promptsforbusiness
Free AI prompt · ChatGPT · Claude · Gemini

Persona signals from customer reviews

Paste reviews (yours or a competitor's) and extract who is buying, what triggered the purchase, what they value and what disappointed them, with every insight tied to a quote.

Task: building a customer persona from reviewsBest in: ClaudeFormat: single promptPack: Customer persona builder

What you'll fill in

  • [business description] What you sell and to whom, in a sentence or two e.g. “Lumen Yoga Studio: a neighborhood yoga studio with drop-in classes and monthly memberships”
  • [reviews] 20-100 reviews pasted as text (Google, Yelp, Amazon, G2, app store, competitor reviews). Include star ratings if you have them. e.g. “5 stars: "I was nervous as a total beginner but the 7am class..." 3 stars: "Great teachers, but parking is a nightmare."”

Copy the prompt

The prompt
You are a customer research analyst. I want to understand who our customers are from what they write in reviews.

Business: [business description]

Reviews:
[reviews]

Evidence rule: ground every claim in the data I pasted. Label each insight as one of: (evidence) when it is directly supported, with a short quote or reference; (inference) when it is a reasonable reading of the data; (assumption) when the data does not cover it. Never invent quotes, demographics, percentages or customer names. If the data is too thin to support a conclusion, say so.

Analyze the reviews and give me:
1. Customer segments: 2-4 distinct groups you can see in the reviews (based on situation, goal or experience level, not invented demographics). For each, quote 2 reviews that show it and estimate how many reviews fit it (count them, do not guess percentages).
2. Triggers: what was happening in their life or business that made them look for this.
3. Desired outcomes: what "success" looks like in their words. Quote phrases verbatim.
4. Delighters: what exceeded expectations.
5. Frustrations and deal-breakers: what disappointed people, especially in 1-3 star reviews.
6. Language bank: 15 exact phrases customers use that would work in marketing copy.
7. What the reviews cannot tell us: important persona questions this data does not answer.

Keep it skimmable. Do not write a polished persona yet; this is raw material.

How to use it

  1. Open a new chat in Claude (the others work too).
  2. Paste the prompt and replace every [bracket] with your details.
  3. Review the output, check any facts, and edit to taste.

Why Claude: Claude and Gemini handle large batches of pasted reviews well and keep quotes accurate.

Tips

  • Competitor reviews work well, especially the 2-4 star ones; they show unmet needs.
  • Paste reviews as plain text; remove reviewer full names if you are sharing the chat.
  • The language bank is reusable for ads, landing pages and email subject lines.

Example output

Segments (from 64 reviews): 1. Nervous beginners (23 reviews) — (evidence) "I was nervous as a total beginner but the 7am class felt welcoming." "Nobody made me feel clumsy." 2. Stressed professionals using yoga to decompress (17 reviews) — (evidence) "My one hour a week without Slack." "Best thing I do for my back after desk days." 3. Experienced practitioners new to the area (9 reviews) — (evidence) "Finally found a studio with real vinyasa after moving here." (15 reviews too short to classify.) Triggers: - (evidence) moved to the neighborhood; back pain from desk work; "my doctor told me to try yoga." - (inference) New Year and post-holiday resets, based on review dates clustering in January. Desired outcomes: "feel less stiff," "an hour that's just mine," "not feel judged." Delighters: teachers learning names (11 mentions); small class sizes. Frustrations: parking (8 mentions, mostly 3-star); 6pm classes full; booking app logouts. Language bank: "total beginner," "nobody made me feel clumsy," "an hour without Slack," "my back thanks me," "real vinyasa," "felt welcoming from day one"... What the reviews can't tell us: why people cancel memberships; price sensitivity; how they first heard of the studio; non-reviewers who tried once and never returned. (assumption) Beginners may be the largest acquisition opportunity, but this needs testing.

FAQ

Does this prompt work in ChatGPT, Claude and Gemini?

Yes. It is plain-language and written to work in all three. Best fit: Claude. Claude and Gemini handle large batches of pasted reviews well and keep quotes accurate.

What do I need to fill in?

Replace each bracket with your own details: [business description], [reviews]. The more specific your input, the more useful the result.

Is this prompt free?

Yes. Copy it above and use it as often as you like. It is one of the free samples from the Customer persona builder pack.

Will I get the same result as the example?

No. The example was made for a fictional business to show the shape and quality of the output. Your result depends on your input and the model, so review and edit before you use it.

Do I need a paid AI plan?

Usually not. Free plans handle most prompts. Paid plans help when you paste in long documents or run a long multi-step sequence in one chat.

More prompts in Customer persona builder

Free2 stepsBest in Claude

Synthesize customer interviews into persona insights

Turn messy interview transcripts or call notes into a structured synthesis: patterns across customers, notable differences, and the quotes that matter, ready to build personas from.

In packBest in Gemini

Pain points and needs from support tickets

Mine support tickets, chat logs or inbound emails for recurring problems, confusion and unmet needs that should shape your persona and your marketing.

In packBest in ChatGPT

Jobs-to-be-done map for your persona

Map the functional, emotional and social jobs your customer is "hiring" you to do, including what they would use instead, so your persona explains why people buy, not just who they are.

In packBest in Claude

Buying journey and objections map

Map how your persona goes from first noticing the problem to buying, with the questions, objections and information they need at each stage.