Start with real signals.
Use analytics, email behavior, reviews, surveys, sales calls, and support questions. The persona should be built from evidence, not invented demographics.
Portfolio archive / Built projects
Book · AI audience strategy
The Audience Machine
A book and workbook that turn audience research into a practical AI-persona testing system for content, email, and offers.
AI personasAudience researchPrompt designContent testingLead magnets
A useful preview, not a vague teaser
Five moves from the AI persona system.
Map behavior, not a stereotype.
Capture motivations, objections, decision criteria, emotional triggers, preferred language, and the moments that make someone act.
Give the AI one job.
Define the audience, context, material being reviewed, and exact output. A specific simulation produces more useful criticism than “What do you think?”
Use it as a skeptical reader.
Ask what feels unclear, generic, untrustworthy, irrelevant, or incomplete—and what single change would most affect the decision.
Validate against reality.
Compare the simulated response with opens, clicks, replies, interviews, conversion data, and A/B tests. Update the persona when the evidence disagrees.
Starter prompt
You are simulating a prospective [audience] based only on the evidence below. Review this [email / page / post]. Tell me: what you think it offers, what feels relevant, what creates doubt, what question is unanswered, and the single change most likely to make you act. Separate evidence from assumptions.
AI personas are a research and testing tool—not a substitute for real customers, interviews, or analytics.
Want the complete system?