Customer research
Group customer feedback without losing the original wording
Group comments by the problem described, not by keywords, and keep an anonymized quote under every theme. Count customers, not messages, and record the comments that contradict each pattern. A theme without its exceptions overstates what you know.
A text to start with
Here are customer comments permitted for this review, each tagged with a non-identifying participant code: [comments]. Group them by the problem described. For each theme, give an exact non-identifying excerpt or mark any removed identifier as [redacted], count distinct participants and show comments that complicate the pattern. Preserve meaning, label paraphrases separately and do not invent quotations. Note comments appearing under more than one theme.
Example request. Change the details to fit your situation.Feedback theme table
Source of comments: [ ] Date range: [ ] Total comments: [ ] Total distinct customers: [ ] Theme name (the problem described): [ ] Exact non-identifying excerpt or marked redaction: [ ] Non-identifying participant codes behind this theme: [ ] Distinct customer count: [ ] Comment that contradicts or complicates it: [ ] Open question: [ ] Comments that fit no theme: [ ] Comments placed in more than one theme: [ ]
Counts that mean something
Check counts against the non-identifying participant codes so repeated messages do not inflate a theme. Compare excerpts with the source material: omissions and redactions must be visible and preserve meaning. Keep originals in the appropriate restricted location instead of adding identifiers to the shared table. Examine broad themes that may hide different problems. Include actual comments that complicate the pattern, and note when a comment appears in multiple themes so category totals are not mistaken for distinct people.
Ask another agent to check the result
Recount distinct participants using the supplied non-identifying codes and identify repeated messages counted twice. Check that excerpts match the source text apart from visible redactions, without changing meaning. Flag identifying detail or paraphrases presented as quotations. Note overlapping themes whose counts cannot be added as separate people.
Decide what the themes can support
Mark which themes rest on several distinct customers and which rest on one or two. Treat small counts as leads for follow-up questions. Choose one or two open questions to bring to a customer conversation or a short survey. Keep the source file beside the table so any quote can be traced to its original message.