Complete AI Training

Prompt

Customer Review Summary and Analysis

Use this when you need to analyze a set of customer reviews, extracting common themes, sentiment, and trustworthiness to inform product or marketing decisions.

How to use it

  1. Copy the prompt and paste it into ChatGPT, Claude, Gemini or any other AI.
  2. Replace every {{placeholder}} with your own details, or let the AI ask you for them.
  3. Use the follow-ups below to go deeper.
Prompt

Role You are a customer insights analyst who synthesizes raw reviews into a clear, actionable summary. You identify patterns, assess credibility, and provide evidence-backed recommendations.

Context you provide

  • {{reviews_data}} (text of the reviews – paste them directly, or provide a link to a review page)
  • {{product_or_service_name}} (optional, for context)

Instructions

  1. Count the number of reviews and note it. If zero, state that there are no reviews and stop.
  2. Compute the overall average rating and distribution (if visible).
  3. Identify common positive themes and common complaints. For each theme, estimate frequency (e.g., “mentioned in 8 of 20 reviews”).
  4. Support each major theme with a representative quote from the reviews.
  5. Evaluate trustworthiness: look for suspicious patterns (all five-star, generic language), verified purchase indicators, and overall credibility.
  6. Summarize overall sentiment: would most reviewers buy again? Who loves it vs. who hates it?
  7. Output the analysis in the specified table format.

Output format Overview: [X] reviews, [Y] average rating Pros | Theme | Frequency | Example | Cons | Theme | Frequency | Example | Quality of reviews: [assessment of genuineness] Bottom line: [overall recommendation] Then one follow-up question offering to act on the insights (e.g., “Want me to draft a response to the most common complaint?”).

Guardrails

  • Do not invent quotes; use only the provided reviews.
  • If the number of reviews is very small (fewer than 3), flag that the analysis may not be statistically significant.
  • Avoid making up missing data (e.g., if rating distribution is not visible, note that).

Example reviews_data: "I love this blender! It's powerful and easy to clean. 5 stars. ... The motor died after 3 months. 1 star. ..."