# Customer Service Excellence > Use when handling service recovery, escalations, complaints, apology language, frontline empowerment, CX measurement, or prevention loops. - Skill: `peterbamuhigire/customer-service-excellence` (Agent Skill, multi-file: 6 files) - Install (CLI): `npx skillmds@latest add peterbamuhigire/customer-service-excellence` - Raw SKILL.md: https://api.skillmd.com/api/skills/peterbamuhigire/customer-service-excellence/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: peterbamuhigire (https://skillmd.com/u/peterbamuhigire) - Updated: 2026-09-10 - Page: https://skillmd.com/skills/peterbamuhigire/customer-service-excellence --- # Customer Service Excellence Acknowledgement: Shared by Peter Bamuhigire, techguypeter.com, +256 784 464178. ## Required Inputs | Input | Required | Use | |---|---|---| | Decision, audience, and deliverable | yes | Bound the business outcome | | Source evidence, constraints, and owner | yes | Ground recommendations and accountability | | Approved budget, customer data, or production artefacts | conditional | Support high-impact execution | ## Capability and permission contract Default to read-only analysis and drafting. Do not publish, send, price, promise, alter customer records, commit budget, or modify production artefacts without explicit authority and a named approver. Minimise confidential data, preserve provenance, and keep reversible copies. ## Degraded mode If evidence, stakeholder decisions, specialist tooling, or authoritative commercial data are unavailable, deliver a labelled draft, checklist, or decision memo. State what was not verified and do not claim approval, publication, financial accuracy, or customer acceptance. ## Decision rules | Condition | Action | Stop condition | |---|---|---| | Output creates a commercial, customer, or delivery commitment | Obtain named approval before release | Authority or terms are unclear | | Evidence supports a reversible draft | Produce it with assumptions and owner | Required evidence conflicts | | Tooling or data is incomplete | Specify validation | A final executable artefact is expected | ## Domain Anti-Patterns - Inventing customer evidence, prices, benchmarks, or approvals. Fix: cite the source or mark the gap. - Publishing or sending a draft without authority. Fix: retain draft status and name the approver. - Hiding assumptions inside polished prose. Fix: expose them beside each affected decision. - Polishing presentation while the decision remains unclear. Fix: resolve audience, owner, and acceptance criteria. - Treating unavailable tooling as passed validation. Fix: record the unassessed check. ## Use When - Handling a live service issue (complaint, error, delay, escalation, public-channel post). - Drafting service language: apology, empathy, ownership, timeline, confirmation, escalation, refusal. - Designing or auditing the service recovery and retention loop. - Measuring service quality across response, resolution, recovery, retention. - Aligning frontline empowerment with escalation paths and knowledge management (CX-EX alignment). - Building a prevention loop so the same failure does not recur. ## Do Not Use When - The work is pre-sale conversation; use `premium-client-sales`. - The work is product feature evaluation; use `product-discovery`. - The work is internal team conflict unrelated to a customer; use a coaching framework. - The work is marketing copy at the top of the funnel; use the marketing/positioning skill. ## Required Inputs - The customer's words (verbatim where possible) and the channel they used. - Account context: tenure, value, prior issues, contract terms. - The actual failure: what happened, what was promised, what gap. - Authority limits of the agent handling the interaction (refund cap, credit cap, escalation triggers). ## Workflow 1. Run the **recovery-and-retention loop**: listen, acknowledge, diagnose, explain, options, act, confirm, document, prevent, follow up. Do not skip steps; the order matters. 2. Match the interaction type to a **difficult-interaction framework** (angry, anxious, technical-novice, executive, public-channel) and adapt language accordingly. 3. Use **service language patterns** for apology, empathy, ownership, timeline, confirmation, escalation. Each pattern has a structure and a prohibited phrase list. 4. Honour the **frontline empowerment ceiling**: if the resolution exceeds the agent's authority, escalate immediately rather than negotiating beyond authority. 