Product Management prompt of the day

Support-ticket mining: recurring problems ranked by how often and how badly they hurt

Support has hundreds of tickets a month and product only hears the loudest. Turn a ticket export into a ranked problem list with evidence.

You are a product analyst mining support tickets for product problems. I will paste an anonymized ticket export with categories, text excerpts, and dates.

Produce:

A) PROBLEM CLUSTERS — group tickets by the underlying user problem, not by the category field. For each cluster: a one-line problem statement, ticket count, 2 representative excerpts by ticket ID, and the trend across the period.

B) SEVERITY — for each cluster, the user impact (blocked / workaround exists / annoyance) based on the ticket text, and whether it hits new users, power users, or both.

C) RANKED LIST — clusters ranked by count and severity together, with the reasoning shown.

D) MISCATEGORIZED — tickets whose category field disagrees with their content, so support tagging can improve.

E) OPEN QUESTIONS — what the tickets cannot tell us, such as how many users hit the problem without writing in.

Inputs: [PASTE ANONYMIZED TICKET EXPORT: ID, DATE, CATEGORY, TEXT EXCERPT] · [PRODUCT AREA IN SCOPE] · [PERIOD COVERED]

Rules: Do not invent tickets, counts, or quotes — every cluster must trace to ticket IDs I pasted, and small clusters are labeled "thin". Verify the counts against the support system before they reach a roadmap discussion. Strip customer names, emails, account identifiers, and any personal data before pasting.

Why it works — Support queues contain the most direct evidence of where a product fails, but it arrives as unstructured text sorted by an inconsistent category field. Clustering by underlying problem, attaching counts and ticket IDs, and grading severity from the text turns the queue into a ranked, auditable list. Naming what tickets cannot show, silent users, keeps the analysis from overclaiming.