📋 Backlog Grooming

Translate Support Themes Into Backlog Items

Clusters repeat customer support tickets into 5-8 underlying product themes and converts each into a backlog item with impact data. Built for PMs who need to turn a noisy support queue into a clear signal.

This prompt clusters a batch of customer support tickets into 5-8 product themes using a problem-pattern approach, then drafts a backlog item per theme with ticket count, customer count, and suggested severity. The output is ready to paste into a backlog tool.

When to use this prompt

Reach for this monthly or after any spike in ticket volume. You will need at least 30-50 recent support tickets (titles and a 1-2 sentence summary each) to get meaningful clusters. It works with any source: Zendesk exports, Intercom transcripts, or a hand-curated list. The prompt assumes support tickets already have some structure; if you have free-text only, the themes will be coarser. The output is a first-pass synthesis that should always be reviewed against your discovery notes before committing to the backlog.

The Prompt

Role: Product Manager Variables: {{tickets}}, {{time_period}}, {{product_area}}, {{total_tickets}} 
You are a product manager synthesizing a batch of customer support tickets into backlog-ready themes. Your job is to find the 5-8 recurring underlying problems and propose backlog items that would eliminate them.

Support tickets (title + summary):
{{tickets}}

Time period: {{time_period}}
Product area: {{product_area}}
Total ticket count: {{total_tickets}}

Produce the output in three parts:

1. THEME CLUSTERS — 5-8 themes, ranked by ticket count. For each theme:
   - Theme name (short, specific)
   - Ticket count in this batch
   - Distinct customer count (dedupe accounts)
   - 2 sentence problem description in customer language
   - Example ticket IDs (up to 3)

2. BACKLOG ITEMS — For each theme, draft one backlog item with:
   - Title (verb-led, under 10 words)
   - User story (As a... I want... so that...)
   - Suggested severity (P0-P4)
   - Expected ticket deflection if solved

3. NOT A THEME — Any tickets that did not cluster into a repeat pattern. List them as possible one-off bugs.

Rank the backlog items by expected impact (customer count x severity). Do not invent ticket IDs; only reference ones in the input. If fewer than 30 tickets are provided, say so and warn that the clusters may be noise.

Example Output

THEME CLUSTERS

1. Email notifications delayed or missing (14 tickets, 11 customers)
   Customers report critical alerts arrive 20+ minutes late or not at all. Several escalated to account managers.
   Example IDs: SUP-1201, SUP-1234, SUP-1256.

2. Export to CSV hangs on large datasets (9 tickets, 8 customers)
   Exports over 10k rows time out. Customers lose trust in data accessibility.
   Example IDs: SUP-1188, SUP-1212, SUP-1291.

3. Session expires during long form entry (7 tickets, 6 customers)
   ...

BACKLOG ITEMS

1. Fix email delivery pipeline latency
   As a workspace admin, I want alerts within 60 seconds so that I can act on critical issues immediately.
   Severity: P1. Expected deflection: 14 tickets per month.

2. Stream large CSV exports
   ...

NOT A THEME: SUP-1244 (single report of timezone bug), SUP-1267 (unique integration issue).

Recommended Tools

Productboard excels at linking themes back to the originating customer tickets, preserving the evidence chain for stakeholders. Dovetail is built for research synthesis and will let you attach ticket transcripts directly to each cluster. Jira then acts as the execution tool for the backlog items after themes are validated. This three-tool chain closes the loop from raw signal to shipped fix.

Productboard73/100 Dovetail61/100 Jira Software75/100

Related Prompts

Identify Duplicates and Stale Backlog Items

Audits a backlog of 30+ items and flags duplicates, near-duplicates, and items that have gone stale based on age and activity. Built for PMs inheriting or cleaning up a bloated backlog.

Product Manager

Refine a Bug Report Into an Actionable Ticket

Takes a raw bug report from customer support or a user email and rewrites it as a well-structured engineering ticket with repro steps, severity, and impact assessment. Designed for PMs triaging inbound bugs.

Product Manager

Draft OKRs From a Team Mission Statement

Translates a team mission statement into 2-3 objectives and 3-4 key results each. Built for PMs and team leads starting a new quarter and needing a first draft to iterate on.

Product Manager

Frequently Asked Questions

When should I use this prompt?

Run it monthly, or immediately after a ticket volume spike. It is especially valuable before quarterly planning because it turns anecdote-driven feature requests into data-driven themes. Do not use it on fewer than 30 tickets; small samples produce false clusters and misleading counts. Also skip it if your support team already runs weekly theme syntheses; duplicating that work confuses ownership.

How do I avoid treating noise as a theme?

Require at least 5 tickets from 3 distinct customers before calling something a theme. Anything below that is an anecdote, however painful the individual report is. Also cross-reference against your product analytics: if the theme is real, usage data for the affected flow should show friction (drop-offs, repeat attempts, rage clicks). A theme with no supporting analytics signal is either too new to measure or mis-clustered; investigate before committing to a backlog item.