Identity Resolution in Marketing Analytics: How to Build a Unified Customer View

Analytics team reviewing connected customer data for identity resolution

Why Identity Resolution Matters More Than Ever

A customer rarely interacts with a business through only one device, browser, or marketing channel.

Someone may discover a brand through paid social on their phone, visit the website later from a work laptop, return through organic search, sign into an account on a tablet, open an email, and eventually complete a purchase through a mobile application.

From the customer's perspective, this is one relationship with one company.

From an analytics perspective, it can look like several completely different people.

That creates one of the most difficult problems in modern marketing measurement:

How do you determine when multiple interactions belong to the same customer?

Identity resolution is the process of connecting identifiers and customer interactions across systems so organizations can build a more consistent understanding of the customer journey.

Without it, businesses can:

  • Overcount users

  • Fragment customer journeys

  • Misunderstand acquisition

  • Duplicate audiences

  • Miscalculate Customer Lifetime Value

  • Deliver inconsistent personalization

  • Waste media spend

  • Struggle to connect marketing with revenue

As marketing technology becomes more sophisticated, identity resolution becomes increasingly important.

But it also needs to be implemented carefully.

The objective is not to identify every anonymous person on the internet.

The objective is to responsibly connect customer interactions when the organization has appropriate identifiers and a legitimate reason to do so.

What Is Identity Resolution?

Identity resolution is the process of determining which identifiers, devices, sessions, transactions, and interactions likely belong to the same customer.

Imagine an organization sees these identifiers:

Website browser:
client_id = 84739

Mobile application:
app_instance_id = 39281

CRM:
customer_id = C10245

Ecommerce system:
account_id = 77142

Individually, these appear to describe four separate identities.

But once the customer signs into both the website and mobile application, the organization may be able to establish that these identifiers belong to:

customer_id = C10245

Those separate interactions can now be associated with a common known customer, subject to the organization's privacy and governance requirements.

This process creates an identity graph or identity mapping that connects identifiers together.

Identity Is Not the Same Thing as a Cookie

Cookies are often discussed when marketers talk about identity.

But a cookie is only one type of identifier.

Organizations may encounter identifiers such as:

  • Browser identifiers

  • Device identifiers

  • Customer IDs

  • Account IDs

  • CRM IDs

  • Loyalty IDs

  • Subscription IDs

  • Hashed identifiers

  • Transaction identifiers

Some identify a browser.

Some identify a device.

Some identify an account.

Some represent a known customer.

Understanding the difference is essential.

A browser identifier should not automatically be treated as proof of a person's identity.

Anonymous Identity vs. Known Identity

A useful identity architecture distinguishes between anonymous and known states.

Anonymous Identity

Before a customer identifies themselves, the organization may only recognize a browser or device.

For example:

anonymous_id = A89273

The organization knows that the same browser performed several actions.

It does not necessarily know who the person is.

Known Identity

Later, the customer may:

  • Sign into an account

  • Complete an authenticated transaction

  • Join a loyalty program

  • Identify themselves through another approved interaction

The organization can then associate future or permitted historical interactions with a known customer ID.

For example:

customer_id = C10245

The important distinction is that identity resolution should rely on defined identity signals rather than assumptions.

Expert Insight: Identity Resolution Is Really an Evidence Problem

The question is not whether two records look similar. The question is whether there is enough reliable evidence to conclude that they belong to the same customer.

This distinction matters.

Suppose two people:

  • Share the same IP address

  • Live in the same household

  • Use the same device

That does not necessarily make them the same person.

A strong identity framework therefore assigns different levels of confidence to different signals.

A login tied to a customer account may provide strong deterministic evidence.

Similar browsing behavior provides much weaker evidence.

Treating weak signals as certainty can create inaccurate customer profiles.

Deterministic Identity Resolution

Deterministic matching relies on known identifiers that provide a direct relationship between records.

Examples may include:

  • Customer ID

  • Account ID

  • Loyalty ID

  • Authenticated user ID

Suppose a customer signs into the same account on:

  • Desktop

  • Mobile web

  • Mobile application

Each environment may initially have a different device identifier.

But the authenticated customer ID provides a common reference.

