As of 13 August 2026, AI can find product improvements from customer feedback.

What to actually do

1. Use a tool built for this

   The route this page recommends

2. Do it yourself

   Second choice

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

   How to actually do it

   1. Open the source containing the feedback, such as your survey export, support inbox export or NPS comments, and remove passwords, payment details, contact details and other unnecessary personal data.
   2. Gather the product description, customer groups, current goals, known constraints and any existing issue or roadmap categories that should be used in the analysis.
   3. Paste the context and the feedback into the prompt, splitting a very large dataset into labelled batches and asking the chatbot to preserve the same theme names across batches.
   4. Ask the chatbot to produce the themes, linked evidence, possible improvements, evidence strength and proposed priority order specified in the prompt.
   5. Copy each proposed theme and improvement into a working document, then compare its supporting quotes with the original feedback and remove any claim that is not supported.
   6. Ask a product colleague or subject owner to check the remaining priorities against customer importance, strategy, feasibility and known constraints before turning any recommendation into a roadmap item.

   Prompt

   Analyse the customer feedback below and identify product improvements.
   
   Context:
   - Product: [describe the product]
   - Main customers: [describe the customer groups]
   - Current product goals: [list the relevant goals]
   - Constraints: [list technical, legal, budget or delivery constraints]
   - Feedback source and period: [describe the source and period]
   
   Feedback:
   [PASTE FEEDBACK HERE]
   
   Return:
   1. The main feedback themes, ranked by the number and strength of supporting comments. Do not claim a theme is common unless the pasted feedback supports that claim.
   2. For each theme, quote or reference the relevant feedback without exposing personal data unnecessarily.
   3. A list of possible product improvements, each linked to the feedback theme it addresses.
   4. For each improvement, state the customer problem, likely benefit, risks, assumptions and an evidence strength of high, medium or low.
   5. Separate requests from underlying problems, and separate product issues from complaints about support, pricing or policy.
   6. A proposed priority order using customer impact, frequency, urgency, strategic fit and implementation uncertainty. Do not estimate revenue, savings or effort unless the supplied information supports it.
   7. Five follow-up questions or research actions that would test the most uncertain recommendations.
   
   Do not invent feedback, customer numbers, causes, competitors, market facts or technical details. State clearly where the evidence is mixed or insufficient. Treat the output as a set of hypotheses for a product team to check, not as an automatic roadmap.

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

3. Hand it to a person

   The distant third

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