As of 13 August 2026, AI can find negative themes in customer reviews.

What to actually do

1. Hand it to a person

   The route this page recommends

   A person who owns the outcome does this end to end, worth it when the failure is dear.

2. Use a tool built for this

   Second choice

3. Do it yourself

   The distant third

   A chat interface, chat-fluent skill, and roughly 30 minutes until you can act on the result.

   How to actually do it

   1. Export the reviews from your review platform into a document or spreadsheet, keeping the review text, rating, date, source and a unique review ID or row number.
   2. Remove passwords, payment details and unnecessary personal information, then split a very large export into manageable batches while preserving the original IDs.
   3. Open a chatbot, paste the prompt, replace the bracketed context fields, and paste the reviews after the final heading.
   4. Ask the model to analyse each batch using the same prompt, then ask it to combine the batch results without merging themes unless the quoted evidence supports the merge.
   5. Open the original review export and search each quoted example and review ID to confirm that every quotation is exact and that the review supports its assigned theme.
   6. Create a simple table with each theme, supporting review IDs, review count where reliable, rating pattern and any contradictory examples, then correct the model's labels where the underlying complaint differs.
   7. Send the checked table to the colleague responsible for customer insight or product decisions and agree which themes merit further investigation before changing campaigns or messaging.

   Prompt

   Analyse the customer reviews below and find the recurring negative themes.
   
   Context:
   - Product or service: [PRODUCT OR SERVICE]
   - Review period: [DATE RANGE]
   - Market or customer segment: [MARKET OR SEGMENT]
   - Rating scale and any known review-source limits: [DETAILS]
   
   Instructions:
   1. Identify complaints, frustrations and negative experiences only. Separate a clear complaint from a neutral observation or a positive comment.
   2. Group reviews into distinct themes based on the underlying customer problem, not just shared words.
   3. For each theme, give a short label, a plain-English description, the number of reviews that support it if that can be counted reliably, and three representative quotations copied exactly from the reviews. Do not invent or rewrite quotations.
   4. Quote the review IDs or row numbers for every example. If IDs are absent, use the order in which the reviews appear.
   5. Distinguish direct evidence from your interpretation. Flag themes that may be caused by the same underlying issue.
   6. Note contradictory or positive evidence that weakens each theme.
   7. Do not infer customer characteristics, motives or causes that the reviews do not state.
   8. End with a table of the themes ranked by evidence strength, followed by the five review IDs or row numbers I should read first.
   9. State any limitations caused by missing context, duplicate reviews, unclear wording, small samples or possible selection bias.
   
   Reviews:
   [PASTE REVIEWS HERE]

   Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.