customer-success-playbook

Files (8)

• README.md
• SKILL.md
• evals.json
• churn-prediction.md · references/
• expansion-playbooks.md · references/
• health-score-model.md · references/
• health-score-models.md · references/
• qbr-templates.md · references/

SKILL.md (first lines)

## Key principles

1. **Proactive, not reactive** - CS that only responds to support tickets is account
   management, not customer success. Intervene before customers feel pain. The best
   save is the one that never needed saving.

2. **Health scores drive action** - A health score that lives in a dashboard but
   never triggers a workflow is decoration. Every health band must have an associated
   motion: what the CSM does, when, and how. Score without action is noise.

3. **Onboarding determines lifetime value** - The first 30-90 days set the trajectory
   for the entire customer relationship. Customers who reach their first "aha moment"
   quickly retain at 2-3x the rate of those who struggle. Invest disproportionately
   in time-to-value.

4. **Expansion is earned, not sold** - Upsells from customers who haven't achieved
   their desired outcomes produce churn, not growth. Expansion should follow proven
   value, not quota pressure. The signal to expand is the customer asking for more,
   not the CSM pitching more.

5. **Segment by value and risk** - Not all customers deserve the same coverage model.
   High-ARR accounts need white-glove, human-led success. Low-ARR accounts need
   scalable, tech-touch programs. Mismatching coverage to tier burns CSM capacity on
   accounts that can't justify the cost and underserves accounts that need attention.

---

## Core concepts

### The CS lifecycle

Every customer moves through predictable phases, each with distinct success criteria:

| Phase | Duration | Goal | Key risk |
|---|---|---|---|
| Onboarding | Days 1-90 | First value realization | Slow time-to-value, scope creep |
| Adoption | Months 3-12 | Broad, deep usage across team | Shallow single-user adoption |
| Renewal | 90 days before renewal | Confirmed ROI, signed renewal | Surprise objections at renewal |
| Expansion | Post-renewal or milestone | Upsell based on proven value | Premature pitching |
| Advocacy | Ongoing | Reference, case study, promoter | Neglect after expansion |

### Health score components

A well-designed health score is a weighted composite of leading indicators that
predict renewal probability. Common dimensions and typical weights:

| Dimension | Signal examples | Typical weight |
|---|---|---|
| Product usage | DAU/WAU, feature adoption depth, seats used/licensed | 30-40% |
| Engagement | CSM touchpoint frequency, sponsor responsiveness | 20-25% |
| Outcomes | Goals achieved vs. committed, ROI metrics | 20-25% |
| Support | Ticket volume, CSAT, unresolved critical issues | 10-15% |
| Relationship | Executive sponsor status, champion stability, NPS | 10-15% |

See `references/health-score-model.md` for the detailed weighted model and threshold design.

### Churn indicators

**Leading indicators (intervene now):**
- Login frequency drops >30% week-over-week for 2+ consecutive weeks
- Champion or executive sponsor changes roles or leaves
- Support ticket volume spikes with CSAT below 3/5
- Renewal conversation not started within 90-day window
- Customer misses two consecutive scheduled CSM touchpoints

**Lagging indicators (harder to reverse):**
- Customer requests data export
- Customer asks about contract termination clauses
- NPS drops to detractor (0-6) territory

### Expansion signals

- **Usage ceiling** - Feature utilization approaching licensed limit (>80% of seats used)
- **Adjacent pain** - Customer raising problems the upsell product directly solves
- **Organizational spread** - Multiple departments asking for access beyond the pilot team
- **Renewal enthusiasm** - Customer signs renewal early or references product in internal materials
- **Executive sponsorship shift** - C-suite starts attending success calls

---

## Common tasks

### Build a customer health scoring system

Design a weighted, multi-dimensional model that produces a single score (0-100) and
a color-coded band (Red / Yellow / Green) with automatic CSM action triggers.

**Step 1 - Define dimensions and weights.**
Select 4-6 dimensions relevant to your business. Product usage should carry the
highest weight (30-40%) because it is objective and hardest to fake.

**Step 2 - Normalize each dimension to 0-100.**
Map each raw metric to a 0-100 sub-score using thresholds. Example for usage:
- <20% of seats active in last 30 days = 0
- 20-49% = 40, 50-74% = 70, 75-100% = 100

**Step 3 - Apply weights and compute composite.**
`health_score = (usage * 0.35) + (engagement * 0.25) + (outcomes * 0.20) + (support * 0.10) + (relationship * 0.10)`

**Step 4 - Define bands and mandatory actions.**

| Band | Score range | CSM action |
|---|---|---|
| Green | 75-100 | Expansion motion, reference request |
| Yellow | 50-74 | Scheduled check-in within 7 days, risk assessment |
| Red | 0-49 | Executive escalation within 48 hours, save plan |

**Step 5 - Build a feedback loop.** Compare health score 90 days prior to renewal
against actual renewal outcome. Tune weights until model achieves >75% predictive
accuracy for churn.

See `references/health-score-model.md` for the full scoring template.

### Design an onboarding playbook

Onboarding ends when the customer achieves their first committed outcome, not when
technical setup is complete. Structure around milestones, not calendar dates.

**Milestone 1 - Technical kickoff (Days 1-7)**

From mkurman/zorai, MIT.

More from mkurman/zorai