Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →How do companies track customers across devices? They collect identifiers and interaction events from websites, apps, servers, purchases, support systems and other touchpoints, then apply matching rules to decide which records may belong to the same customer. The resulting links can power an identity graph or a unified profile for analytics and activation. That profile is usually a connected view—not a replacement for the original records—and every inferred link carries privacy, accuracy and governance responsibilities.
What customer identity resolution actually does
Identity resolution is the process of associating records that contain different identifiers, such as an anonymous web ID, an app device ID, an email address, a phone number, an account ID or a purchase record. A resolution service evaluates the identifiers and rules available at the time an event arrives. If the criteria are satisfied, it connects the event to an existing profile key or to other identifiers in a graph.
As an Amazon Associate I earn from qualifying purchases.
For example, a person might browse a site with an anonymous cookie, use a retailer’s mobile app, sign in later, make a purchase and contact support. If the sign-in event supplies a trusted account or email identifier, earlier anonymous activity may be associated with the newly known account. That is a conceptual pathway, not a promise that every anonymous session can be recovered or that every device can be identified.
Free tools Windows power users keep installed
One-click scans. No signup required.
Twilio Segment describes an identity graph spanning web, mobile, server and partner interactions. Its documented identifiers include cookie IDs, device IDs, email addresses and custom external IDs. Other providers use different channels, identifiers and data sources, so the name of a product does not tell you how broadly it can resolve people.
#1 Best Overall
Identity graph versus unified customer profile
An identity graph represents identifiers and the relationships between them: one account ID connected to an email, several device IDs and a sequence of event records. A unified customer profile is a view assembled from those links, often with selected fields such as name, address or preferred contact method.
These terms are related but not interchangeable. A graph can preserve multiple identifiers and their provenance while a profile presents one value for a field. Salesforce Data 360 documentation describes rulesets that link source profiles and reconciliation rules that choose values for fields that cannot hold multiple values. It also says reconciliation does not delete or alter the source profiles or update the connected source systems. The profile is therefore a presentation and activation layer, not automatically a legally authoritative “golden record.”
How matching works: deterministic and probabilistic methods
| Method | How a link is made | Strengths | Risks and limits |
|---|---|---|---|
| Deterministic | An explicit identifier or rule is treated as a known match, such as a shared authenticated account, verified email or another first-party identifier. | Usually easier to explain and audit when the identifier is trustworthy and collected correctly. | Requires the person to use or provide the identifier; duplicate, stale or mis-keyed source data can still create errors. |
| Probabilistic | A model estimates that records belong to the same person from a set of signals. The FTC has described device type, operating system, fonts and IP address as examples of fingerprinting signals. | Can associate activity when no shared login or direct identifier exists. | It is an estimate, not proof. A false join can place one person’s activity in another profile or trigger an inappropriate message. |
Twilio’s educational material presents verified first-party identifiers as the basis of deterministic resolution. The Federal Trade Commission’s 2015 workshop materials contrasted sign-in-based account linking with probabilistic fingerprinting. Some services use both methods. Experian describes cross-channel linking that combines deterministic and probabilistic resolution.
Recommended Free Tools
Probabilistic matching should not be described as inherently accurate, anonymous or privacy-safe. The FTC has warned that some cross-device techniques may be less visible to consumers and harder to control. A responsible implementation records the method and confidence behind a link, limits what low-confidence links may do, and provides a way to reverse or suppress an incorrect association.
Rank #2
- Ideal for Gifting
- Ideal for a bookworm
- Compact for travelling
What cross-device tracking can and cannot infer
Cross-device tracking associates activity from more than one internet-connected device. In a legitimate first-party use, it can provide continuity when a customer starts on a phone and finishes on a laptop, improve measurement of a journey, or prevent a service from treating a signed-in customer as a stranger on every device. In advertising and personalization contexts, the same linking can expose a much broader behavioral history.
The available evidence determines the scope of the link. An authenticated account is a direct bridge. A shared, verified email can be strong evidence when it is unique and current. A device characteristic or network signal is weaker and may be shared by households, change over time or be affected by technical settings. A system should not claim to identify every person across every device.
