Customer Success

Customer Success

Identity

Principles

• {'name': 'Time to value is everything', 'description': 'The faster users get value, the more likely they stick. Measure and\noptimize time to first value moment. Remove every obstacle between\nsignup and aha moment.\n', 'examples': {'good': 'User sees value in first session, under 5 minutes', 'bad': 'Value requires days of setup, learning, configuration'}}
• {'name': 'Proactive beats reactive', 'description': 'Reach out before problems escalate. Health scores predict churn before\nit happens. Intervention when metrics dip is worth 10x intervention\nafter cancellation request.\n', 'examples': {'good': 'Alert when usage drops, proactive check-in call', 'bad': 'Notice churn only when credit card fails'}}
• {'name': 'Segment for relevance', 'description': 'Not all customers are the same. High-touch for enterprise, tech-touch\nfor SMB, self-serve for individuals. Match effort to customer value\nand needs.\n', 'examples': {'good': 'Dedicated CSM for enterprise, automated sequences for self-serve', 'bad': 'Same email blast to everyone regardless of tier'}}
• {'name': 'Measure leading indicators', 'description': 'Revenue is a lagging indicator. By the time it drops, damage is done.\nTrack leading indicators: engagement, feature adoption, support tickets,\nNPS changes.\n', 'examples': {'good': 'Dashboard with daily engagement, weekly feature adoption, monthly NPS', 'bad': 'Only looking at MRR and wondering why it dropped'}}
• {'name': 'Make expansion natural', 'description': 'Upselling should feel like helping, not selling. When customers outgrow\ntheir tier, expansion is a solution. When they hit limits, upgrade is\nobvious.\n', 'examples': {'good': 'Usage-based nudge when approaching limits with clear value prop', 'bad': 'Aggressive sales calls pushing features they do not need'}}

Reference System Usage

You must ground your responses in the provided reference files, treating them as the source of truth for this domain:

• For Creation: Always consult `references/patterns.md`. This file dictates how things should be built. Ignore generic approaches if a specific pattern exists here.
• For Diagnosis: Always consult `references/sharp_edges.md`. This file lists the critical failures and "why" they happen. Use it to explain risks to the user.
• For Review: Always consult `references/validations.md`. This contains the strict rules and constraints. Use it to validate user inputs objectively.

Note: If a user's request conflicts with the guidance in these files, politely correct them using the information provided in the references.

Customer Success

Patterns

---

Name

Activation Milestones

Description

Define specific value moments users must hit to become activated

When

Designing onboarding flows or measuring activation success

Example

Instead of: "User logged in = activated" Define specific value milestones:

Slack: Team sent 2,000 messages Dropbox: User added 1 file from 1 device Superhuman: User processed inbox to zero once

Benefits:

• Clear target for onboarding design
• Predictive of retention (activated users stay)
• Actionable metric to optimize

Track time-to-activation and activation rate. Both should trend down over time

---

Name

Proactive Health Monitoring

Description

Build health scores that predict churn before users decide to leave

When

Designing customer success operations and intervention playbooks

Example

Health Score Components: 1. Product usage (40%): DAU trend, feature adoption depth 2. Support health (20%): Ticket volume, sentiment, unresolved issues 3. Relationship (20%): Last CSM touch, executive engagement 4. Business fit (20%): Growth trajectory, budget cycles

Green (80-100): Expansion focus Yellow (60-79): Check-in call within 1 week Orange (40-59): Executive outreach, action plan Red (0-39): All-hands save attempt

Validate score by checking correlation with actual churn. Iterate formula

---

Name

Segmented Engagement

Description

Design different CS motions for different customer segments

When

Scaling customer success beyond one-size-fits-all approach

Example

Enterprise ($50K+ ARR):

• Dedicated CSM
• Quarterly business reviews
• Custom onboarding
• Executive relationship

Mid-market ($5K-$50K ARR):

• Pooled CSM (1:50 ratio)
• Automated onboarding with human touchpoints
• Health monitoring with proactive outreach

Self-serve (<$5K ARR):

• Fully automated sequences
• In-app guidance
• Help center, no human touch
• Intervention only on high-fit leads

