Skill Details
Runtime
(gpt-5.6)
Example user message
[userMessage]: A regional telecom operator with 80,000 home-broadband customers. Data over the past 6 months: 32,000 highly-satisfied, 16,000 mid, 8,000 low (3,000 explicit survey dissatisfaction), 24,000 not-rated (2,000 with predicted poor experience ≥80%). Top causes among high-risk: equipment faults 1,200; delivery/supply 800; billing 500; slow support 300; other 200. Goal: lift satisfaction, intercept complaints, improve resource efficiency in 3 months. Plan it.
Example agent response
Plan: 1) Segmentation — 3,000 high-risk, 21,000 mid/low, 2,000 latent, 32,000 highly-satisfied. 2) Warning list — rescue 3,000 first, then 2,000 latent. 3) Actions — open 3,000 full-year recovery files routed by root cause (equipment → field team; billing → finance); 21,000 care work orders; 2,000 pre-emptive inspection; 32,000 highly-satisfied → model-identify easy-to-interview, ~10-second rating script. 4) Rhythm — month 1 build foundation + rescue high-risk; month 2 expand care + inspection; month 3 solicit high ratings + review. 5) Closure standard: issue fully resolved AND customer explicitly satisfied.
Key insight: 67% of high-risk complaints are equipment/supply — route them, don't let customer support carry them all.
Instant access
Commercial use
Money‑backBy purchasing this skill, you agree to our terms of service
2
Views
GPT-5.6
A battle-tested customer-satisfaction operations skill. Paste complaints, survey scores, or retention data and it segments customers into priority tiers (explicit dissatisfaction → mid/low ratings → latent poor quality), then plans targeted recovery: recovery files for the high-risk, care work orders for the mid, pre-emptive inspection for the silent, and rating solicitation for the happy. Refined from a real telecom project that lifted satisfaction by 31%. Built for CX, operations, and support
...more
Added 4 days ago
