As a brand, you have to interact with your customers across many channels. Having a unified, coherent view of each one has become essential for any company that wants to deliver personalized, seamless experiences. Customer Data Platforms (CDPs) such as Segment bring solid answers to that challenge. One of Segment's core functions is identity resolution, which merges each user's full history into a single profile and makes behavior legible across channels and devices. That is exactly what we want from any CDP.
Contents:
- What is identity resolution?
- The challenges of identity resolution
- How Segment simplifies identity resolution
- Rigid and flexible approaches to merge rules
- Guarding against merge errors
- Use case: from anonymous to identified
Let me walk you through it. 😉
1. What is identity resolution?
Identity resolution is the process by which a company attributes and consolidates user behavior across different touchpoints into a single customer profile — web, mobile or any other channel. That unification is essential for personalized customer experiences and for optimizing marketing campaigns. The process is often made complex, though: at Smart Bees we support a lot of clients on CDP implementations, and this step is regularly complicated by intricate client architectures — a multitude of identifiers and scattered data sources in particular.
2. The challenges of identity resolution
The first obstacle many companies hit is the false impression that identity is already resolved because they have identifiers such as email addresses. In practice that approach limits unification to separate segments, creating fragmented, incomplete profiles — which makes life harder for the marketing and technical teams trying to activate that data coherently.
A concrete example: a user browses a site with an anonymous ID (say anon_123), then signs up and gets a user ID (use_123). Without a centralized identification strategy, the past events tied to anon_123 stay isolated and never connect to the full use_123 profile.
So why not the email?
Taking the email as the primary identifier in identity resolution can look logical at first: it is unique, everyone has one, and it already sits in your CRM. Not so fast. The practice raises several serious problems when you are trying to unify customer data effectively. Here is a quick tour of why, at Smart Bees, we do not consider the email a suitable universal identifier:
2.1 People have several email addresses
Plenty of people hold several email addresses for different purposes: one for online shopping, another for work, a third for social networks. If a company uses email as the primary identifier, each address is treated as a distinct user, which creates multiple profiles for the same person. The user's interaction history ends up split across several profiles, and a single, complete customer view becomes very hard to get.
A quick example : Pierre uses pierre@smart-bees.fr to create an account on a site and pierre.smartbees@gmail.com to buy products. Without an identity resolution mechanism, the company ends up with two separate profiles for the same person.
2.2 Limited accuracy and reliability
An email can be changed by the user, which makes it hard to keep identifying the same person over time. Someone may switch address for security reasons, to tidy an inbox, or simply out of preference (“I want to change my email address”). If the platform has no way of tracking that change and tying it to an existing profile, it loses the trail of the user's earlier interactions.
Another example: Pierre uses pierre.smartbees@gmail.com for years, then switches to pierre@smart-bees.fr. If the system cannot connect the two addresses, Pierre's historical data stays siloed, you get two separate journeys, and the 360° view never materializes. Not ideal for getting a return on your CDP.
2.3 Shared or dormant identifiers
An email is not necessarily a unique, permanent identifier. I am seeing this right now with a client. Family members or friends sometimes share one address to manage accounts or subscriptions. They may even change the address to share the account. That distorts data attribution and any understanding of individual journeys, and you quickly hit a dead end.
Example : a family shares smartbees.laruche@gmail.com to manage all their honey orders. That creates one profile covering several people with distinct behaviors, which makes segmentation and personalization much harder for the marketing team.
2.4 Data protection and consent
The last point, and not the least, concerns privacy and compliance with data protection regulation such as GDPR. Users can withdraw consent for their email address to be used for tracking, which makes it hard to attach other interactions to their profile without an alternative identifier.
One last example : a user may decline consent for their email to be used for tracking, or exercise their right to be forgotten. If the email is the primary identifier, that sharply limits the company's options for continuing to follow the user compliantly and effectively.
For all of these reasons, at Smart Bees we would urge you not to lean on the email alone as the primary identifier in an identity resolution strategy. Companies can rely on CDPs like Segment to handle identity resolution through a multi-identifier model. That model uses external identifiers such as user_id, device_id, anonymous_id and other custom ids to unify interactions and build a complete, durable view of users. It gets past the email's limitations and keeps customer data far more consistent.
Right, enough about email — back to identity resolution. 😙
Just remember that Segment uses a canonical identifier model (canonical_id) that links and consolidates various external identifiers (externalIDs) such as user_id, anonymous_id, device_id and many more. That approach gives you a holistic view of each customer.
3. How Segment simplifies identity resolution
Segment simplifies identity resolution through its Identity Graph, which creates and merges user profiles in real time. Identity resolution in Segment can be configured to:
> Work with your existing data: no extra coding required.
> Merge anonymous sessions: group temporary sessions under a parent identity.
> Map user-to-account relationships: ideal for B2B companies.
> Maintain a persistent ID: tie several external identifiers to a single ID.
So what happens technically?
Segment looks for identifiers such as userId, anonymousId and email in incoming events, compares them against existing profiles, and decides to:
- Create a new profile if no matching profile exists.
- Add data to an existing profile if a match is found.
- Merge several existing profiles into one if multiple matches are found.
I go into more detail below.
4. Rigid and flexible approaches to merge rules
Effective identity resolution needs a balance between flexibility and rigidity. Too much flexibility produces unpredictable results and inconsistent profiles. Segment favors a deterministic approach where merges rest on exact matches, which — assuming everything is implemented properly — buys you more precision.
