Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Design customer service operations including tiered support (L1/L2/L3), response templates, SLA definitions, escalation procedures, and complaint handling. Use this skill when the user needs to set up a CS team, create service standards, design escalation flows, or improve response quality — even if they say 'our CS is a mess', 'how should we handle complaints', 'set up support tiers', or 'create CS SOPs'.
.claude/skills/asgard-ai-platform-cs-sop/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -11% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 49% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 23% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 50% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 114% | 0% |
IRON LAW: Tier the Support, Not the Customer
Every customer deserves quality service. But not every issue needs a
senior specialist. Route by ISSUE COMPLEXITY, not by customer "importance."
L1 handles 70-80% of volume (simple, repeatable)
L2 handles 15-20% (requires expertise)
L3 handles 5% (requires engineering or management)| Tier | Handles | Skills Required | Resolution Target | |------|---------|----------------|------------------| | L1 (Basic) | FAQ, order status, password reset, simple returns | Script-following, product basics, empathy | < 5 minutes, first-contact resolution | | L2 (Specialist) | Technical issues, billing disputes, complex returns, product defects | Deep product knowledge, judgment, negotiation | < 24 hours | | L3 (Expert) | System bugs, legal/compliance, executive escalations, crisis | Engineering, legal, or management involvement | < 72 hours, case-by-case |
| Category | Examples | Priority | SLA (First Response) | |----------|---------|----------|---------------------| | Critical | Service outage, security breach, safety issue | P1 | < 15 minutes | | High | Payment failure, account locked, order error | P2 | < 1 hour | | Medium | Product question, feature request, general complaint | P3 | < 4 hours | | Low | Feedback, suggestion, general inquiry | P4 | < 24 hours |
| Trigger | Escalate To | Timeline | |---------|-----------|---------| | L1 can't resolve in 15 min | L2 | Immediate warm handoff | | Customer requests supervisor | L2 or Team Lead | Within 5 minutes | | Issue involves refund > NT$X | L2 (approval authority) | Same interaction | | Legal threat or media mention | L3 + Legal + PR | Immediate | | Repeat contact (3+ on same issue) | L2 + investigation | After 3rd contact |
[Greeting] Hi {name}, thank you for contacting us.
[Acknowledge] I understand you're experiencing {issue}.
[Action] Here's what I've done / Here's what we'll do:
1. {specific action}
2. {timeline}
[Next steps] {what the customer should expect / do next}
[Close] Is there anything else I can help you with?markdown# Customer Service SOP: {Business} ## Support Tiers | Tier | Scope | Team Size | Tools | |------|-------|----------|-------| | L1 | {scope} | {N people} | {tools} | | L2 | {scope} | {N} | {tools} | | L3 | {scope} | {N} | {tools} | ## SLA Targets | Priority | First Response | Resolution | Escalation | |----------|--------------|-----------|-----------| | P1 | {time} | {time} | {to whom} | | P2 | ... | ... | ... | ## Top 10 Contact Reasons | # | Reason | Volume % | Resolution | Template? | |---|--------|---------|-----------|----------| | 1 | {reason} | {%} | L1/L2 | Y/N | ## Escalation Flowchart {Decision tree for when to escalate} ## Quality Metrics | Metric | Target | |--------|--------| | First Contact Resolution | > 70% | | CSAT | > 4.2/5 | | Avg Response Time | < {X} hours | | Escalation Rate | < 20% |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-10 | pass→pass | 11,908 | 13,067 | +10% | 1 | 1 | 0% | 1,700 | 3,331 | +96% | 0 | 0 | — |
case-01 | fail→pass | 32,003 | 23,754 | -26% | 1 | 1 | 0% | 6,089 | 5,412 | -11% | 0 | 0 | — |
case-02 | fail→fail | 46,633 | 17,721 | -62% | 1 | 1 | 0% | 6,576 | 4,878 | -26% | 0 | 0 | — |
case-03 | fail→fail | 40,952 | 38,582 | -6% | 1 | 1 | 0% | 8,312 | 6,284 | -24% | 0 | 0 | — |
case-04 | pass→pass | 15,399 | 19,310 | +25% | 1 | 1 | 0% | 2,401 | 4,257 | +77% | 0 | 0 | — |
case-05 | pass→pass | 14,620 | 11,447 | -22% | 1 | 1 | 0% | 2,331 | 3,131 | +34% | 0 | 0 | — |
case-06 | fail→pass | 24,813 | 18,224 | -27% | 1 | 1 | 0% | 2,013 | 2,990 | +49% | 0 | 0 | — |
case-07 | fail→pass | 18,044 | 12,028 | -33% | 1 | 1 | 0% | 2,309 | 2,840 | +23% | 0 | 0 | — |
case-08 | fail→pass | 18,736 | 24,029 | +28% | 1 | 1 | 0% | 2,991 | 4,497 | +50% | 0 | 0 | — |
case-09 | pass→pass | 12,514 | 9,077 | -27% | 1 | 1 | 0% | 1,817 | 2,411 | +33% | 0 | 0 | — |
case-11 | pass→pass | 7,832 | 4,051 | -48% | 1 | 1 | 0% | 870 | 1,714 | +97% | 0 | 0 | — |
case-12 | pass→pass | 10,872 | 19,833 | +82% | 1 | 1 | 0% | 1,431 | 4,022 | +181% | 0 | 0 | — |
case-13 | pass→pass | 15,629 | 30,404 | +95% | 1 | 1 | 0% | 2,320 | 5,252 | +126% | 0 | 0 | — |
case-14 | fail→pass | 12,881 | 18,855 | +46% | 1 | 1 | 0% | 1,737 | 3,713 | +114% | 0 | 0 | — |
case-15 | pass→pass | 7,880 | 3,814 | -52% | 1 | 1 | 0% | 1,158 | 1,663 | +44% | 0 | 0 | — |
case-16 | pass→pass | 9,989 | 11,100 | +11% | 1 | 1 | 0% | 1,621 | 2,514 | +55% | 0 | 0 | — |
case-17 | fail→pass | 10,818 | 11,167 | +3% | 1 | 1 | 0% | 1,791 | 3,081 | +72% | 0 | 0 | — |
case-18 | fail→pass | 11,656 | 7,371 | -37% | 1 | 1 | 0% | 2,165 | 2,455 | +13% | 0 | 0 | — |
case-19 | pass→pass | 9,106 | 3,312 | -64% | 1 | 1 | 0% | 1,319 | 1,736 | +32% | 0 | 0 | — |
case-20 | fail→pass | 8,902 | 4,505 | -49% | 1 | 1 | 0% | 1,330 | 1,888 | +42% | 0 | 0 | — |
case-21 | pass→pass | 29,767 | 22,099 | -26% | 1 | 1 | 0% | 4,154 | 5,227 | +26% | 0 | 0 | — |
case-22 | pass→fail | 23,531 | 29,589 | +26% | 1 | 1 | 0% | 3,468 | 5,830 | +68% | 0 | 0 | — |
case-23 | pass→pass | 22,752 | 28,114 | +24% | 1 | 1 | 0% | 3,128 | 4,430 | +42% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted. The headline lift of +30 percentage points is the difference between those two pass rates over the 23 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.