Entity Resolution, Explained: The Importance of Data Matching

The process of working with data to get to that singular, verified view of a business is called entity resolution. Within the wide scope of master data management (MDM), entity resolution is one of the most critical activities for bringing structure to disordered data in pursuit of the “single source of truth.” It’s widely regarded by data management experts as a key measure of success of data mastering programs.

As defined by Gartner®, entity resolution is “the capability to consolidate multiple labels for individuals, products, or other data classes into a single resolved entity and analyze relationships among these entities.”¹ As Gartner asserts in its 2024 Market Guide for Master Data Management Solutions: “Broadly speaking, entity resolution can be equated with data quality.”
 

Entity Resolution and Data Matching: Is There a Difference?

Essentially, these terms are interchangeable among data experts. To resolve an entity, you need to successfully match a collection of data points against a trusted reference to identify the object that those data points truly represent. This process can be simple, or it can be complex, depending on factors including completeness of the data and capability of the process — human- or machine-driven — to fulfill the match request.

Whether you call it “matching” or “entity resolution,” it’s challenging and resource intensive. It seems like a simple concept to grasp, but it’s actually a difficult science to perfect. Understanding the limitations of their ability to develop robust in-house data harmonization, deduplication, and matching solutions is often what drives companies to pivot from home-grown master data solutions to those offered by an external MDM provider.
 

Benefits of Entity Resolution: Optimizing Your Data

Matching, a critical first step to take in data management, helps to ultimately optimize data by keeping it up to date, complete, and relevant.

The matching process aggregates data from across the enterprise, eliminating data silos and enabling identification and merging of duplicate records. The goal is to achieve accurate, complete views of entities (customers, suppliers, prospects, etc.). Once matched, “golden records” can be enriched with third-party data to reveal more about these companies. Incomplete fields can be corrected, and inconsistent formatting can be addressed. Additional data fields can be appended to the data, and related data can be linked, whether to give a sales team more context or a finance team a bigger picture of the customer. 

It’s important to note that when building out an entity resolution initiative, a flawed implementation can have a ripple effect, putting a company at risk of fragmented revenue reporting, data privacy concerns, and regulatory and compliance issues. 
 

Entity Resolution Use Cases

There are many use cases that rely on proper identification of businesses. Here are just a few examples:

Audience Targeting 

Entity resolution creates greater confidence that the data being leveraged for marketing campaigns is accurate and that the right targets are being included in segmentation, list building, message delivery, and personalization. When customer profiles are generated with accurate entity data, marketing spend is optimized and campaigns can yield higher engagement and conversion rates. And the resulting analytics from executed campaigns can help marketers design more effective future strategies. 

Regulatory Compliance 

When the identity of a customer or third party is resolved, due diligence processes — determining ultimate beneficial ownership (UBO), detecting watchlist or sanctions violations, uncovering corrupt or criminal activity — become more streamlined and accurate. Less time is required to gather information for investigations, and the number of false positives is reduced. 

Fraud Detection 

Particularly in financial institutions, entity resolution can help uncover duplicate records and create a unified view of customer accounts; this makes it easier to detect unusual activity that might result from attempted fraud. During onboarding, a bank can use entity resolution to link multiple accounts with similar details that may represent an incipient money laundering scheme.

Supply Management

Disconnected supplier entity data may be spread among several different departments or divisions of an enterprise. When that disconnected data is brought together to create a unified supplier profile, it creates strategic advantages in sourcing and negotiating. Supplier matching also enables a better understanding of the hierarchies and ownership of suppliers, leading to better management of risks associated with geographies, financial status, and other factors. 

Customer Relationship Management

Customer names may be entered in a database multiple times, with different name spellings, addresses, and phone numbers; for example, a company like JPMorganChase might be entered as JP Morgan Chase, JP Morgan, JPMC, and so on. In this case, entity resolution serves the purpose of deduplication, creating a single clean record in the CRM system that will prevent duplicate emails, poor customer experiences, and missed sales opportunities.

Consider this scenario: The finance team has a customer record that they want to match with a corresponding record from the sales team to consolidate information. The organization can match company name to company name. But maybe the finance team spelled the company name differently than the sales team did. There may be other attributes, such as the business address field, that can be used to match these two records for a common view of the customer across the business.

But then think about the magnitude of matching and merging hundreds, thousands, or even millions of records with multiple parameters of varying completeness and quality. This is where it gets real for many enterprises — and where the value of entity resolution within MDM is demonstrated.
 

Entity Resolution and Referential Matching

We mentioned earlier that entity resolution requires a collection of data points to be matched against a reliable reference to produce a consolidated, verified record. Referential data — another important consideration for MDM — is external data that an organization can generally trust to be accurate, complete, and consistent, regardless of how it’s internally applied.

External referential data is like a “window to the outside world.” It comes with subject matter expertise and built-in data governance around a given entity and the attributes associated with it, such as:

• Standard definitions for what is, and what is not, a business

• Global, cross-border consistency in the data, which improves accuracy and completeness

• Segmentation data to drive analytics like industry codes, sales figures, and employee counts

• Relationships to other business entities

Referential data sets a standard of quality that can show you where there are opportunities for improvement in your own data. This helps to make data maintenance more scalable and sustainable.

