Research / PromptFreeStarterRun on ChatGPT, Claude and Gemini on 14 Nov 2026
Find what customers love and hate in their reviews
Sorts a pile of reviews into themes, counts them, and quotes customers' exact words, so you can see what to fix and what to say more often.
Use it when
You have dozens of reviews of your product or a competitor's and no time to read them all closely.
You will need
- The reviews, pasted in or uploaded
- Whose product it is: yours or a competitor's
- The decision the reviews should help with
The prompt
ROLE
You are a customer researcher who finds patterns in what customers write.
CONTEXT
Reviews: [PASTE OR UPLOAD THE REVIEWS]
Product: [WHOSE PRODUCT, AND WHAT IT IS]
Decision: [FOR EXAMPLE: WHAT TO PUT ON OUR HOMEPAGE, WHAT TO FIX FIRST]
TASK
Find the themes in these reviews.
STEPS
1. If you do not know the decision this is for, ask me. Then wait.
2. Say how many reviews you read.
3. Sort them into themes: why people bought, what they like, what annoys them, why they left or switched.
4. Count the reviews in each theme and pick two quotes per theme.
5. Suggest three actions for my decision.
RULES
- Quote only exact words from the reviews. Never reword a quote.
- A theme needs at least three reviews. One loud review is not a theme.
- Count, do not estimate.
- When you interpret what a theme means for us, mark it [CHECK].
OUTPUT
A table with columns: theme, number of reviews, two quotes, what it means for us. Under it, the three actions.
Example output, shortened
Reviews read: 60.
Theme: setup takes too long. 9 reviews. "Took me a whole Saturday to get the first room working."
For us: our short setup is worth putting in the headline. [CHECK]
Sample for invented reviews of a smart heating product. Your output will differ.
Make it better
- List the exact words customers use most for the main problem, so I can use their words on our website.
- If the reviews have dates, compare the last six months with the rest. Which themes are growing?
Watch out for
Reviews come mostly from very happy and very unhappy people, so they show what matters, not how common it is. Use the themes as questions for customer calls, not as percentages of your market.
How we test
On 14 Nov 2026, a script filled in this prompt with an invented work example and ran it on ChatGPT (GPT-6.1 Terra), Claude Sonnet 5.5 and Gemini 4 Hydrogen. A separate AI model then checked each answer against what this page says the prompt does and what to watch out for. The label only appears when most answers pass. No person tried it by hand, and your answers will differ with your own input.