Churn Prediction - Complete Reference

Beyond Health Scores: Velocity-Based Risk Modeling

A health score is a snapshot; churn prediction is a trajectory. The core insight is that the rate of change in health signals is often more predictive than the absolute level.

The Velocity Framework

For each health signal, track three values:
  1. Current value (today's score)
  2. Velocity (rate of change over 30 days)
  3. Acceleration (change in velocity - is the decline speeding up?)

Risk Assessment Matrix:

Current Score | Velocity      | Risk Level
--------------|---------------|------------------
High (70+)    | Stable/Up     | Low - monitor
High (70+)    | Declining     | Medium - investigate
High (70+)    | Rapid decline | High - act now
Mid (40-69)   | Improving     | Medium - continue support
Mid (40-69)   | Stable        | Medium-High - proactive outreach
Mid (40-69)   | Declining     | High - escalate
Low (<40)     | Improving     | Medium - recovery in progress
Low (<40)     | Stable        | Critical - recovery stalled
Low (<40)     | Declining     | Emergency - exec escalation

Calculating Velocity

Signal Velocity = (Current_30day_avg - Previous_30day_avg) / Previous_30day_avg

Example:
  DAU/MAU last 30 days: 0.35
  DAU/MAU previous 30 days: 0.50
  Velocity = (0.35 - 0.50) / 0.50 = -0.30 (-30% decline)

Velocity Thresholds:
  > +10%:  Positive trend (green)
  -10% to +10%: Stable (neutral)
  -10% to -25%: Declining (yellow)
  < -25%: Rapid decline (red)

Early Warning System Design

Signal Priority for Early Warning

Not all signals degrade at the same rate before churn. The typical degradation timeline for B2B SaaS (from earliest warning to cancellation):

Timeline Before Churn    | Signal
--------------------------|------------------------------------------
6-12 months              | Executive sponsor engagement drops
4-8 months               | Feature adoption breadth narrows
3-6 months               | Login frequency declines
2-4 months               | CSM meeting cancellations increase
1-3 months               | Support ticket sentiment turns negative
1-2 months               | Renewal conversation stalls
2-4 weeks                | Customer asks about data export/migration
1-2 weeks                | Cancellation request submitted

The earlier signals are harder to detect but give you more time to recover. Invest in tracking executive engagement and feature adoption breadth - they are the 6-month early warning system most CS teams ignore.

Alert Design

Structure alerts in three tiers:

Tier 1 - Automated Watch (no human action required yet):
  Trigger: Any single signal crosses yellow threshold
  Action:  Log to account timeline, include in weekly CSM digest
  Example: "Login frequency dropped 15% MoM for Acme Corp"

Tier 2 - CSM Action Required:
  Trigger: Two or more signals cross yellow, OR any signal crosses red
  Action:  Task assigned to CSM with 48-hour SLA to investigate
  Example: "Acme Corp: Login decline (-25%) + Support sentiment negative"

Tier 3 - Escalation Required:
  Trigger: Health score drops below 30, OR velocity multiplier hits 2.0x
  Action:  CS manager + CSM sync within 24 hours, exec sponsor assigned
  Example: "CRITICAL: Acme Corp health score dropped from 62 to 28 in 30 days"

Churn Cohort Analysis

Analyze churn patterns by cohort to identify systemic issues vs. individual account problems.

Building a Churn Cohort View

Group churned accounts by:

1. Signup cohort (when they became a customer)
   - Are recent cohorts churning faster than older ones?
   - If yes: onboarding or product-market fit problem

2. Segment (Enterprise / Mid-Market / SMB)
   - Is churn concentrated in a specific segment?
   - If yes: pricing, coverage model, or product fit issue for that segment

3. Churn reason category
   - Product gaps, poor adoption, budget, competitor, bad fit, M&A
   - Track distribution over time - is "competitor" growing as a reason?

4. Lifecycle stage at churn
   - First 90 days: onboarding failure
   - 90 days - 1 year: adoption/value failure
   - 1+ years: relationship, evolving needs, or competitive displacement

5. Last known health score trajectory
   - Sudden drop (event-driven): champion departure, outage, billing issue
   - Gradual decline (trend-driven): fading engagement, slow disengagement
   - Never healthy (fit-driven): should not have been sold in the first place

Interpreting Churn Cohort Data

Pattern                          | Root Cause                | Action
---------------------------------|---------------------------|---------------------------
New cohorts churn faster         | Onboarding degraded       | Audit onboarding, time-to-value
Churn spikes in specific months  | Renewal bunching          | Stagger renewals, prep earlier
One segment churns 2x+ others   | Product-market fit gap    | Investigate segment-specific needs
"Competitor" reason trending up  | Competitive pressure      | Win/loss analysis, feature gap review
Most churn from "never healthy"  | Sales/CS misalignment     | Tighten ICP, improve handoff process

Predictive Modeling Approaches

Simple: Rules-Based Scoring

Best for teams with <500 accounts or limited data science resources.