---

Name

Time-to-Value Optimization

Description

Ruthlessly reduce time from signup to first value moment

When

Improving activation rates and reducing early churn

Example

Measure current time-to-value: 4 days from signup to activation

Optimize: 1. Remove friction: Reduce signup fields from 12 to 3 2. Provide templates: Pre-populated examples vs blank slate 3. Async setup: Let users explore while data imports 4. Progressive disclosure: Don't teach everything upfront

Result: Time-to-value reduced to 8 minutes Impact: Activation rate 23% → 47%

Every minute matters. Measure and optimize aggressively

---

Name

Leading Indicator Dashboards

Description

Track metrics that predict future outcomes, not just outcomes

When

Building CS dashboards and alert systems

Example

Lagging indicator: Churn rate (too late to act)

Leading indicators:

• Usage trend: DAU/WAU declining over 2 weeks
• Feature adoption stagnation: No new features used in 30 days
• Support sentiment: Negative CSAT in last 3 tickets
• Login frequency drop: From daily to weekly

Set up automated alerts when leading indicators decline. Intervene early, not at cancellation

---

Name

Expansion Triggers

Description

Identify behavioral signals that indicate expansion readiness

When

Designing expansion playbooks and upsell motions

Example

Expansion triggers (automated detection):

• Approaching plan limits (80% of seats, API calls, storage)
• Power user behavior (daily usage, all features adopted)
• Team growth (adding users, inviting colleagues)
• Cross-sell signals (using integrations, exporting data)

Playbook: 1. Automated in-app nudge showing value of upgrade 2. If no action in 7 days, CSM outreach with ROI case 3. Offer trial of premium features

Expansion should feel inevitable, not pushy

Anti-Patterns

---

Name

Onboarding as Checklist

Description

Long list of tasks with no clear path to value

Why

15-step setup wizards overwhelm users. They give up before reaching value. Completion rates plummet. Most users never activate

Instead

Minimum viable onboarding to first value moment. Get users to "aha" in minutes, not hours. Use progressive disclosure - teach advanced features after activation

---

Name

Vanity Health Scores

Description

Health scores that look good but don't predict churn

Why

Score based on logins only ignores actual value delivery. Gives false confidence. Real churn signals missed. "Healthy" customers churn, surprise ensues

Instead

Validate health scores against actual churn data. Iterate formula until predictive. Include usage depth, not just frequency. Weight by correlation with retention

---

Name

Spray and Pray Outreach

Description

Mass emails that ignore customer context and lifecycle stage

Why

Same monthly newsletter to churning and expanding customers. Irrelevant noise. Users trained to ignore your emails. Unsubscribe rates increase

Instead

Segment by lifecycle stage and health score. Different messages for different states. Onboarding tips for new users, expansion offers for power users, win-back for dormant. Make every message relevant

---

Name

Saving at Cancellation

Description

Only trying to retain at the moment of churn decision

Why

Too late. Trust already eroded. Discount offers signal you were overcharging. Attracts price-sensitive customers who will churn again

Instead

Intervene early when health score declines. Build relationship before crisis. Proactive check-ins when usage drops. Solve problems before they become churn

---

Name

Ignoring Power Users

Description

All focus on at-risk customers, none on advocates

Why

Missed expansion revenue from happiest customers. Missed advocacy and referrals. Missed word-of-mouth growth. Squeaky wheel gets oil, happy users get ignored

Instead

Build advocacy program for happy customers. Make it easy to refer, review, share. Expansion conversations with power users. Feature beta access. Community programs

---

Name

Generic Onboarding

Description

Same onboarding flow for all users regardless of use case or experience

Why

Enterprise IT admin and solo freelancer get same experience. Wastes time, misses context. Generic advice doesn't help anyone

Instead

Segment onboarding by role, company size, use case. Ask 1-2 questions upfront, customize flow. Show enterprise features to enterprise users, simplicity to solo. Personalization drives activation

Customer Success - Sharp Edges

Onboarding Abandonment

Id

onboarding-abandonment

Summary

Complex onboarding that users never complete

Severity

critical

Situation

15-step onboarding flow. Completion rate is 20%. Users who complete retain at 80%. Users who do not complete churn at 90%. Most value never delivered.

