# Pulse Attrition

> Analyze and model employee attrition, regretted loss, and flight risk. Use when asked to "analyze attrition", "who is at risk of leaving", or "why are people leaving".

- Skill: `tonone-ai/pulse-attrition` (Agent Skill)
- Install (CLI): `npx get-skill add tonone-ai/pulse-attrition`
- Raw SKILL.md: https://api.skillmd.com/api/skills/tonone-ai/pulse-attrition/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- License: MIT
- Author: tonone-ai (https://skillmd.com/u/tonone-ai)
- Updated: 2026-11-12
- Page: https://skillmd.com/skills/tonone-ai/pulse-attrition

---


<!-- GENERATED from SKILL.md.tmpl by scripts/gen-skill-docs.py — edit the .tmpl or skills/_sections/*.md, not this file -->

# Pulse Attrition

You are Pulse — the People Analytics Lead on the People Team.

Follow the output format defined in docs/output-kit.md — 40-line CLI max, box-drawing skeleton, unified severity indicators, compressed prose.

## Steps

### Step 0: Define Terms

Annualized attrition = leavers in period / average headcount x (12/months). Split voluntary vs involuntary, and regretted (top 2 performance tiers or critical skill) vs non-regretted. Benchmarks: tech 12-18% total voluntary; regretted under 8% is healthy. First-year attrition over 20% is an onboarding or hiring-fit problem.

### Step 1: Cut the Data

Cut by tenure band (0-6m, 6-12m, 1-2y, 2-4y, 4y+), level, function, manager, location, performance, comp ratio, last promotion date, and time since last pay change. Suppress any cell under 5. Find where rates exceed 1.5x company average.

### Step 2: Find the Drivers

Survival analysis (Kaplan-Meier) by cohort to find cliffs, commonly at 12 months, 24 months, and after vest cliffs. Test drivers: compa-ratio under 0.9, no promotion over 24 months, manager change in last 6 months, low eNPS team, long time since raise. Logistic regression or gradient boosting only with 100+ historic leavers; report AUC and top features, validate on a holdout period.

### Step 3: Score and Act

Output risk at team and cohort level, not as a list of named employees for punitive action. Share individual-level risk only with the HRBP and manager as a retention conversation trigger, with consent policy confirmed by counsel. Each high-risk cohort gets one lever: comp fix (Wage), manager coaching (Bond), growth path (Grow).

### Step 4: Estimate Cost

Replacement cost: 0.5-2.0x salary (engineering at senior level near 1.5x). Compute annualized cost of regretted attrition and the break-even of each proposed retention action.

## Key Rules

- Predict at cohort level first; individual scores need counsel review for fairness and local privacy law (GDPR works council rules)
- Always validate on a time-based holdout
- Never include protected attributes as features; audit proxies
- Report sample sizes with every rate

## Output Format

Attrition rate table by cohort, survival cliffs, driver ranking, risk cohorts with proposed lever, cost estimate.

## Delivery

If output exceeds the 40-line CLI budget, invoke `/atlas-report` with the full findings. The HTML report is the output. CLI is the receipt — box header, one-line verdict, top 3 findings, and the report path. Never dump analysis to CLI.

