Support Ticket Theme Miner - E-commerce
Use this skill when
The user wants to turn support tickets into store improvements.
Common requests:
- "Analyze these support tickets."
- "What questions keep repeating?"
- "What should we add to product pages or FAQ?"
- "Find automation candidates for support."
Required input
Minimum useful input:
- Support tickets, chats, or email samples.
- Time period and channel.
- Tags or categories if available.
- Help center, policy, or FAQ links if available.
Optional:
- Product page URLs.
- Order status tags.
- Resolution time.
- Customer sentiment.
- Refund or return outcomes.
Before analysis
- Confirm whether tickets are pre-purchase, post-purchase, support, returns, or mixed.
- Remove or avoid exposing sensitive customer data where possible.
- Do not recommend full automation before identifying theme quality.
- Keep escalation topics human.
Analysis workflow
- Cluster tickets by theme:
- order status
- shipping time
- returns/exchanges
- sizing/fit
- compatibility
- product usage
- discount/payment issues
- damaged/wrong item
- policy confusion
- Identify repeated questions and root information gaps.
- Map themes to fixes:
- product page
- checkout
- help center
- transactional email
- support macro
- automation candidate
- human escalation
- Prioritize by frequency, buyer impact, and operational load.
- Give every theme a denominator. A theme is a share of total tickets in a stated period, not a raw count, otherwise a busy month reads as a new problem and a quiet one hides an old one.
Decision and evidence standard
Every finding should include:
- Evidence tag:
export, screenshot, url, review_cluster, support_ticket, return_reason, policy, feed_diagnostics, margin_csv, inventory_export, hypothesis, or needs_data.
- Severity: low, medium, high, or critical.
- Confidence: low, medium, or high.
- Business impact: revenue, margin, cashflow, retention, conversion, support_load, or risk.
- Effort: XS, S, M, or L.
- Owner decision: do_now, test, investigate, monitor, ignore, or approval_needed.
If the evidence is weak, mark the finding as hypothesis or needs_data and lower confidence.
Output format
Support verdict
Top repeated themes and the likely store-side causes.
Theme table
| Theme |
Frequency signal |
Likely cause |
Recommended fix |
Automation fit |
Content updates
FAQ, PDP, policy, and help center updates to consider.
Human escalation topics
Issues that should stay with a human agent.
Example input and output
Input:
- support ticket export
- chat transcripts
- policy links
- affected product URLs
Good output excerpt:
| Finding |
Evidence |
Severity |
Confidence |
Business impact |
Effort |
Owner decision |
| Customers repeatedly ask whether Product A fits Product B |
support_ticket, url |
high |
high |
conversion/support_load |
S |
do_now |
| Refund edge cases need human escalation |
support_ticket, policy |
medium |
high |
risk/support_load |
S |
approval_needed |
What not to do yet: automate refunds, cancellations, or sensitive complaints by default.
Guardrails
- Do not expose personal customer data.
- Do not write final support macros unless policy and tone are provided.
- Do not automate refunds, cancellations, or sensitive complaints by default.
- Do not treat high frequency as proof that customers are wrong.
- Do not ignore product or fulfillment root causes.
1---2name: support-ticket-theme-miner-ecommerce3description: Analyzes e-commerce support ticket exports, chat transcripts, email samples, and help center content to identify repeated questions, policy confusion, product information gaps, and automation candidates. Use when the same questions keep arriving, support load grows faster than orders, or the store wants to know which page or policy is generating its own tickets.4---56# Support Ticket Theme Miner - E-commerce78## Use this skill when910The user wants to turn support tickets into store improvements.1112Common requests:1314- "Analyze these support tickets."15- "What questions keep repeating?"16- "What should we add to product pages or FAQ?"17- "Find automation candidates for support."1819## Required input2021Minimum useful input:2223- Support tickets, chats, or email samples.24- Time period and channel.25- Tags or categories if available.26- Help center, policy, or FAQ links if available.2728Optional:2930- Product page URLs.31- Order status tags.32- Resolution time.33- Customer sentiment.34- Refund or return outcomes.3536## Before analysis37381. Confirm whether tickets are pre-purchase, post-purchase, support, returns, or mixed.392. Remove or avoid exposing sensitive customer data where possible.403. Do not recommend full automation before identifying theme quality.414. Keep escalation topics human.4243## Analysis workflow44451. Cluster tickets by theme:46 - order status47 - shipping time48 - returns/exchanges49 - sizing/fit50 - compatibility51 - product usage52 - discount/payment issues53 - damaged/wrong item54 - policy confusion552. Identify repeated questions and root information gaps.563. Map themes to fixes:57 - product page58 - checkout59 - help center60 - transactional email61 - support macro62 - automation candidate63 - human escalation644. Prioritize by frequency, buyer impact, and operational load.655. Give every theme a denominator. A theme is a share of total tickets in a stated period, not a raw count, otherwise a busy month reads as a new problem and a quiet one hides an old one.6667## Decision and evidence standard6869Every finding should include:7071- Evidence tag: `export`, `screenshot`, `url`, `review_cluster`, `support_ticket`, `return_reason`, `policy`, `feed_diagnostics`, `margin_csv`, `inventory_export`, `hypothesis`, or `needs_data`.72- Severity: low, medium, high, or critical.73- Confidence: low, medium, or high.74- Business impact: revenue, margin, cashflow, retention, conversion, support_load, or risk.75- Effort: XS, S, M, or L.76- Owner decision: do_now, test, investigate, monitor, ignore, or approval_needed.7778If the evidence is weak, mark the finding as `hypothesis` or `needs_data` and lower confidence.7980## Output format8182### Support verdict8384Top repeated themes and the likely store-side causes.8586### Theme table8788| Theme | Frequency signal | Likely cause | Recommended fix | Automation fit |89|---|---|---|---|---|9091### Content updates9293FAQ, PDP, policy, and help center updates to consider.9495### Human escalation topics9697Issues that should stay with a human agent.9899## Example input and output100101Input:102103- support ticket export104- chat transcripts105- policy links106- affected product URLs107108Good output excerpt:109110| Finding | Evidence | Severity | Confidence | Business impact | Effort | Owner decision |111|---|---|---|---|---|---|---|112| Customers repeatedly ask whether Product A fits Product B | `support_ticket`, `url` | high | high | conversion/support_load | S | do_now |113| Refund edge cases need human escalation | `support_ticket`, `policy` | medium | high | risk/support_load | S | approval_needed |114115What not to do yet: automate refunds, cancellations, or sensitive complaints by default.116117## Guardrails118119- Do not expose personal customer data.120- Do not write final support macros unless policy and tone are provided.121- Do not automate refunds, cancellations, or sensitive complaints by default.122- Do not treat high frequency as proof that customers are wrong.123- Do not ignore product or fulfillment root causes.
mardab96