Core Concepts:

Identity Resolution Technologies

At the heart of Identity Resolution are several core concepts and technologies that work together to link and consolidate data.

Key Concepts

• Identifiers: These are the data points used to recognize an individual. They range from Personally Identifiable Information (PII) like names and email addresses to pseudonymous data like cookie IDs and device IDs.
• Identity Graph: This is the central database that stores all identifiers linked to a customer, creating a map of their interactions. It merges the complete history of each customer into a single, unified profile.

Different Data Matching Methods

• Deterministic Matching: This high-confidence method links data based on exact matches of known identifiers, such as a shared email address or phone number. It is highly accurate but limited by the availability of exact matches.
• Probabilistic Matching: Also known as “fuzzy matching,” this approach uses statistical algorithms to infer the likelihood that different data points belong to the same person. It analyzes non-PII data like IP addresses, device types, and Browse behavior to make educated guesses, offering broader reach but lower accuracy than deterministic matching
• Real-Time Principle Based Matching: The latest evolution of intelligent data matching abandons the need for thousands of complex, hard-coded rules. Instead, it operates on a set of universal principles—generalized, built-in knowledge about how identifying attributes behave in the real world. This makes the system incredibly flexible, allowing it to work with high accuracy out of the box and add new data sources with unprecedented ease. 

Enabling Technologies

• Customer Data Platforms (CDPs): These platforms centralize customer data from various sources to build unified profiles and often include Identity Resolution capabilities.
• Identity Resolution Platforms: These are specialized tools dedicated to connecting consumer identifiers across different channels and devices.
• Artificial Intelligence (AI) and Machine Learning (ML): AI and ML enhance accuracy through advanced pattern recognition, automate the linking of data, and enable real-time resolution. A modern Customer Identity Resolution AI can analyze vast datasets to handle inconsistencies and predict fraudulent behavior.
• Graph Databases: These databases are structured to efficiently store and navigate the complex relationships between identifiers, making them ideal for resolving large identity networks.

What’s The Difference?

Entity vs Identity Resolution

Identity Resolution is a specific subset of a broader data quality and management process called Entity Resolution. While Identity Resolution aims to identify and link records corresponding to people, Entity Resolution applies the same process to any real-world “entity,” which can include people, organizations, locations, accounts, or products.

Identity Resolution in Relation to Entity Resolution

Identity Resolution is a specific subset of Entity Resolution that focuses exclusively on identifying and linking information related to individuals. While the terms are sometimes used interchangeably, Identity Resolution has a narrower focus on individuals, often within marketing, customer relationship management, and personalization.

Entity Resolution is a broader data management process focused on identifying, linking, and merging records that correspond to the same real-world entity across various data sources. An “entity” can be any distinguishable real-world object, including people, organizations, locations, products, or events. The goal is to create a single, unified, and accurate representation of each entity.

Both employ similar underlying techniques for record linkage, data matching, and deduplication, including deterministic and probabilistic matching algorithms, data cleansing, normalization, and the use of identity graphs or entity registries. However, Entity Resolution has broader applications in areas like supply chain management, product information management, and regulatory compliance, while Identity Resolution primarily serves user and customer-centric use cases like marketing, customer service, and personalization.  Learn more in-depth about Entity Resolution.

In the context of fraud, while Identity Resolution is crucial for creating a unified view of individuals and detecting individual bad actors, the broader capabilities of Entity Resolution can help uncover more complex, multi-entity fraudulent schemes by linking related entities like businesses, addresses, or IP addresses that are part of a coordinated operation.

A Critical Defense:

Identity Resolution for Fraud Detection & Prevention

Fraudsters often open multiple accounts using slight variations of names, email addresses, and other personal information across different channels. In isolation, each application or transaction might appear legitimate. Without a unified view, financial institutions, e-commerce businesses, and government agencies struggle to detect these coordinated fraudulent activities, leading to severe consequences:

• Increased Financial Losses: Undetected fraudulent transactions and applications result in significant financial losses.
• Security Breaches: Bad actors exploit fragmented data to gain unauthorized access to sensitive information.
• Inefficient Investigations: Investigating disconnected data silos is time-consuming and often yields incomplete results, allowing criminals to evade detection.
• Erosion of Trust: Customers and citizens lose faith in organizations that cannot protect them from fraud or manage their data securely.

