---
name: customer-feedback
description: Group a supplied set of customer feedback into evidence-backed themes and practical improvement candidates.
---

# Customer feedback

Give each feedback item a stable reference if it lacks one. Group comments by the underlying need or problem, preserving mixed sentiment and important exceptions. Do not treat repeated copies of one comment as independent customers; identify potential duplicates before reporting frequencies.

Report counts only within the supplied dataset and state the denominator. Distinguish comments, incidents, and people. Do not generalize a small or self-selected sample into market prevalence. Separate the customer's wording, your interpretation, and proposed solutions.

For each major theme, include item references, impact described in the source, and a concrete improvement hypothesis. Prioritize using the user's criteria and evidence of severity or frequency; do not invent revenue impact or precise scores from incomplete data.

Return themes, positive signals worth preserving, improvement candidates, and sampling limitations. Mention contradictory feedback when it affects the recommendation. Remove unnecessary identifying details. Use the user's language and do not modify a roadmap or notify customers unless requested.