That allows the organization to connect those environments much more confidently.

Deterministic matching is generally easier to explain and audit because the relationship is based on an explicit identifier.

Probabilistic Identity Resolution

Probabilistic identity resolution attempts to estimate whether records belong to the same person based on multiple signals.

Potential signals might include:

  • Device characteristics

  • Geography

  • Timing

  • Behavioral patterns

  • Network information

A model might conclude that two device profiles are highly likely to belong to the same customer.

But probability is not certainty.

This creates additional considerations around:

  • Accuracy

  • False matches

  • Privacy

  • Governance

  • Appropriate use

For many organizations, deterministic first-party identity should be the foundation before considering more complex probabilistic approaches.

What Is an Identity Graph?

An identity graph is a structured representation of relationships between identifiers.

Consider:

Customer ID C10245

connected to:

  • Browser ID A734

  • Browser ID B929

  • Mobile App ID M382

  • Loyalty ID L10245

The graph provides a way to understand that multiple identifiers are associated with the same customer profile.

As additional authenticated interactions occur, the graph can evolve.

The important point is that the customer ID acts as the durable business identity, while device identifiers describe environments through which that customer interacted.

Identity Resolution and the Customer Journey

Without identity resolution, customer journeys frequently become fragmented.

Imagine a customer:

  1. Clicks a social advertisement on mobile

  2. Browses products

  3. Returns through organic search on a laptop

  4. Creates an account

  5. Receives an email

  6. Purchases three days later

Without identity resolution, reporting might show:

User 1: Paid social visitor

User 2: Organic search visitor

User 3: Email purchaser

The business may conclude that email generated the customer.

In reality, email may have been only the final interaction in a much longer journey.

Identity resolution does not automatically solve attribution.

But it improves the underlying customer-journey data that attribution depends on.

Identity Resolution and Omnichannel Analysis

Identity becomes even more important when customers move between channels.

Consider a retailer with:

  • Website

  • Mobile app

  • Physical stores

  • Email

  • Loyalty program

A customer may research online and purchase in-store.

Without a shared identifier, the website activity and store transaction may never connect.

With a loyalty or account identifier, the organization may gain a more complete view of the relationship.

This is one of the foundations of effective Omnichannel Analysis.

The objective is not simply to track more interactions.

It is to understand how those interactions relate to the same customer journey.

Why Authentication Is So Valuable

Authentication provides one of the strongest opportunities for deterministic identity resolution.

When customers sign in, organizations can establish a consistent account identifier across:

  • Sessions

  • Devices

  • Applications

This is one reason businesses increasingly consider the value exchange around authenticated experiences.

Customers may receive:

  • Saved preferences

  • Faster checkout

  • Order history

  • Loyalty benefits

  • Personalized recommendations

The business receives a more stable first-party relationship.

This can strengthen First-Party Data Activation by making customer data more useful across analytics and marketing systems.

The Customer ID Should Be Stable

A strong identity system depends on stable identifiers.

A customer ID should ideally remain consistent over time.

Avoid designing the identity architecture around values that frequently change.

For example, email addresses can change.

A customer might:

  • Change employers

  • Update their personal email

  • Use several email addresses

The internal customer ID should generally remain stable while attributes such as email can change around it.

This creates a more durable identity model.

Do Not Use Email as Your Entire Identity Architecture

Email can be valuable for matching in specific approved use cases.

But using raw email addresses as the primary customer key creates several problems.

Email addresses:

  • Can change

  • Can be shared

  • May have multiple versions

  • Contain personal information

A stronger internal architecture typically uses a durable customer identifier.

The relationship might look like:

customer_id = C10245

associated with approved attributes and identifiers maintained by the customer system.

That separates the business identity from a particular contact field.

Identity Resolution and Customer Lifetime Value

Customer Lifetime Value becomes difficult to calculate when transactions cannot be connected to the same customer.

Suppose one person makes four purchases using multiple devices.

Without identity resolution, analytics may treat those purchases as coming from several different users.

The organization may underestimate:

  • Purchase frequency

  • Retention

  • Customer lifespan

  • Lifetime revenue

Identity resolution allows organizations to build more accurate longitudinal customer histories.

