As of 13 August 2026, AI can only partly find repeated complaints about a product.

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 1 hour until you can act on the result.

   How to actually do it

   1. Open your helpdesk, review platform and other complaint channels, then export the relevant records with their text, dates, source and record ID.
   2. Remove passwords, payment details, health information and other unnecessary personal data while keeping the wording, date, channel and an anonymous record ID.
   3. Combine the cleaned exports into one document or spreadsheet, add the product name and label each record with its source.
   4. Paste the product name, complaint data and any named public links into the prompt, then ask the chatbot to analyse only the supplied material.
   5. Open the source records cited for each theme and compare the quoted wording, count, dates and duplicate flags with the original records.
   6. Give the verified themes to the customer-service or product owner, and record which issues need a separate investigation before changing the product or process.

   Prompt

   Analyse the complaint data below for repeated complaints about [PRODUCT]. Group complaints into distinct issue themes, combining different wording that describes the same underlying problem. Do not invent complaints, counts, causes or customer impact. For each theme, provide: the theme name, the number of records supporting it, the percentage of all supplied records if the total allows an exact calculation, the record IDs or source references, two short verbatim excerpts, the earliest and latest dates supplied, and a plain description of what customers are reporting. Separate genuine repeated themes from one-off complaints, vague dissatisfaction and complaints that cannot be classified. Flag possible duplicates and records that could belong to more than one theme. Rank themes by record count, then by recency only where the dates support that comparison. State clearly that the findings cover only the material supplied and cannot prove that no complaints exist elsewhere. If web access is available, analyse only the named public sources below and cite each source; do not claim to have searched sources you could not access.
   
   Product: [PRODUCT]
   Complaint data or review text:
   [PASTE DATA HERE]
   
   Named public sources, if any:
   [PASTE LINKS HERE]
   
   Return a table followed by a short limitations section and three practical questions for a customer-service manager to investigate.

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