How Identity Resolution Uncovers Malicious Activity

Identity Resolution tackles these vulnerabilities by creating a single, persistent view of each individual, allowing organizations to connect seemingly unrelated activities. The focus shifts from simple data consolidation to identifying patterns and anomalies that indicate malicious intent.

This is achieved through several enhanced techniques:

• Data Ingestion from High-Risk Sources: In addition to standard customer data, organizations incorporate information from sources relevant to fraud detection, such as IP address logs, device fingerprinting data, transaction histories, and watchlists.
• Link Analysis: By visualizing connections between data points, analysts can uncover hidden relationships. For example, multiple accounts using different names but sharing the same phone number or device ID can reveal a potential fraud ring.
• Behavioral Analytics Integration: Identity Resolution can be combined with behavioral analytics to establish a baseline of normal user behavior. Any deviation from this baseline, such as a sudden change in transaction volume, can trigger an alert for suspicious activity.
• Real-Time Risk Scoring: By resolving an individual’s identity in real-time during a transaction or login attempt, organizations can assess the associated risk based on historical data and behavioral patterns, preventing fraud before it occurs.

Critical Applications in Combating Fraud

The application of Identity Resolution for fraud prevention spans various sectors:

• Financial Services: Banks use it to detect account takeovers, credit card and loan fraud, and money laundering by linking seemingly unrelated accounts and transactions.
• E-commerce: Online retailers can prevent fraudulent purchases and reduce chargebacks by linking Browse behavior, purchase history, and device information to flag suspicious orders.
• Government Agencies: Public sector agencies use Identity Resolution to prevent individuals from fraudulently claiming benefits under multiple identities, assist law enforcement in tracking criminal networks, and help tax agencies identify tax evasion schemes.

DIY Danger:

Should You DIY In-House Identity Resolution?

The complexity of accurately linking diverse and evolving data sources requires deep expertise in matching algorithms and data structures. A DIY system may lack the flexibility to adapt to new channels and identifiers, demanding constant and costly maintenance. Furthermore, simply identifying users is not enough; the resolved data must be accessible and actionable for business teams, which requires a comprehensive suite of integrated tools.

Attempting an in-house Identity Resolution solution presents significant pitfalls:

• Evolving Complexity: Customer engagement strategies constantly evolve, with new channels, platforms, and identifiers emerging. Initial DIY logic might not be flexible enough to accommodate these changes, requiring constant updates.
• Inherent Complexity: Building a robust Identity Resolution solution requires a deep understanding of data structures, matching algorithms (deterministic and probabilistic), and data quality issues.
• Underestimation of Tooling Needs: Simply identifying unified customer profiles is not enough; the data needs to be accessible and usable by various platforms. This typically requires investments in event streaming, ETL, reverse ETL, data warehousing, and data transformation tools, along with the expertise to connect and maintain them.
• Ongoing Maintenance and Resources: Identity Resolution is not a one-time project; it demands continuous monitoring, testing, debugging, and auditing to ensure accuracy and reliability. Staying compliant with evolving legal, regulatory, and privacy requirements adds another layer of complexity and necessitates dedicated resources.
• Lack of Expertise and Perspective: Data scientists, while skilled, might underestimate the complexity of building a real-time, principle-based Identity Resolution system that is both highly accurate and provides explainable results.
• Significant Budget (more than you’d think): To develop a fast, accurate and reliable Identity Resolution system is notoriously difficult and shockingly expensive.  We’ve worked with organizations who have spent millions in development, only to throw in the towel and use our off-the-shelf Identity Resolution SDK (with better results).

A Robust Solution:

Advantages of a Pre-Built Solution for Identity Resolution