That can improve:

  • CLV analysis

  • Cohort analysis

  • Retention measurement

  • Customer segmentation

Identity Resolution and Audience Segmentation

Audience segmentation also depends on reliable identity.

Imagine the business wants to create a segment of:

High-value customers who purchased at least three times during the last 12 months.

If one customer's transactions appear under several disconnected identifiers, that customer may never qualify.

A stronger identity layer allows Audience Segmentation to operate on customer-level behavior rather than isolated device activity.

This can improve targeting for:

  • Loyalty campaigns

  • Retention programs

  • Cross-selling

  • Suppression

  • Personalization

Identity Resolution Can Reduce Audience Duplication

Fragmented identity can also cause the same customer to appear in several marketing audiences.

For example, one customer might simultaneously appear as:

  • New prospect

  • Existing customer

  • Cart abandoner

  • Recent purchaser

If systems do not communicate effectively, the customer could receive conflicting campaigns.

A recent purchaser might continue receiving acquisition advertising.

A loyal customer might receive a first-purchase discount.

A unified identity strategy can help reduce these contradictions.

Identity Resolution and Advertising Suppression

One practical application is customer suppression.

Suppose an organization is running an expensive new-customer acquisition campaign.

Existing customers may not need to receive those ads.

A reliable first-party identity framework can help create approved suppression audiences.

This can reduce unnecessary media exposure and improve customer experience.

However, the effectiveness of suppression still depends on:

  • Match rates

  • Platform capabilities

  • Consent

  • Data quality

Identity resolution improves the foundation, but it does not guarantee perfect matching.

Identity Resolution and Personalization

Personalization also becomes more useful when customer identity is consistent.

Without identity resolution, a returning customer may repeatedly receive generic experiences because every device appears to be new.

With a known customer relationship, organizations may be able to personalize experiences based on:

  • Previous purchases

  • Customer status

  • Product interests

  • Loyalty tier

  • Lifecycle stage

That can support broader Personalization at Scale.

But personalization should always have a clear customer benefit.

Recognizing a customer simply because technology allows it does not automatically create a better experience.

The Role of the Data Layer

Identity should be considered when designing the data layer.

For authenticated experiences, events may need an approved internal customer identifier in addition to anonymous session or device identifiers.

For example:

event = purchase

customer_id = C10245

transaction_id = 984573

The data layer should not expose unnecessary sensitive information.

Instead, it should provide the minimum approved identifiers needed for the measurement architecture.

This connects identity resolution directly with broader data-layer governance.

Identity Resolution and Server-Side Tracking

Server-side architecture can provide another controlled point for managing identity.

For example:

Browser

First-Party Endpoint

Identity and Event Processing

Approved Destinations

The server may receive:

  • Anonymous identifier

  • Authenticated customer ID

  • Consent state

  • Business event

It can then determine which information is appropriate for each downstream destination.

This creates an opportunity for:

  • Validation

  • Data minimization

  • Identifier transformation

  • Routing

But server-side tracking does not remove privacy obligations.

Identity use still needs to follow the organization's approved governance framework.

Identity Resolution and the Data Warehouse

For many organizations, the data warehouse becomes the strongest environment for customer-level identity analysis.

A warehouse may contain:

  • Transactions

  • CRM records

  • Website events

  • App events

  • Customer support information

  • Loyalty data

These datasets can be connected through defined customer identifiers.

The resulting model might look conceptually like:

Customer Table

customer_id

Transaction Table

customer_id + transaction_id

Behavior Table

customer_id + events

Marketing Table

customer_id + acquisition data

A centralized architecture can support:

  • Customer Lifetime Value

  • Segmentation

  • Retention analysis

  • Cohorts

  • Predictive analytics

Strong Data Engineering is critical for maintaining these relationships consistently.

Identity Resolution and a Single Source of Truth

Identity is also fundamental to creating a single source of truth.

Different systems frequently assign their own identifiers.

For one customer:

CRM might use:

CRM-10823

Ecommerce might use:

EC-98373

Loyalty might use:

LT-34821

A centralized identity map can establish:

customer_id = C10245

and connect the system-specific identifiers to it.

