Psychological Targeting

Based on Sandra Matz's public method. Not affiliated with or endorsed by Sandra Matz.

When to use

Use this skill to design or review a personalized invitation, recommendation, campaign, or behavior support intervention when psychological needs could change what people receive. It also applies when someone proposes personality inference from digital traces or claims that tailored messages improve decisions. Start with a concrete audience and goal, rather than a dataset looking for a use. Matz distinguishes gaining psychological insight from using it to influence choices, and describes both helpful and exploitative applications. [S1] [S2]

Produce a psychological personalization decision brief, including usable variants or recommendations and a plan to evaluate them. For an audit, describe the existing system in the same fields and identify necessary corrections. The package translates public research into an operational workflow; its controls are implementation recommendations, not a questionnaire, clinical diagnostic method, legal assessment, or verbatim protocol published by Matz. Source attribution and evidence limits live in references/method.md.

Procedure

1. Set the decision boundary. Read method rule M1 in references/method.md. Identify the recipient's stated goal, sponsor objective, intended behavior, context, and decision owner. State conflicts and missing information. Do not equate a sponsor's sales target with a recipient benefit. Use the brief fields in templates/output-template.md.

2. Separate insight from influence. Follow M2 in references/method.md. Write what the system would estimate and what it would change. Require separate evidence for signal validity and intervention effectiveness; accurate prediction alone establishes neither permission nor benefit.

3. Screen the proposed use. Apply M3 in references/method.md. Redirect designs that exploit vulnerability, discriminate, or covertly steer critical civic choices. Offer a transparent route using volunteered preferences, consistent facts, and recipient control. Document the reason rather than drafting an exploitative variant.

4. Inventory evidence and permission. Use M4 in references/method.md and the signal table in templates/output-template.md. Distinguish deliberate self-presentation from incidental traces. Record provenance, purpose, permission, retention, and expected use; do not interpret public availability as consent to profiling.