Website Identity, Account Identity, and Household Identity Answer Different Questions

Website identity can add context to anonymous traffic, but the level of resolution matters. A company match can suggest which organization is engaging. A household match can connect activity to a household-level record. A person-level match makes a different and more specific claim. These should never be treated as interchangeable.

For a buyer evaluating Cortex Website Identity, the first question is not simply how many visitors can be matched. It is what each match means and which business action that meaning can support.

Begin with the unit you actually need

A B2B team planning account research may find company-level context useful. A local business evaluating serviceable demand may need a different location or customer relationship context. Neither automatically requires identifying a specific individual.

Define the decision before choosing a matching workflow. If the goal is to prioritize accounts for research, a household inference may not answer the question. If the goal is to understand aggregate audience overlap, exporting named contacts may be unnecessary.

Use the least specific information that supports the agreed purpose. This can simplify both analysis and operations.

A match is an inference with conditions

Matching methods can rely on different identifiers and evidence. Shared devices, corporate networks, household relationships, remote work, and changing identifiers can affect interpretation. Ask which level is supported and how uncertainty is represented.

Do not turn a company match into the claim that a particular executive read a page. Do not assume that one household member’s activity describes every member. The sales message should remain appropriate even if the inferred context is incomplete.

A good handoff might identify a relevant business topic to discuss. It should not encourage a salesperson to recite supposed private browsing behavior as a certainty.

Measure coverage and correctness separately

Coverage asks how much eligible traffic receives a match. Correctness asks whether those matches are accurate under a defined validation method. A high match rate does not prove high accuracy, and a precise match on a small subset does not establish broad coverage.

Ask for the denominator in every claim. Is it all visits, unique browsers, eligible visitors, or records with certain identifiers? Ask how known bots, internal traffic, and existing customers are handled.

Also ask what validation data supports accuracy. Marketing counters without documented methodology are not a substitute for an inspectable evaluation.

A hypothetical account handoff

An engineering supplier observes engagement associated with a target company and a relevant service page. The useful next step may be to research the account’s public needs and prepare a technical discussion through an appropriate existing relationship.

The observation does not establish which employee visited, whether a funded project exists, or whether outreach is permitted through every channel. Sales still needs to confirm need, role, timing, and fit.

If the account later becomes qualified, the outcome can help evaluate the usefulness of that signal. It should not be treated as proof that the original identity inference was correct in every detail.

Matching is not consent

Technical ability to connect records does not create permission to use them for every purpose. Review collection context, agreements, applicable requirements, and the rules of each activation destination. Keep those checks separate from technical matching success.

For example, Google’s Customer Match policy, reviewed October 12, 2026, restricts uploads to customer information collected in a first-party context and includes additional requirements. An identity-enriched record should not be assumed eligible merely because it contains a matchable field.

Confirm the specific workflow with the platform and your responsible reviewers before activation. This article does not establish eligibility for a particular account or dataset.

Connect identity to useful learning

Cortex’s two-way intent approach can use suitable website behavior alongside customer outcomes to refine a customer-specific strategy. Identity can help connect the pieces, but it should not erase uncertainty or replace qualification.

Preserve resolution level, event timing, and customer status in the analysis. If a match is later corrected, the feedback process should be able to reconcile that correction.

Book a Cortex Website Identity consultation. Ask for a sample explanation of the match level, validation method, permitted uses, and the next action each type of signal can reasonably support.