This does not mean every system needs to use exactly the same database key.

It means the organization has a documented way to reconcile them.

Identity Resolution Requires Clear Source Hierarchies

Organizations should establish which system is authoritative for different identity attributes.

For example:

Customer ID

Authoritative source: CRM

Account Status

Authoritative source: subscription system

Transaction

Authoritative source: ecommerce database

Website Behavior

Authoritative source: analytics event stream

This prevents teams from treating every system as equally authoritative.

Identity resolution is much easier when source ownership is explicit.

Identity Stitching Can Create Serious Errors

One of the biggest risks in identity resolution is incorrectly merging two different people.

Consider a shared household computer.

Person A browses products.

Person B later signs into their account on the same device.

If the system automatically associates all previous device activity with Person B, some of Person A's behavior may be incorrectly attached to Person B.

This is sometimes called false stitching.

The consequences can include:

  • Incorrect personalization

  • Distorted customer profiles

  • Bad segmentation

  • Misleading analytics

Identity rules should therefore define carefully when anonymous historical activity can be connected to a known identity.

Shared Accounts Create Another Challenge

Sometimes multiple people intentionally use one account.

Examples might include:

  • Household streaming accounts

  • Family retail accounts

  • Business purchasing accounts

In these cases:

Account identity is not necessarily the same as person identity.

Analytics teams should understand what the identifier actually represents.

If an account ID represents a household or organization, reporting should not automatically describe it as an individual person.

Terminology matters.

Logout Behavior Matters

Identity handling should also define what happens when users log out.

Imagine:

  1. Customer A signs into a shared device.

  2. Customer A logs out.

  3. Customer B begins browsing anonymously.

If Customer A's identifier remains attached to subsequent events, Customer B's behavior could incorrectly be attributed to Customer A.

Implementations should explicitly clear or change identity state when authentication changes.

This should be part of analytics QA.

Identity QA Should Be Part of Tracking QA

Identity should be tested just like events and parameters.

A QA plan might include:

Anonymous Visit

Confirm no known customer identifier is assigned.

Login

Confirm the correct customer ID becomes available.

Cross-Device Login

Confirm the same internal customer ID is used.

Logout

Confirm the known identity is cleared appropriately.

Account Switch

Confirm one customer's identity does not persist into another customer's session.

Purchase

Confirm the correct customer and transaction identifiers are associated where intended.

These scenarios can reveal serious measurement issues.

Identity Resolution and Privacy

Identity resolution increases the importance of privacy governance because it intentionally connects customer interactions.

Organizations should clearly understand:

  • Which identifiers are collected

  • Why they are collected

  • Where they are stored

  • Which systems receive them

  • How long they are retained

  • What permissions apply

The principle of data minimization remains important.

Just because an organization can connect more information does not mean it should.

A Data Privacy Compliance Audit can help evaluate whether identity-related data flows align with the organization's broader privacy and governance requirements.

Hashing Does Not Automatically Mean Anonymous

A common misconception is that hashing an identifier removes all privacy considerations.

Hashing transforms information.

It does not necessarily make it anonymous.

If an identifier can still be consistently matched between systems, it may continue to function as an identity signal.

Organizations should therefore distinguish between:

  • Encryption

  • Hashing

  • Pseudonymization

  • Anonymization

These concepts serve different purposes.

Technical transformation should not be confused with the complete elimination of identity.

Build an Identity Resolution Framework Before Buying Technology

Organizations sometimes begin by evaluating:

  • CDPs

  • Identity vendors

  • Data clean rooms

  • Customer-data platforms

before defining what identity problem they actually need to solve.

The process should begin with the business requirements.

Step 1: Define the Use Cases

Examples:

  • Cross-device analytics

  • Customer Lifetime Value

  • Audience suppression

  • Personalization

  • Omnichannel measurement

Step 2: Inventory Identifiers

Document identifiers across:

  • CRM

  • Website

  • Mobile app

  • Ecommerce

  • Loyalty

  • Marketing platforms

Step 3: Define a Durable Customer Key

Determine the organization's primary internal customer identifier.

