類似視聴者の説明

Lookalike audiences help advertisers reach people who resemble a seed group, such as prior purchasers or subscribers. Platforms may use first-party data, interactions, and other platform signals to build a modeled audience, but the exact features and controls are proprietary and can change. The resulting audience is not a list of people who share the same identity, nor proof that they have a particular interest or will convert. Seed quality affects the result: a narrow, outdated, or biased list can shape who is reached. Customer lists may contain personal data, so advertisers need to check applicable privacy obligations, platform terms, consent or other lawful basis, minimization, and deletion processes. Sensitive information should not be used to infer protected traits or target vulnerable people. Performance should be evaluated against a meaningful baseline because a campaign may perform well due to creative, bidding, or seasonality rather than the audience model. Ad platforms also impose limits on audience use and may provide different forms of audience expansion. Advertisers should read current documentation rather than assume a setting behaves the same over time. Lookalike targeting can support discovery beyond existing customers, but it should be treated as probabilistic reach optimization, not a customer identity system or a guarantee of similarity at an individual level. A seed audience can reflect who previously had access to a product rather than everyone who might benefit. Advertisers should audit who is included and avoid sensitive inferences.