qa-aman/churn-risk-analysis

Overview

Based on "Customer Success" by Mehta, Steinman & Murphy, churn rarely happens by surprise. The signals appear weeks or months before a customer cancels - they just go unread. This skill structures those signals into a scored risk model and prescribes specific interventions for each risk tier.

The core insight: churn is a lagging indicator. The leading indicators are low usage, missing stakeholders, unresolved complaints, and strategic misalignment. Act on those, not the cancellation notice.

Workflow

Step 1: Score the account across six risk dimensions

Rate each dimension 1 (healthy) to 3 (at risk):

| Dimension | 1 - Healthy | 2 - Watch | 3 - At Risk | |-----------|-------------|-----------|-------------| | Usage | MAU >80% of licenses, core feature adoption >70% | MAU 50-80%, spotty feature use | MAU <50%, logins declining MoM | | Stakeholder coverage | Champion active, exec sponsor engaged | Champion disengaged, no exec relationship | Champion left, no internal advocate | | Support load | <3 tickets/month, all resolved | 4-8 tickets/month, 1-2 lingering | >8 tickets/month or critical unresolved | | NPS / sentiment | NPS 8-10, positive qualitative signals | NPS 6-7, neutral | NPS 0-5, complaints escalating | | Goal progress | On track for agreed outcomes | Behind on 1-2 goals | No measurable progress, ROI questioned | | Renewal proximity | >6 months out | 3-6 months | <3 months with unresolved issues |

Composite score = sum of all dimensions (6-18)

• 6-9: Healthy - standard cadence
• 10-13: Watch - increase touchpoint frequency
• 14-18: At Risk - immediate intervention required

Step 2: Identify which risk flags are present

Check for these specific churn signals from the CS literature:

• Champion left: The person who bought and championed the product is no longer at [customer name]
• Budget scrutiny: Customer mentions cost reviews, budget freezes, or consolidation initiatives
• Silent period: No product login or CSM response for 3+ weeks without explanation
• Executive disengagement: Exec sponsor declined last QBR or stopped attending
• Competitive mention: Customer references a competitor by name in conversation
• Support frustration: Repeated tickets on the same issue, escalating tone in communications
• Failed expansion refusal: Customer declined an upsell or said "we need to right-size first"
• Org change: Merger, acquisition, leadership change, or restructuring announced

Document which flags are active for [customer name].

Step 3: Assign risk tier and intervention playbook

Tier 1 - Healthy (score 6-9) Standard success motion:

• Monthly check-in call
• Quarterly health review
• Proactive feature enablement aligned to goals

Tier 2 - Watch (score 10-13) Elevated engagement:

• Bi-weekly calls until score improves
• Build or refresh the Success Plan with new goals
• Loop in 1-2 additional stakeholders to reduce champion dependency
• Identify 1 quick win to re-establish momentum

Tier 3 - At Risk (score 14-18) Immediate intervention:

1. CSM escalates to CS Manager within 24 hours
2. Executive sponsor meeting booked within 2 weeks
3. Remediation plan prepared with specific commitments and dates
4. Internal coordination: loop in sales, support, and product if needed
5. Weekly status cadence until score drops to Tier 2

Step 4: Map root cause to intervention type

| Root Cause | Primary Intervention | |------------|---------------------| | Low usage | Activation sprint - identify unused high-value features, run targeted enablement | | Champion left | Stakeholder mapping - identify new champion, executive introduction via your sponsor | | Unresolved support issue | Escalation - assign DRI, set resolution SLA, provide daily updates | | ROI not demonstrated | Value audit - pull outcome data, schedule executive value review | | Competitive threat | Competitive positioning call - address gaps directly, accelerate roadmap items | | Budget pressure | Value consolidation story - quantify ROI, provide cost justification data |

Step 5: Write the intervention brief

For at-risk accounts, produce a one-page brief for internal stakeholders:

Account: [customer name] ARR: [value] Renewal date: [date] Risk score: [X/18] - Tier [1/2/3] Active flags: [list] Root cause hypothesis: [1-2 sentences] Proposed intervention: [specific actions, owners, dates] Success criteria: [what "recovered" looks like with a measurable signal]

Anti-Patterns

1. Waiting for the customer to say they're leaving Bad: Only flagging churn risk when a customer sends a cancellation email. Good: Scoring all accounts monthly. Catching at-risk signals 90+ days before renewal.

2. One-dimensional health scoring Bad: Defining "at risk" as only low usage. Good: Scoring across six dimensions. An account with high usage but no executive sponsor and a recent support escalation is still at risk.

3. Intervention without root cause Bad: "Account is red, scheduling an EBR." Good: "Champion left 6 weeks ago. We have no internal advocate. First step is stakeholder mapping to identify a new champion before any executive outreach."

4. No internal escalation path Bad: CSM handles Tier 3 accounts alone without manager involvement. Good: Tier 3 automatically triggers manager review and cross-functional coordination.

5. Treating all at-risk accounts the same Bad: Same playbook for low-usage accounts and accounts with executive disengagement. Good: Match intervention to root cause. Low usage needs activation. Lost champion needs relationship rebuilding. Support frustration needs escalation and resolution.

Quality Checklist

• [ ] Account scored across all six risk dimensions (not just usage)
• [ ] Composite score calculated and risk tier assigned
• [ ] Active churn flags identified and documented
• [ ] Root cause hypothesis stated (not just symptoms)
• [ ] Intervention playbook matches the root cause, not just the tier
• [ ] For Tier 3: manager notified, executive meeting booked, weekly cadence set
• [ ] Intervention brief written for at-risk accounts with ARR and renewal date visible
• [ ] Success criteria defined - what does "recovered" look like?
• [ ] No personal names, internal tool names, or company-specific references in template