To leverage the benefits of a reference data set, a match between the source data of a company and the reference data must be made. Referential matching identifies duplicates, links records, and provides a reliable foundation for improving data quality while driving efficiencies in data management processes.  

The best entity resolution processes involve bringing first-party data (and second-party data, if available) together and enriching it with third-party referential data to benefit a broad range of enterprise functions. Generally, the matching of first-party data to referential data is made using third-party data providers’ proprietary processes that exist outside the MDM infrastructures of their customers.
 

How the Entity Resolution Process Works

Before starting the matching process, data teams should start small and then apply what they learn to additional sets of data. Data mastering — which has entity resolution as a component — can be prioritized by domain, business segment, sales channel, geography, vertical market, or application. The business use case should also be considered, as the level of matching stringency can vary for different cases. For example, compliance and risk use cases often require very stringent criteria for entity resolution; sales and marketing segmentation can have broader acceptance criteria. 

MDM applications employ sophisticated matching algorithms to identify and merge duplicate records using combinations of exact and fuzzy matching techniques. Generally, the entity resolution process utilizes a method called clustering, where the goal is to create a group (cluster) of candidates which are all likely to share the same attributes. The MDM software will use powerful statistical algorithms to make inferences about candidates for the cluster, where logical links can be made based on similarities shared across entities. The more records being matched, the more resources are required by those algorithms to complete the matching process. 

For example, if there were only 5 records being matched together in an MDM hub, there would be 10 possible combinations that the MDM software would, at a minimum, need to evaluate: 

| 1-2 | 1-3 | 1-4 | 1-5 | 2-3 | 2-4 | 2-5 | 3-4 | 3-5 | 4-5 |

This process is further complicated by the fact that two records, when individually compared, may not necessarily be obviously related to each other. They may actually be related via a shared connection to a third candidate. 

To help improve possible match results beyond one-to-one matching, the match algorithms used by MDM software will try to draw inferences across multiple candidates. Match associations, with varying degrees of probability, can be made in situations where two records indirectly share a characteristic through a third candidate. 

For instance, if Candidate Record A is related to Candidate Record B (for example, due to similarities in the company name), and Candidate Record B is related to Candidate Record C (for example, by sharing a similar address), then it’s logical to assume that Candidate Record A is probably or potentially related to Candidate Record C. This process, usually referred to as transitive matching, relies on the application of complex match algorithms. The accuracy and efficiency of those varied algorithms is a competitive differentiator of MDM software providers.

Clustering records together can help data teams see commonalities that can be applied to their organization’s data governance practices and data stewardship activities. One example: a pattern in the way customer names and/or addresses are being input — for example, including ancillary comments like “do not use” or “DBA” — which can be adjusted to improve match rates. 
 

Alternative Matching Techniques 

Data doesn’t naturally exist in a vacuum, and the data going through the entity resolution process may be quite aged. Many data points — company names, physical locations, phone numbers, associated individuals — may have changed since the data was initially collected. This can result in unmatched and low-confidence records that still need to be resolved so that the quality of the data set can be maintained.

In these cases, alternative matching can be leveraged to increase match and enrichment rates. Alternative matching techniques can include:   

• Matching in the local country language of record origin and using language translation tools

• Matching using alternate data attributes like National ID and/or URL

• Multi-pass matching: mining additional entity name and address fields (like Ship-to) that may exist inside internal documents such as tables

• Performing company look-ups to try to narrow down the match candidates and to establish rules and data governance practices that will improve match and enrichment. An example: associating a record to a family tree if there is a unique business name but not enough address information available to differentiate one location from another.

One example of the payoff of alternative matching: A Dun & Bradstreet client was trying to improve match rates for its customer data. The client’s data team noticed that match rates for entities in France were lagging behind those in other regions. Once it was discovered that France uses a national identification number, this data attribute was incorporated for matching, and the match rate for French entities immediately jumped by 12 percentage points. (Read the full case study.)
 

Entity Resolution Best Practices Are MDM Best Practices

Successful entity resolution initiatives do more than enrich business intelligence on crucial business entities. They also provide scalable data practices that work within the framework of the company’s overall data mastering program. MDM is not a one-time project; new records and change notifications come in every day, along with opportunities for learning and refining the program. Through the adoption of entity resolution best practices alongside monitoring and change management, enterprises are able to not just sustain but continue to improve data accuracy and internal efficiency.  

The landscape of data today has undergone rapid change, with even more dramatic shifts still to come in the era of generative AI. Where data is concerned, it’s important to know what you don’t know — and the criticality of entity resolution as part of an MDM program is a prime example of why it makes sense to consult with data experts. 

Dun & Bradstreet has a team of MDM experts who specialize in optimizing match performance for specific use cases. These data professionals can help you convert data from an obstacle to an asset that brings your company’s ultimate business goals within closer reach.

¹ Gartner, Market Guide for Master Data Management Solutions, 17 December 2024. GARTNER is a registered trademark and service mark of Gartner, Inc. and/or its affiliates in the U.S. and internationally and is used herein with permission. All rights reserved.