Purpose
Turn raw engagement survey output into a prioritized diagnosis: identify the themes driving scores up or down, separate high-impact problems from low-impact noise using importance-performance logic, and produce a sequenced action plan leadership can execute. Solves the "we have the data but don't know what to fix first" problem that stalls most post-survey action.
When to Use
Invoke this skill when the user asks to:
- Analyze or interpret employee engagement survey results
- Understand why engagement or eNPS scores dropped since last cycle
- Identify the top themes in open-ended survey comments
- Determine which engagement drivers to prioritize for action
- Build a post-survey action plan for leadership or managers
- Compare engagement across departments, tenure, or demographic segments
Inputs Required
Before starting, collect:
- Overall scores by driver: Engagement index, eNPS, and category/driver scores (e.g., leadership, growth, recognition, workload, manager)
- Prior-period scores: Same drivers from the previous survey, if available, for trend analysis
- Verbatim comments: Open-ended responses (paste or summarize by question)
- Demographic cuts: Score breakdowns by department, tenure, level, location, or other segments
- Response rate: Overall and by segment if available
- Benchmark: Industry or internal benchmark scores, if known
- Organizational context: Recent changes (reorg, layoffs, leadership change, RTO policy) that could affect scores
If prior-period scores or benchmarks are missing, analyze on absolute scores and flag where trend context would sharpen the diagnosis.
Instructions
Step 1: Score Diagnosis and Trend Read
Establish the headline picture before diving into causes.
- Segment the drivers: Sort all driver scores into Strengths (top third), Neutral (middle), and Risk Areas (bottom third)
- Trend flags: For each driver, mark the change vs. prior period — improved (+3 or more), stable (±2), declined (−3 or more)
- Response rate check: If overall response rate is below 60%, or any segment is below 50%, flag that segment's results as directional, not conclusive
- Benchmark gap: Where benchmarks exist, note which drivers fall meaningfully below (−5 or more) — these are competitive risks, not just internal weaknesses
Step 2: Verbatim Sentiment and Theme Extraction
Turn open text into structured themes.
- Sentiment tagging: Classify comments as Positive / Neutral / Negative and report the approximate distribution
- Theme clustering: Group comments into 5–8 recurring themes (e.g., unclear career pathing, workload/burnout, manager communication, recognition, tooling, compensation fairness)
- For each theme, capture:
- Frequency: How often it appears (High / Medium / Low)
- Sentiment: Predominantly positive or negative
- Representative quote: One anonymized verbatim that captures it (paraphrase if needed to protect anonymity)
- Linked driver: Which quantitative driver this theme explains
Step 3: Importance-Performance Grid
Map every major driver onto a 2×2 to determine action priority.
- Importance (Y-axis): How strongly the driver correlates with overall engagement/intent to stay. Use stated importance if measured; otherwise infer from theme frequency in verbatims tied to that driver
- Performance (X-axis): The driver's actual score
Place each driver into one quadrant:
- Fix First (High Importance / Low Performance): Top-priority action targets — these depress engagement most
- Maintain (High Importance / High Performance): Protect these; don't neglect strengths
- Monitor (Low Importance / Low Performance): Low scores that don't move engagement — deprioritize
- Deprioritize (Low Importance / High Performance): Fine as-is; no resource allocation needed
Step 4: Segment Analysis
Identify where engagement risk concentrates.
- Hotspot detection: Flag any segment scoring 10+ points below the org average on the engagement index
- Cross-cut patterns: Note if a Risk Area driver is universal or concentrated in specific segments (e.g., workload low only in Product; recognition low across all)
- Tenure curve: Check whether engagement drops at a specific tenure band (a common 6–18 month or 3–5 year dip signals onboarding or growth-ceiling issues)
Step 5: Root Cause Analysis
For each "Fix First" driver, move from symptom to cause.
- State the symptom (the low score), the supporting verbatim evidence, and the most likely root cause
- Apply a 5-whys logic on the top 2–3 drivers to avoid treating surface symptoms
- Distinguish structural causes (systems, policy, staffing) from local causes (specific manager or team)
Step 6: Prioritized Action Plan
Translate the Fix First quadrant into sequenced actions. For each action, specify:
- Action: The specific intervention
- Target driver/theme: What it moves
- Owner: Function or role accountable (HRBP, manager, senior leadership)
- Timeline: Quick win (0–30 days) / Near-term (30–90 days) / Structural (90+ days)
- Success measure: The driver score or behavioral indicator that should improve by next survey
Sequence so that at least 2 credible quick wins land first — visible early action is what preserves trust in the survey process.
Output Format
- Structure: Diagnosis → theme analysis → importance-performance grid → segment findings → root cause → action plan
- Length: 1,000–1,400 words
- Format: Structured document with tables for the importance-performance grid, theme summary, and action plan; prose for root cause narrative
- Sections: Score Diagnosis → Verbatim Themes → Importance-Performance Grid → Segment Hotspots → Root Cause Analysis → Prioritized Action Plan
- Tone: People analytics lead — evidence-led, candid about risk, oriented toward decisions rather than reassurance
Quality Checklist
Before delivering output, verify:
- [ ] Every driver is placed into exactly one importance-performance quadrant
- [ ] 5–8 verbatim themes are extracted, each with frequency, sentiment, a quote, and a linked driver
- [ ] Every "Fix First" driver has a corresponding root cause and at least one action
- [ ] The action plan includes at least 2 quick wins (0–30 days) and specifies owner and success measure for each action
- [ ] Segments below 50% response rate or 10+ points below average are flagged explicitly
- [ ] No action is recommended for a Monitor or Deprioritize driver without stated justification
Notes
- The most common analysis failure is treating all low scores as equally urgent — the importance-performance grid exists specifically to prevent spreading action across low-leverage drivers
- Protect anonymity: never attribute a verbatim to an identifiable individual or a segment small enough to expose one (generally under 5 respondents)
- A tenure-band engagement dip almost always points to onboarding or growth-path problems, not compensation — check the verbatims before assuming pay
- If scores dropped after a known organizational event (reorg, RTO, layoff), separate event-driven volatility from structural decline in the root cause analysis
How to use this skill
This file is a Claude Code skill in Anthropic's standard format — markdown with YAML frontmatter at the top.
Installing in Claude Code:
~/.claude/skills/<folder-name>/SKILL.md— auto-loads in every session anywhere on your machine (recommended for skills you'll use across multiple projects)<your-project>/.claude/skills/<folder-name>/SKILL.md— auto-loads only when working in that project (recommended for project-specific skills)
Restart Claude Code or open a new session to register.
Using in Claude.ai, ChatGPT, Gemini, or other AI tools:
- Drag this file into a new chat, or copy and paste its contents
- Send your task or request as your next message — the AI will apply the skill where it fits
For Claude Code skill details: https://docs.anthropic.com/en/docs/claude-code/skills