Step 4: Establish Identity Relationships

Document how other identifiers relate to the customer.

Step 5: Define Matching Rules

Determine what evidence is strong enough to merge identities.

Step 6: Define Privacy and Consent Rules

Establish permitted uses and destinations.

Step 7: Build the Identity Map

Implement relationships across the relevant systems.

Step 8: Validate the Data

Test for:

  • Duplicate profiles

  • False stitching

  • Missing relationships

  • Invalid identifiers

Step 9: Activate Carefully

Use identity for approved:

  • Analytics

  • Segmentation

  • Personalization

  • Marketing activation

Step 10: Monitor Continuously

Identity graphs change as customers:

  • Create accounts

  • Change emails

  • Merge profiles

  • Use new devices

  • Delete accounts

Identity governance is an ongoing process.

Common Identity Resolution Mistakes

Treating Device Identity as Person Identity

A browser or device may be shared.

Using Email as the Only Customer Key

Contact information can change.

Merging Records Too Aggressively

False matches can corrupt customer profiles.

Ignoring Logout and Account Switching

Identity state must change correctly.

Failing to Document Source Systems

Teams need to know which system owns each identifier.

Collecting More Identity Data Than Necessary

Data minimization should remain part of the architecture.

Assuming a CDP Automatically Solves Identity

Technology still depends on good identifiers, matching rules, and governance.

Ignoring Match Rates

Not every customer journey can be connected.

Measurement should acknowledge those limitations.

How Identity Resolution Supports Advanced Marketing Analytics

Identity resolution becomes increasingly valuable as organizations adopt more advanced analytical methods.

Customer Lifetime Value

Connects purchases across time.

Predictive Analytics

Creates richer customer histories for modeling.

Audience Segmentation

Allows behavioral and transactional signals to be combined.

Omnichannel Analysis

Connects interactions across digital and offline environments where reliable identifiers exist.

Attribution

Improves the continuity of customer journeys.

Data Clean Rooms

Provides stronger first-party identifiers for approved matching use cases.

Identity is therefore not a standalone analytics capability.

It is infrastructure supporting many other capabilities.

The Goal Is Not a Perfect 360-Degree Customer View

The phrase "360-degree customer view" is often used in marketing technology.

In reality, no organization has perfect visibility into every customer interaction.

Customers:

  • Use anonymous sessions

  • Switch devices

  • Clear browser data

  • Use shared accounts

  • Decline tracking

  • Interact offline

  • Use channels the organization cannot observe

A mature identity strategy acknowledges these limitations.

The goal should be:

Build the most accurate, governed customer view that the available data can responsibly support.

That is more realistic and more useful than pretending every customer journey can be perfectly reconstructed.

Final Thoughts

Identity resolution sits at the center of modern customer analytics.

Without it, the same person can appear as several disconnected users across browsers, devices, applications, and business systems.

That fragmentation affects:

  • Customer counts

  • Attribution

  • Segmentation

  • Personalization

  • Customer Lifetime Value

  • Omnichannel analysis

  • Marketing activation

A strong identity strategy begins with stable first-party identifiers and clearly defined relationships between systems.

It distinguishes:

  • Devices from people

  • Anonymous users from known customers

  • Strong identity signals from weak assumptions

And it applies governance around how customer information is connected and used.

The objective is not to identify everyone.

It is to create enough continuity that marketing and analytics teams can make more accurate decisions about the customers they legitimately know.

As analytics infrastructure becomes more sophisticated, identity resolution becomes one of the foundational capabilities connecting customer behavior with business outcomes.

Build a More Connected Customer Data Foundation

If customer interactions are fragmented across analytics platforms, CRM systems, devices, and transaction databases, identity resolution can help create a more consistent foundation for measurement and activation.

At RBG Analytics, we help organizations connect first-party data, analytics architecture, data engineering, segmentation, and measurement systems so customer insights can be built on more reliable data.

No pressure. Just a quick conversation about your goals and opportunities.

Previous
Previous

Offline Conversion Tracking: How to Connect Leads and Revenue Back to Marketing

Next
Next

Data Layer Architecture: How to Build Scalable Marketing Analytics