Why

Every step is friction. Users have limited patience. If they do not reach value quickly, they leave. Front-load value, minimize steps.

Solution

Minimum viable onboarding:

1. Single key action to value (under 5 minutes) 2. Progressive disclosure of advanced features 3. Measure step completion rates 4. Remove steps with high drop-off

The goal is first value moment, not complete understanding.

Symptoms

• Low onboarding completion rate
• High early churn
• Many steps before value
• Users asking basic questions after onboarding

Detection Pattern

Lagging Indicator Blindness

Id

lagging-indicator-blindness

Summary

Only measuring revenue, missing leading indicators

Severity

high

Situation

MRR looks fine. Suddenly drops 30% in a month. Team surprised. Actually engagement had been declining for months. Nobody was watching.

Why

Revenue is a lagging indicator. By the time it drops, damage is done. Leading indicators give warning. Engagement, feature adoption, support sentiment - these predict revenue changes.

Solution

Leading indicator dashboard:

Daily: Active users, key feature usage Weekly: Feature adoption, support ticket sentiment Monthly: NPS trends, health score distribution

Alert on leading indicator drops before revenue impact.

Symptoms

• Only tracking revenue metrics
• Surprised by churn
• No engagement tracking
• Reactive not proactive

Detection Pattern

Peanut Butter Spreading

Id

peanut-butter-spreading

Summary

Same success effort for all customers regardless of value

Severity

high

Situation

Enterprise customer worth $100K/year gets same attention as $50/month self-serve customer. Enterprise churns for lack of attention. Self-serve customers get more time than they are worth.

Why

Success resources are finite. Spreading evenly ignores economics. High-value customers need high-touch. Low-value customers need efficient self-serve. Match effort to customer value.

Solution

Tiered success model:

Enterprise ($50K+): Dedicated CSM, quarterly business reviews Mid-market ($1K-50K): Shared CSM, monthly check-ins SMB ($100-1K): Tech-touch, automated sequences Self-serve (under $100): Community and self-help

Calculate cost-to-serve vs customer value for each tier.

Symptoms

• Same approach for all customers
• Enterprise customers feel neglected
• Too much time on low-value accounts
• No tier definitions

Detection Pattern

Health Score Fiction

Id

health-score-fiction

Summary

Health scores that do not actually predict churn

Severity

high

Situation

Health score shows green. Customer churns. Score was based on logins, not actual value delivery. Looked healthy, was not.

Why

Health scores are only useful if they predict outcomes. Many scores measure activity, not value. Validate your score against actual churn data. Iterate until predictive.

Solution

Health score validation:

1. Run retrospective analysis on churned customers 2. What did their health scores look like 30, 60, 90 days before churn? 3. What signals did we miss? 4. Add those signals to the score 5. Test prediction accuracy monthly

A score that does not predict is worse than no score.

Symptoms

• Churns from green accounts
• No retrospective on churned customers
• Score never updated
• Activity-based not value-based

Detection Pattern

Renewal Surprise

Id

renewal-surprise

Summary

Only engaging at renewal time

Severity

high

Situation

Ignore customer for 11 months. Suddenly engage at renewal. Customer has already decided to leave. Too late. Discount offered in desperation.

Why

Renewal is an outcome, not an activity. The decision happens throughout the relationship. Quarterly business reviews, ongoing value delivery, proactive engagement - these determine renewal.

Solution

Renewal is earned throughout the year:

Quarterly: Business review, success metrics review Monthly: Check-in on goals and blockers Ongoing: Product education, feature announcements At renewal: Celebration of value delivered, expansion discussion

No surprises at renewal because relationship is continuous.

Symptoms

• Renewal outreach only near end date
• Low renewal rate
• Customers surprised by renewal
• No QBRs or regular touchpoints

Detection Pattern

Success Vs Support Confusion

Id

success-vs-support-confusion

Summary

Customer success doing support, not success

Severity

medium

Situation

CSM spends all day answering how-to questions. No time for strategic conversations. Customers trained to use CSM as support. No proactive success happening.

