---
name: "voice-of-customer-and-feedback"
description: "Use when tickets pile into patterns: mine the batch for repeat themes, file them as sourced evidence, and draft the product-ready brief."
triggers: ["customer feedback themes", "repeat issues in tickets", "pull customer quotes", "product-ready feedback brief", "feature request from a customer", "top customer complaints", "survey responses"]
version: "1"
---
# Voice of customer and feedback
Use this when tickets pile into patterns: mine the batch for repeat
themes, file them as sourced evidence, and draft the product-ready
brief.
You need the ticket or thread batch with dates, survey or review quotes
when they exist, and the product or engineering intake path.
## Mine the batch
Read every thread, pull the exact customer words with speaker and
date, and group them into themes. Each theme carries its count, two
verbatim quotes, the origin channel per quote, and the ticket IDs
behind it. Complaints under-represent quiet unhappiness, so say where a
theme is likely under-counted — but never scale a count by an assumed
ratio, and never rank themes on an estimate you cannot source. Three
mentions is a hint, ten is a pattern; say which each theme is.
## File each theme as sourced evidence
State the claim in the customer's words, who said it and when, and
what it blocks. Missing-feature asks get the offer to file a defect or
feature issue; feedback with no breakage gets a thank-you draft, never
a roadmap promise.
Draft the product brief: themes ranked by count and pain, the top
knowledge-base (KB) candidate, and the one ask per theme with its evidence
attached. Show it in chat, take one round of edits, then stage it for the
owner.
Log the themes against the intake with dates so next week's batch
dedupes against them.
Deliver the ranked theme brief plus staged issue and thank-you drafts.
Offer to file issues when connected, as drafts they approve first.
## What not to do
A thin batch means fewer themes with honest counts, never padded ones.
Say which extra threads would settle each INFERENCE.