Segment admins can configure custom merge rules, setting priorities between identifiers (user_id, email and so on) and putting limits in place to avoid incorrect merges.
Here is a simplified example of a limit configuration:
If an event exceeds the limit set for an identifier (a unique user_id, say), Segment applies the priority rules to handle the merge.
Practical examples of Segment's merge and unification rules:
Take an e-commerce company using user_id, email and anonymous_id to track its customers' interactions. In that context, admins can configure rules that give user_id priority over the others so the main profile stays constant and reliable.
A concrete case : Pierre starts by browsing the site without logging in, creating an anonymous_id: anon_123. When he registers, a user_id: use_789 is generated. If a new event arrives with a user_id already present in a profile and exceeds the identifier's limit (set to 1), Segment applies the defined priority — user_id first here — to decide whether to merge or create a new profile.
Value limits per identifier type:
To avoid creating inconsistent profiles, you can set limits on how many values are associated with each identifier type. Segment lets admins cap the number of anonymous_id values at 5 per week, for instance.
Scenario : a user accesses a service from several devices. Each logged-out session generates a new anonymous_id. The cap of 5 anonymous_ids a week keeps the profile from filling up with temporary values that would make it unstable. If a sixth identifier appears, Segment applies the priority rules and drops the older anonymous_ids to avoid incorrect merges.
Priority and conflict resolution:
Ranking identifiers matters a great deal when an event exceeds the limit set for an identifier type. A user profile may have one user_id (priority 1) and several email addresses (priority 2). If an incoming event carries a new user_id different from the existing one but the same email address, Segment checks the priority.
Back to Pierre: he has a profile with user_id: user_123 and email: pierre@gmail.com. A new event arrives with user_id: user_456 and the same email. Segment sees that the user_id limit (set to 1) is exceeded. Since user_id ranks above email, Segment creates a new profile for user_456 rather than modifying the existing one.
Customizing the merge rules:
Admins can adjust the unification rules to their company's specific needs. A B2B company might choose to prioritize group_id in order to track the relationships between users and their company accounts.
In a SaaS platform, for instance, an admin decides group_id should take priority in order to link the interactions of employees from the same company. If several users (with different user_ids) share the same group_id, Segment merges the interactions under one account-level profile while keeping distinct individual profiles for each employee.
5. Guarding against merge errors
Merge errors — non-unique anonymous IDs, for example — can damage profile quality. To prevent that, Segment applies a trust-priority algorithm and a list of blocked values (-1, null, anonymous and so on) to exclude problematic ones.
Example of recommended blocked values:
6. Use case: from anonymous to identified
As you will have gathered, one of the great strengths of Segment's identity resolution is its ability to link anonymous sessions to authenticated ones. When a user moves from anonymous browsing to creating an account, all their earlier interactions are consolidated into a single profile.
To continue with our scenarios: a user anonymously visits a product page (anonymous_id: anon_123). Later, they sign up and receive a user_id: use_123. Segment ties anon_123 to use_123, producing a complete profile with the full interaction history.
Cross-device and cross-app identification also lets you follow a user across devices (mobile, desktop). That matters for marketing campaigns — it stops you sending an abandoned-basket notification to a device where the purchase has already gone through.
In closing
To make identity resolution work, you need to:
- Centralize and standardize the data.
- Use robust, varied identifiers.
- Configure limits and priorities that suit your business.
Segment offers a solid framework for handling the complexity of identity resolution, giving you a unified customer view and effective marketing. It is one of the CDP solutions that offer this kind of identity resolution.
If you would like to talk this through, get in touch with us at Smart Bees.
FAQ
What is identity resolution in a Customer Data Platform?
Identity resolution is the process of bringing a single user's different identifiers (email, cookie, mobile ID, CRM ID) together into one customer profile. It lets you recognize that an anonymous visitor on your site is the same person as a known buyer in your CRM, giving you a 360° view of the customer journey.
What is the difference between deterministic and probabilistic matching?
Deterministic matching relies on exact identifiers (email, customer ID): two profiles share the same email, so it is the same person — 100% precision. Probabilistic matching relies on indirect signals (IP, behavior, browser fingerprint): less precise, but it reconciles more profiles. Modern CDPs combine both in an Identity Graph.
How does Segment Unify handle identity resolution?
Segment Unify creates an identity space where each profile is built from traits and events. It automatically links anonymous sessions to known profiles through analytics.identify() calls, letting you reconstruct the whole journey from the first anonymous visit through to loyalty. Merge rules define which identifiers are used to combine profiles.
What is the difference between identity resolution and data unification?
Identity resolution establishes that this is the same person across several touchpoints. Data unification goes further: once identity is resolved, it merges all the behavioral, transactional and contextual data into a single profile. Identity resolution is a prerequisite for unification.
What are the challenges of identity resolution after third-party cookies?
Without third-party cookies, tracing users across sites and devices is harder. The answers include building first-party data strategies (email through forms, account creation), using universal identifiers (RampID, UID 2.0), and leaning on authentication APIs (Google Sign-In, Apple Sign-In) as identity anchors.
What is the business impact of good identity resolution?
Good identity resolution lets you attach 30 to 70% of sessions to a known profile, improves personalization, reduces ad pressure on users who have already converted, and optimizes media budgets by removing duplicates from audiences. It also improves attribution and gives a sharper view of LTV.