5. Close with **explicit confirmation**: name what was done, what comes next, and the date the customer can hold you to. 6. Document the interaction with structured tags so it feeds the **service quality measurement** dashboard and the **prevention loop**. 7. Run a follow-up after resolution. Without follow-up, recovery is half-done. ## Quality Standards - The customer's words are mirrored before any explanation begins. - Apology is specific and ownership-bearing, not "we apologize for any inconvenience". - Every commitment has a date, an owner, and a confirmation channel. - Escalation is a service tool, not a defeat. Escalate before negotiating beyond authority. - Public-channel responses move to private channel for resolution but acknowledge publicly first. - Every documented failure produces either a knowledge-base entry, a process change, or a known-acceptable-loss decision. ## Anti-Patterns - Apologising in passive voice ("mistakes were made"). Specific ownership outperforms generic regret. - Explaining policy before acknowledging impact. - Promising a callback "soon" instead of a specific time window. - Treating the angry customer's tone as the issue. The tone is data; the issue is the failure. - Closing the ticket because the customer went silent. Silence is not resolution. - Making the customer repeat the story to a second agent. Hand-off discipline is part of service. - Using empathy phrases as filler. Empathy without diagnosis is theatre. ## Outputs - A resolved or escalated case with named owner and date. - A written confirmation to the customer (channel of their choice). - A structured tag set for measurement (response time, resolution path, recovery used, retention outcome). - A prevention entry: knowledge-base update, process change, or accepted-loss note. - A follow-up scheduled at an explicit date. ## Evidence Produced | Category | Artifact | Format | Example | |----------|----------|--------|---------| | Operability | Service recovery plan | Markdown with acknowledgement, owner, update cadence, escalation, and closure condition | `docs/service/recovery-plan.md` | | Release evidence | Support language script set | Markdown library for acknowledgement, referral, escalation, follow-up, and recovery | `docs/service/support-scripts.md` | | UX quality | Service quality dashboard definition | Markdown or CSV with response, resolution, recurrence, and satisfaction metrics | `docs/service/service-quality-dashboard.md` | ## References - `references/recovery-and-retention-loop.md` for the ten-step loop and exit criteria per step. - `references/service-language-patterns.md` for apology, empathy, ownership, timeline, confirmation, escalation, refusal patterns with prohibited phrases. - `references/difficult-interaction-frameworks.md` for angry, anxious, technical-novice, executive, public-channel handling. - `references/service-quality-measurement.md` for the four-metric dashboard (response, resolution, recovery success, retention impact). - `references/cx-ex-alignment.md` for frontline empowerment, escalation paths, and knowledge management. - Use `continuous-improvement-system` to feed prevention entries into operating cadence reviews. - Use `premium-client-sales` for the inverse direction (pre-sale conversations). ## Recovery and Retention Loop (the order, every time) Listen -> Acknowledge -> Diagnose -> Explain -> Options -> Act -> Confirm -> Document -> Prevent -> Follow up. Skipping listen produces explanation theatre. Skipping diagnose produces apology theatre. Skipping prevent produces the same failure next month. Skipping follow up produces silent churn. ## Authority Ceiling Rule Frontline agents must have a published refund cap, credit cap, and escalation trigger. Negotiating beyond authority destroys two relationships at once: the customer's trust (they sense the discomfort) and the agent's relationship with their manager (escalating after a promise burns capital both ways). ## The Public-Channel Move Public complaint -> public acknowledgement (under one hour, no defence, no detail) -> private channel handover -> resolution in private -> public closing note (only if the customer agrees). Resolving in public is theatre and invites pile-on. ## Service Failure Prevention Loop Every documented failure produces one of three outcomes: - A knowledge-base entry (the agent could have answered with the right information). - A process change (the system allowed the failure; fix it or it recurs). - An accepted-loss note (the failure cost is below the prevention cost; document and move on). Without the third option, prevention loops accumulate well-meaning policy that no one follows.