Salesforce documentation gives a concrete limitation: its real-time matching workflow uses exact or exact-normalized matching and can return no profile when no existing profile satisfies those criteria. Its scheduled resolution process may use different rules. “Real time” therefore does not necessarily mean broader or more accurate matching than a scheduled job.
From incoming events to a profile
- Collect the event and provenance. Record where the event came from, which identifier was supplied, when it was collected and what permission or notice applied.
- Normalize identifiers. Apply documented transformations such as case normalization where permitted. Keep the original value or a reference to it so the transformation can be audited.
- Apply the permitted match rules. Decide whether the event satisfies a deterministic rule, a probabilistic threshold or neither. Do not silently broaden a rule because a profile is missing.
- Create or update the link structure. Add the identifier-to-profile relationship, preserving source records and the method used to establish the link.
- Reconcile profile fields. If several sources provide a single-valued field, apply a declared precedence, recency or verification rule. Retain conflicting source values rather than erasing them.
- Expose only approved outputs. Send the profile or events to analytics, customer relationship systems or marketing destinations according to purpose, consent and access policies.
Twilio Segment documents configurable ID rules, anonymous identity stitching, user-to-account relationships and merge protection. Those controls illustrate why the implementation details matter more than the phrase “single customer view.” Ask which identifiers are accepted, how merges are prevented, and whether an operator can undo a mistaken link.
Timing, freshness and profile semantics
Resolution systems commonly combine scheduled processing with event-driven or real-time processing. Scheduled jobs can reassess larger data sets and apply broader rules, but a profile may lag behind a recent purchase or consent change. Real-time paths can make a decision during an event, yet may restrict matching to a narrower set of exact criteria.
When evaluating a system, document the latency that matters for each use: identity linking, field reconciliation, consent or opt-out propagation, and delivery to an activation destination may each have a different update time. A profile that appears current in a dashboard may still contain stale values in a downstream system.
Also separate these layers:
- Source systems: the systems that originally collected account, transaction, support or digital-event data.
- Link structure: identifiers, relationships, match methods and confidence information.
- Unified view: the query or activation representation assembled from linked sources.
- Reconciliation rules: logic selecting a displayed value when a field cannot contain multiple values.
- Destinations: analytics, customer relationship, service or marketing tools receiving approved data.
Calling the output a “golden record” can mislead stakeholders into believing that every value is verified, complete and authoritative. The safer description is a governed view whose field-level values have an identifiable source and selection rule.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Fix the driver behind crashes, sound loss and screen glitches3Repair Windows errors before they cause bigger problemsHow to compare identity-resolution offerings
| Evaluation axis | Questions to ask | Why it changes the outcome |
|---|---|---|
| Identifier and channel coverage | Which cookie, device, account, email, phone, custom and offline identifiers are supported? Are sources first-party, licensed or enriched? Is resolution individual or household? | A broad channel list is not useful if your organization cannot lawfully collect or consistently populate those identifiers. |
| Match method and controls | Which workflows are deterministic, probabilistic or both? Can you set thresholds, inspect evidence, protect against merges and reverse false joins? | Method and controls determine explainability, false-join exposure and how much a match may influence activation. |
| Timing and freshness | What is scheduled, event-driven or real time? How often are profiles reconciled, and do real-time rules differ from scheduled rules? | Stale identity or consent data can produce an incorrect decision even when the match itself was valid earlier. |
| Profile semantics | Are source values retained? How are conflicts resolved? Does the service write back to systems of record? | You need to know whether the output is a view, a copied record or a change to an authoritative system. |
| Activation and security | Which destinations and APIs are available? Where are credentials stored, and which services can read customer data? | An accurate graph still creates risk if access is exposed too widely. |
| Governance | How are notice, consent, opt-out, sensitive data, retention, access reviews and deletion requests handled? | Cross-device linking expands the scope of data a profile can reveal and share. |
Vendor documentation establishes what a provider says its system does; it does not establish comparative accuracy, suitability or independent validation. Test candidate rules against your own labeled records, edge cases and false-join scenarios, and require an explanation of what happens when no match is found.
Rank #4
Privacy and security safeguards
The FTC’s report announcement of January 23, 2017, recommended truthful disclosure of tracking practices, consumer choice, heightened protection for sensitive information and reasonable data security. It also encouraged affirmative express consent before cross-device tracking on sensitive topics and before collecting and sharing precise geolocation. This is historical US federal staff guidance, not a complete statement of current law in every jurisdiction or legal advice.
