What is Identity Resolution?
Identity resolution is the process of connecting fragmented data points across devices, channels, and touchpoints to build a single, unified profile of an individual customer. When a shopper browses a product on mobile, clicks a retargeting ad on desktop, and completes a purchase in-store, those three interactions look like three separate users in most analytics systems. Identity resolution links them to one person.
For marketers, the practical value is significant. Without it, frequency capping fails, attribution models undercount conversions, and personalization breaks down at channel boundaries. With it, brands can accurately measure reach, suppress converted customers from acquisition campaigns, and deliver coherent experiences across touchpoints.
How Identity Resolution Works
The process relies on matching identifiers. Some are deterministic, meaning exact. Others are probabilistic, meaning inferred. Most enterprise identity systems use both in combination.
Deterministic Matching
Deterministic matching links profiles using verified, unique identifiers such as email addresses, phone numbers, loyalty IDs, or login credentials. When a user signs into the same app on both a phone and a laptop, the system knows with near-certainty these devices belong to the same person.
Retailers with large authenticated audiences hold a structural advantage here. Amazon links purchases, browsing, Alexa usage, and Prime Video activity through a single login across hundreds of millions of accounts. This gives the company a match rate accuracy that anonymous web tracking cannot replicate.
Probabilistic Matching
Probabilistic matching infers connections using behavioral and contextual signals: shared IP addresses, device fingerprints, browser configurations, location patterns, and browsing behavior. If two devices consistently connect from the same home network and browse similar categories at similar times, a matching algorithm may assign a high probability that they belong to one household or individual.
Commercial systems express match accuracy as a confidence score, typically between 0 and 1. Most commercial systems treat a score above 0.85 as a reliable match, though thresholds vary by platform and risk tolerance.
The Identity Graph
The output of identity resolution is an identity graph: a persistent database mapping known identifiers (emails, cookies, mobile advertising IDs, hashed phone numbers) to anonymous signals (device IDs, IP addresses, browser fingerprints). Each node represents an identifier; edges represent confirmed or probabilistic links between them.
LiveRamp, a data connectivity platform, operates one of the largest commercial identity graphs in the US, covering an estimated 250 million individuals. Its Authenticated Traffic Solution translates publisher first-party data into RampIDs, a pseudonymous identifier that travels across programmatic supply chains without exposing raw personal data.
Key Metrics and Calculations
| Metric | Definition | Formula |
|---|---|---|
| Match Rate | Percentage of anonymous IDs resolved to known profiles | Matched records / Total records × 100 |
| Resolution Accuracy | Percentage of matches confirmed as correct | True matches / Total matched records × 100 |
| Identity Coverage | Percentage of addressable audience with resolved profiles | Resolved profiles / Total unique visitors × 100 |
| Cross-Device Reach | Unique individuals reached across all devices | Total device impressions / Average devices per person |
A typical e-commerce brand with 2 million monthly visitors might achieve a 30 to 45 percent identity resolution rate without a login wall, rising to 60 to 75 percent when email capture programs are active. Each percentage point of improvement directly reduces wasted ad spend on already-converted customers.
Use Cases in Marketing
Cross-Channel Attribution
Identity resolution enables marketers to credit the correct touchpoints for a conversion. Without it, a customer who sees a display ad on mobile, reads a brand email on desktop, and converts via paid search appears in attribution reports as three separate users from three separate channels. Unified identity collapses that path into one journey, making multi-touch attribution models significantly more accurate.
Audience Suppression
Suppression lists, which exclude existing customers from acquisition campaigns, depend on identity resolution working correctly. A mismatch between a CRM email record and a device ID means a paid customer continues to see “new customer” promotions, which wastes budget and can erode trust. Procter & Gamble reportedly saved tens of millions of dollars in wasted programmatic spend after restructuring its identity and suppression practices between 2017 and 2019.
Frequency Management
Without resolved identity, a user who browses on three devices could receive the same ad nine times before the advertiser’s frequency cap of three triggers. Resolved identity applies the cap at the person level rather than the device level, improving user experience and reducing cost per impression for the brand.
Personalization
A resolved profile carries browsing history, purchase history, and channel preferences across sessions. This allows a brand to continue a personalized experience from a mobile app into a web browser or a physical store, rather than starting fresh each time. Nordstrom has built much of its omnichannel strategy around first-party data identity, linking loyalty program activity to email, app, and in-store behavior to drive relevant recommendations.
Privacy Constraints and the Cookieless Shift
Third-party cookies historically powered much of the probabilistic matching infrastructure in digital advertising. As browsers phase out cookie support and regulators enforce stricter consent requirements under GDPR and CCPA, identity resolution has shifted toward authenticated, consent-based approaches.
The major responses from the industry include:
- Universal IDs: Pseudonymous identifiers built from hashed emails, shared across publishers and advertisers with user consent. Examples include Unified ID 2.0 (developed by The Trade Desk) and ID5.
- Data clean rooms: Secure environments where advertisers and publishers match first-party data without either party exposing raw records. Google’s Ads Data Hub and Amazon Marketing Cloud operate on this model.
- On-device matching: Processing identity signals locally on the user’s device rather than in centralized servers, preserving privacy while still enabling personalization.
Brands investing in customer data platform infrastructure tend to hold an advantage in this environment, as CDPs aggregate first-party identifiers that remain usable regardless of third-party cookie availability.
Deterministic vs. Probabilistic: When to Use Each
| Factor | Deterministic | Probabilistic |
|---|---|---|
| Accuracy | Very high (95%+) | Moderate (70–85%) |
| Scale | Limited to authenticated users | Broad, covers anonymous users |
| Privacy risk | Higher (tied to PII) | Lower (no direct PII) |
| Best for | CRM matching, suppression, loyalty | Prospecting, reach extension |
Frequently Asked Questions
What is identity resolution in marketing?
Identity resolution in marketing is the process of linking data from multiple devices, channels, and sessions to build a single unified customer profile. It allows brands to recognize the same person whether they browse on mobile, respond to an email on desktop, or purchase in-store, enabling accurate attribution, personalization, and audience management across every touchpoint.
What is the difference between deterministic and probabilistic identity resolution?
Deterministic identity resolution uses verified identifiers like email addresses or login credentials to match profiles with near-certain accuracy (95%+). Probabilistic identity resolution infers connections from behavioral signals like shared IP addresses, device fingerprints, and browsing patterns, achieving broader reach across anonymous users at lower accuracy (70–85%). Most enterprise systems use both methods together.
What is an identity graph?
An identity graph is a persistent database that maps known identifiers, such as emails, cookies, and mobile advertising IDs, to anonymous signals like device IDs and browser fingerprints. It is the core output of an identity resolution system, allowing marketers to recognize the same individual across sessions, devices, and channels over time.
How does identity resolution work without third-party cookies?
Without third-party cookies, identity resolution relies on authenticated first-party data: email sign-ups, loyalty logins, and universal IDs built from hashed email addresses. Tools like Unified ID 2.0 and data clean rooms allow advertisers and publishers to match audiences with user consent, without relying on browser-level tracking.
What is a good identity resolution match rate?
A match rate of 30 to 45 percent of monthly visitors is typical for an e-commerce brand without a login wall. With active email capture programs, that rate can rise to 60 to 75 percent. Each percentage point of improvement directly reduces wasted ad spend on already-converted customers.