Using AI to build and test an email campaign persona

Executive overview

Most email optimisation is guesswork — too little data, no clear signal on what to change. ChatGPT can stand in as a synthetic customer persona and review every email in a sequence before it reaches a real audience.

Prompt ChatGPT to generate a persona, feed it your emails one by one, then use that same persona to simulate different buyer behaviours and A/B test variants.

Build a persona first, then let it judge your copy — before any real subscriber sees it.

Building the persona and getting first impressions

• Prompt ChatGPT with the target product URL and role (e.g. Freelance Copywriter) to generate a detailed customer persona.
• Give the persona a name — the tutorial uses "Alex" — to anchor subsequent prompts.
• Share the first email and ask for Alex's impressions and expectations given the action that triggered the sequence.
• Repeat for each email in the sequence to get first-impression feedback throughout.

Simulating behaviour mid-sequence

• Use a structured prompt with three variables: context (user's current mindset), delay trigger (time or event before email sends), and desired action (goal of the sequence).
• Set context to reflect hesitation — e.g. "you're not ready to sign up for a paid plan" — to test how emails land on a resistant reader.
• Run every email through the simulated behaviour, not just the first touch.

A/B testing variants

• Submit Version A, then Version B to ChatGPT in the same conversation.
• Test any and all variants before choosing which to send live.
• Treat results as directional feedback, not ground truth — useful signal, but not a replacement for real data.