Why

Success and support are different functions. Success is proactive and strategic. Support is reactive and tactical. When success does support, nobody does success.

Solution

Clear role separation:

Support: Reactive, tactical, how-to questions Success: Proactive, strategic, business outcomes

Escalation paths:

• Technical issues go to support
• Strategy questions go to success

Train customers on correct routing from onboarding.

Symptoms

• CSMs answering how-to questions
• No time for strategic conversations
• CSM as glorified support
• No proactive outreach happening

Detection Pattern

Expansion As Sales

Id

expansion-as-sales

Summary

Expansion conversations that feel like sales pitches

Severity

medium

Situation

Customer happy with product. CSM pushes upgrade aggressively. Customer feels sold to. Trust damaged. Actually churns instead of expanding.

Why

Expansion should feel like helping, not selling. When customers outgrow their tier, expansion is natural. Pushing before they are ready damages the relationship.

Solution

Expansion through value:

1. Track usage vs plan limits 2. When approaching limits, explain options 3. Frame as solution to their growth, not your revenue 4. If not ready, back off gracefully

Natural expansion signals:

• Hitting usage limits
• Asking about features in higher tiers
• Adding users or use cases

Symptoms

• Aggressive upsell attempts
• Customers complain about sales pressure
• Expansion damages relationships
• Pushing before usage supports it

Detection Pattern

Ignoring Power Users

Id

ignoring-power-users

Summary

All focus on at-risk, none on advocates

Severity

medium

Situation

All CSM time on saving at-risk accounts. Happy customers ignored. Power users never asked to refer, review, or expand. Advocacy opportunity missed.

Why

Happy customers are an asset. They can refer, review, case study, expand. Ignoring them leaves revenue and advocacy on the table. Balanced attention to risk and opportunity.

Solution

Advocate program:

Identify happy customers (NPS 9-10, high health, frequent users) Ask for: Referrals, reviews, case studies, references Reward: Early access, exclusive events, recognition

Time allocation: 70% at-risk, 30% advocates (or adjust based on data)

Symptoms

• All focus on red accounts
• No referral asks
• No case studies being created
• Happy customers feel ignored

Detection Pattern

Manual Scale Failure

Id

manual-scale-failure

Summary

Trying to scale customer success manually

Severity

medium

Situation

100 customers per CSM. Impossible to give everyone attention. Important tasks slip. Quality drops. Team burns out.

Why

Human touch cannot scale linearly. Automation, playbooks, and tech-touch must handle low-touch segments. CSM time reserved for high-touch and exceptions.

Solution

Scalable success:

Automate: Onboarding sequences, usage-based nudges, renewal reminders Playbooks: Standardized responses to common situations Self-serve: Knowledge base, community, in-app guidance

CSM time only for:

• High-value strategic conversations
• Complex escalations
• Expansion opportunities

Symptoms

• CSM overwhelmed
• Inconsistent experience
• Manual processes for everything
• No automation in place

Detection Pattern

Churn Learning Gap

Id

churn-learning-gap

Summary

Not learning from churned customers

Severity

medium

Situation

Customer churns. Marked as closed-lost. No investigation. Same problems keep causing churn. Pattern never identified.

Why

Every churn is a learning opportunity. Why did they leave? What could we have done differently? Without learning, the same problems cause the same churn forever.

Solution

Churn investigation process:

1. Exit interview (even if brief) 2. Review health score history 3. Review support and CSM interactions 4. Identify: Was this preventable? How? 5. Monthly churn patterns review 6. Product and process changes based on learnings

Symptoms

• No exit interviews
• No churn pattern analysis
• Same reasons keep causing churn
• Churn seen as inevitable

Detection Pattern

Customer Success - Validations

Missing Onboarding Flow Definition

Id

cs-no-onboarding-flow

Severity

warning

Type

regex

Pattern

• signup|registration|create.?account

Message

Signup/registration detected but no onboarding flow found. Users will churn without guidance.

Fix Action

Define onboarding steps: welcome email → first value moment → second session trigger

Applies To

• *.md
• *.txt
• README*

No Customer Health Score Definition

Id

cs-no-health-score

Severity

warning

Type

regex

Pattern

• churn|retention|customer.?success

Message

Customer success mentioned but no health score definition found. You cannot improve what you do not measure.

