 bucket. ([source](https://aws.amazon.com/entity-resolution/features/))
- Snowflake Native: No — Reads S3 data on AWS. ([source](https://aws.amazon.com/entity-resolution/features/))
- Databricks Native: No — Reads S3 data on AWS. ([source](https://aws.amazon.com/entity-resolution/features/))
- Microsoft Fabric Native: No — Reads S3 data on AWS. ([source](https://aws.amazon.com/entity-resolution/features/))

### Deployment

- SaaS Deployment: Yes — Managed AWS service with nothing to install. ([source](https://aws.amazon.com/entity-resolution/))
- Self-Hosted Deployment: No — Runs only in AWS Regions. ([source](https://aws.amazon.com/entity-resolution/pricing/))

### Open Source

- Open Source: No — Paid AWS service, "per 1,000 records processed". ([source](https://aws.amazon.com/entity-resolution/pricing/))

## Integrations

5 integrations, each sourced from the vendor docs.

- Data Enrichment: LiveRamp, TransUnion
- Data Integration: AWS Glue
- Data Platform: AWS Clean Rooms
- Data Storage: Amazon S3

## Alternatives

- [Tilores](https://toolsfordata.com/tools/tilores/): The closest real-time alternative on AWS: rules plus ML scoring as records arrive, an auditable entity graph and a GraphQL API, managed or in your own account.
- [Zingg](https://toolsfordata.com/tools/zingg/): Consider Zingg to train your own ML model on 40 to 50 labeled pairs and match inside Databricks, Fabric, AWS or GCP, with incremental runs in the paid editions.
- [Senzing](https://toolsfordata.com/tools/senzing/): Consider Senzing to resolve each record in real time inside your own application, with built-in rules, no training and explanations for every match.
- [Splink](https://toolsfordata.com/tools/splink/): The free option: probabilistic record linkage in Python on DuckDB or Spark, estimated without labels, if you can run the jobs yourself.
- [Reltio by SAP](https://toolsfordata.com/tools/reltio/): Consider Reltio when match groups aren't enough: a SaaS MDM hub with field-level survivorship, steward review and unmerge.