A practical governance program should:
- Document the origin, permitted purpose and retention period for every identifier.
- Propagate consent, opt-out and suppression signals to graphs, profiles and downstream destinations, rather than changing only a front-end preference.
- Use data minimization: collect and link only what the stated purpose requires.
- Restrict sensitive attributes and low-confidence links from automated activation until reviewed.
- Keep source provenance, match method, confidence and timestamps for audit and correction.
- Run tests for false joins, duplicate people, shared household devices, recycled identifiers and account changes.
- Limit administrative access, rotate credentials and monitor exports and API use.
Segment’s documentation provides a specific security example: its Profile API is intended for server-side use because exposing an access secret could reveal customer data. The general lesson applies to any identity platform: do not place privileged profile credentials in browser code or other publicly accessible clients.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.A governance decision framework
Before creating a link
Define the business purpose, the permitted data sources, the match evidence required and the consequences of a false join. A link used only for aggregate measurement may warrant a different threshold from one that changes an individual’s communications or account experience.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minutePC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11When a link is created
Store the rule or model version, evidence, confidence, timestamp and source. Make the link reversible. If an identifier later becomes invalid or a customer withdraws permission, the system should be able to suppress or disconnect associated events without destroying the underlying source record.
Best Value
- It can be a gift option
- Comes with secure packaging
- Helpful in various ways
After activation
Audit where the profile traveled, whether downstream systems honored consent, and whether field reconciliation produced a misleading value. Review match-quality samples regularly; a model or identifier that worked last year may behave differently after a product, platform or data-collection change.
What the FTC record says about consumer expectations
In a March 2015 workshop announcement, Jessica Rich, then director of the FTC’s Bureau of Consumer Protection, said: “With the advent of new tracking methods, though, it’s important to ensure that consumers’ privacy remains protected as businesses seek to target them across multiple devices.” The statement is historical, but it captures the central trade-off: more connected data can improve continuity while also making a larger and potentially more sensitive behavioral history available to an organization.
Transparency should explain the categories of identifiers, the purposes of linking, the kinds of partners or destinations involved and the choices available. A vague statement that a company uses “analytics” does not tell a person whether an account identifier is being used to connect activity on several devices.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Common implementation mistakes
- Treating every shared signal as proof. A household network, recycled phone number or shared device can connect different people.
- Assuming real time means universal. Real-time workflows may accept only exact or exact-normalized identifiers.
- Overwriting source data. A reconciled display value should not silently replace the source value or its history.
- Ignoring no-match outcomes. A missing profile is a valid result and should not trigger increasingly aggressive inference without policy approval.
- Sending more data than the destination needs. Activation should be purpose-limited and credentialed.
- Leaving opt-outs in one system. Suppression must reach the graph, profile and downstream tools that could otherwise reactivate the identifier.
Bottom line for financial and data leaders
Customer identity resolution is a set of linking decisions, not a magic database that knows every device owner. Deterministic identifiers can provide an explainable bridge; probabilistic methods can fill gaps but introduce uncertainty and potential false joins. Identity graphs preserve relationships, while unified profiles select a usable view of those relationships without necessarily changing source records. Choose a system by its identifiers, match controls, timing, reconciliation behavior, security and governance—not by a “360-degree” label—and treat cross-device capability as both a measurement tool and a privacy-sensitive practice.
Frequently Asked Questions
Does identity resolution prove that two devices belong to the same person?
No. An authenticated account or other trusted identifier can provide a strong, explainable link, while probabilistic signals only estimate that records are related. Systems should preserve uncertainty and allow incorrect links to be reversed.
Will a unified profile overwrite the original customer records?
Not necessarily. Salesforce documents reconciliation rules that select values for a unified field while leaving source profile data unchanged. Confirm the behavior of the specific implementation before treating the profile as a system of record.
What should a company do when no exact match is found?
Treat no match as a legitimate outcome. Create or retain an unlinked record, follow the documented confidence policy and avoid progressively weaker inference unless the purpose, risk and governance controls justify it.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
