Build an Evidence-Based Customer Persona from Interview Notes or Reviews

Turns your pasted interviews, reviews or survey answers into a customer persona where every trait cites a quote or data point, with untested guesses listed separately.

You getPersona Snapshot, evidence Table with these columns and top 3 Messaging Angles

Prompt

Prompt
Act as a customer research analyst who builds personas only from evidence.

I will paste customer interview notes, reviews, support tickets or survey answers right after this message. Build one customer persona from that material.

Rules:
1. Every trait, goal, pain point and objection must cite its source. Use either a short direct quote from my material in quotation marks or a specific data point, such as "4 of 11 reviews mention setup time."
2. If a trait seems likely but nothing I pasted supports it, leave it out of the persona. List it under Hypotheses to Test instead.
3. Use the customers' own words for pains and goals. Do not rewrite them into marketing language.
4. Count how many sources support each trait. Mark any trait backed by only one source as "weak."
5. Do not invent names, ages, job titles, income or demographics that are not in the material.

Output format:
1. Persona Snapshot: 3 sentences, using evidence only.
2. Evidence Table with these columns: Trait | Category (goal, pain, behavior, objection, trigger) | Evidence | Source count
3. Top 3 Messaging Angles: link each one to traits in the table.
4. Hypotheses to Test with these columns: Guess | Why it seems plausible | How to test it

End with one line on gaps in the material and which customer types are missing from it.

Example output

At a glancePersona traits by number of supporting sourcesFrom 5 interview notes and 14 app-store reviews
  • No-shows are the biggest cost: 10
  • Books clients by text between grooms: 5
  • Long setup is a dealbreaker: 5
  • Wants deposits without awkward asking: 4
  • Worries older clients won't use links: 3
  • Started searching after a bad week: 1 (weak)

Example for: a booking app for independent dog groomers, built from 5 interview notes and 14 app-store reviews.

1. Persona Snapshot: "The Solo Groomer Buried in Messages"

Runs a one-person grooming business and handles bookings by text and phone between appointments (5 of 5 interviews). Looked for an app mainly to stop no-shows, which cost them money every week (4 interviews, 6 reviews). Wary of anything that takes long to set up or confuses older clients (3 interviews, 5 reviews).

2. Evidence Table

TraitCategoryEvidenceSource count
Books clients by text between groomsBehavior"I'm answering texts with wet hands all day" (Interview 2)5
No-shows are the biggest costPain"Two no-shows on a Saturday is my whole profit" (Interview 4)10
Wants deposits without awkward askingGoal"I hate chasing people for a deposit" (Review 7)4
Worries older clients won't use linksObjection"Half my regulars are retirees, they just call" (Interview 1)3
Long setup is a dealbreakerObjection5 of 14 reviews say setup took "an evening" or longer5
Started searching after a bad weekTrigger"After three no-shows in one week I downloaded four apps" (Review 11)1 (weak)

3. Messaging Angles

  1. "Stop losing Saturdays to no-shows." This draws on the no-show pain, the strongest trait with 10 sources. It uses the customers' own Saturday framing.
  2. "Deposits collected for you, no awkward texts." This links to the deposit goal (4 sources) and to texting between grooms.
  3. "Your regulars can still just call." This answers the older-client objection (3 sources). It could pair with a claim about setup time if you can back one up.

4. Hypotheses to Test

GuessWhy it seems plausibleHow to test it
Many users are mobile groomers working from vansTwo interviews mention "the van," but nobody was asked directlyAdd one question to the onboarding survey
Would pay more for SMS remindersReminders are praised in 3 reviews, but price never comes upShow a test pricing page with an SMS tier to half of visitors
Prefers setup on a phone over a laptopReviews describe setup "on my phone at night," though the evidence is thinCheck device data for completed setups

Gaps: Every source is a current user. No groomers who churned or chose a competitor are included, so objections are probably undercounted.

Written by AI from this prompt. Your results will vary.

Tips and common edits

Why this works

  • Requiring a quote or count for every trait stops the model from inventing generic persona filler like age, hobbies and income.
  • Moving unsupported guesses into a Hypotheses to Test table keeps useful hunches without passing them off as findings.
  • Source counts and the "weak" label show which traits are solid enough to build messaging on.

Use this before writing ads, landing pages or emails. Have at least 5 interviews or 10 reviews ready to paste. It works in ChatGPT, Claude or Gemini. Claude handles long pasted transcripts well.

Common edits

  • Ask for two or three personas if your material shows clearly different customer groups.
  • Add a column for which product feature or page each trait should shape.

Prompt details

By
Prompts-GPT editors
Published

Did this prompt work?

Keep exploring

More in Marketing