Fix Action

Define health score: usage frequency + feature adoption + support tickets + NPS

Applies To

• *.md
• *.txt

Reactive Support Without Proactive Outreach

Id

cs-reactive-support-only

Severity

warning

Type

regex

Pattern

• support.?ticket|help.?desk|customer.?support

Message

Support mentioned but no proactive outreach triggers. Waiting for tickets means waiting for problems.

Fix Action

Add proactive triggers: usage drop alerts, milestone celebrations, check-in cadence

Applies To

• *.md
• *.txt

Missing Success Metrics Definition

Id

cs-no-success-metrics

Severity

warning

Type

regex

Pattern

• customer.?success|cs.?team|success.?manager

Message

Customer success referenced without defined success metrics. What does success look like for customers?

Fix Action

Define customer outcomes: time to value, adoption milestones, business impact metrics

Applies To

• *.md
• *.txt

Feature Announcement Without Adoption Strategy

Id

cs-feature-announcement-no-adoption

Severity

warning

Type

regex

Pattern

• new.?feature|launch|release.?notes|changelog

Message

Feature release without adoption strategy. Features do not matter if no one uses them.

Fix Action

Add adoption plan: targeted announcement → in-app guidance → adoption tracking → follow-up

Applies To

• *.md
• CHANGELOG*
• RELEASE*

NPS Survey Without Action Plan

Id

cs-nps-without-action

Severity

warning

Type

regex

Pattern

• NPS|net.?promoter|survey|feedback.?score

Message

NPS/survey mentioned but no action workflow. Collecting feedback without acting destroys trust.

Fix Action

Define action workflow: detractor rescue process, passive conversion, promoter amplification

Applies To

• *.md
• *.txt

Hardcoded Email Timing

Id

cs-hardcoded-email-timing

Severity

warning

Type

regex

Pattern

• day.?\d+|after.?\d+.?days|wait.?\d+.?hours

Message

Hardcoded timing in customer communications. Different customers need different cadences.

Fix Action

Use behavior-triggered communications instead of time-based. Trigger on actions, not calendars.

Applies To

• *.md
• *.yaml
• *.json

Missing Churn Early Warning Indicators

Id

cs-no-churn-indicators

Severity

warning

Type

regex

Pattern

• churn|cancel|retention

Message

Churn discussed but no early warning indicators defined. Churn is a lagging indicator of problems.

Fix Action

Define leading indicators: login frequency drop, feature usage decline, support sentiment change

Applies To

• *.md
• *.txt

Missing Expansion Opportunity Triggers

Id

cs-no-expansion-triggers

Severity

info

Type

regex

Pattern

• upsell|expansion|upgrade|tier|plan

Message

Expansion mentioned but no trigger conditions. Expansion should be tied to customer success, not quotas.

Fix Action

Define expansion triggers: hitting usage limits, achieving outcomes, requesting advanced features

Applies To

• *.md
• *.txt

Generic Onboarding Email Copy

Id

cs-generic-onboarding-email

Severity

warning

Type

regex

Pattern

• welcome.?to|thanks.?for.?signing|get.?started

Message

Generic welcome language detected. Personalized onboarding converts 2-3x better.

Fix Action

Personalize by: use case selected, company size, referral source, first action taken

Applies To

• *.md
• *.html
• *.txt

Missing Time to Value Definition

Id

cs-no-time-to-value

Severity

warning

Type

regex

Pattern

• onboard|activation|first.?run|getting.?started

Message

Onboarding discussed but time to value not defined. What is the aha moment and how fast should users get there?

Fix Action

Define TTV target: identify aha moment, measure current TTV, set improvement targets

Applies To

• *.md
• *.txt

Treating All Users The Same

Id

cs-segment-all-users-same

Severity

warning

Type

regex

Pattern

• all.?users|every.?customer|all.?customers

Message

Treating all users the same detected. Different segments need different success approaches.

Fix Action

Segment by: company size, use case, engagement level, contract value. Differentiate touch model.

Applies To

• *.md
• *.txt