Churn Risk Score = weighted sum of risk factors

Risk Factors:
  Health score < 50:              +30 points
  Health velocity negative:       +20 points
  Days to renewal < 90:           +15 points
  Executive sponsor departed:     +25 points
  2+ negative support tickets:    +15 points
  CSM meetings declined 2x:       +10 points
  No login in 14+ days:           +20 points

Risk Levels:
  0-25:   Low risk
  26-50:  Moderate risk
  51-75:  High risk
  76-100: Critical risk

Intermediate: Logistic Regression

Best for teams with 500+ accounts and 50+ churn events to train on.

Approach:
  1. Define the target: churned within 90 days (binary: yes/no)
  2. Feature set: health score, velocity, engagement metrics, tenure,
     contract value, segment, support ticket count/sentiment
  3. Train on historical data (12-24 months)
  4. Output: probability of churn within 90 days for each account
  5. Set action thresholds: >70% probability = critical, >40% = high

Advantages:
  - Interpretable (you can explain WHY an account is high risk)
  - Works with relatively small datasets
  - Easy to update quarterly with new data

Disadvantages:
  - Assumes linear relationships
  - May miss complex interaction effects

Advanced: Survival Analysis

Best for teams that need to predict WHEN churn will happen, not just IF.

Approach:
  1. Model time-to-churn as a survival function
  2. Accounts that haven't churned are "censored" (still at risk)
  3. Cox Proportional Hazards model identifies which factors accelerate
     or decelerate time-to-churn
  4. Output: hazard ratio for each factor and predicted survival curves

Example output:
  "Executive sponsor departure increases churn hazard by 2.3x"
  "Each additional integrated tool reduces churn hazard by 0.85x"
  "This account has a 35% probability of churning within 6 months"

Involuntary Churn Prevention

Involuntary churn (failed payments, expired cards) is often 20-40% of total churn for self-serve and SMB segments. It deserves its own playbook.

Dunning sequence (pre-failure):
  Day -30:  Alert customer that card expires soon
  Day -14:  Second reminder with one-click update link
  Day -7:   Final warning, highlight what they'll lose access to

Dunning sequence (post-failure):
  Day 0:    Payment failed - retry immediately
  Day 1:    Email: "Payment issue - update your card" (friendly tone)
  Day 3:    Retry payment + second email
  Day 7:    Retry + "Your account will be paused in 7 days" email
  Day 10:   Retry + final warning
  Day 14:   Account paused (not deleted), email with reactivation link
  Day 30:   Account cancelled, data retained for 90 days

Best practices:
  - Retry failed payments 4-6 times over 14 days (different times/days)
  - Use card updater services (Visa Account Updater, Mastercard ABU)
  - Send in-app notifications, not just email
  - Make the update flow one click, no login required
  - Track recovery rate by dunning step to optimize the sequence

Churn Prevention Metrics

Track these metrics to measure the effectiveness of your churn prediction and prevention programs:

Metric                        | Target         | Measurement
------------------------------|----------------|-------------------------------------
Prediction accuracy (AUC)     | >0.75          | % of churned accounts flagged 90d prior
False negative rate           | <15%           | Churned accounts that were scored Green
False positive rate           | <30%           | Red accounts that did not churn
Recovery rate                 | >40%           | Red accounts saved within 90 days
Time to intervention          | <48 hours      | Time from alert to first CSM action
Churn reason accuracy         | >80%           | Validated churn reasons match prediction
Save offer acceptance rate    | 20-35%         | Accounts that accept a retention offer

Expansion Playbooks - Complete Reference

Expansion Types

Understanding the different expansion motions helps target the right approach for each account.

Type          | Definition                              | Typical Trigger
--------------|-----------------------------------------|---------------------------
Seat expansion| More users on existing plan              | Team growth, new hires
Tier upgrade  | Move to a higher plan/tier               | Feature limits hit
Module add-on | Purchase additional product modules      | New use case identified
Usage upgrade | Higher usage limits or volume tier       | Approaching capacity
Cross-sell    | New product from your portfolio          | Adjacent need discovered
Professional  | Services, training, custom development   | Complex implementation need

Signal Detection Framework

Quantitative Signals (data-driven)

Signal                          | Detection Method              | Readiness
--------------------------------|-------------------------------|----------
Seat utilization > 80%          | License usage report          | Hot
API calls approaching limit     | Usage monitoring dashboard    | Hot
New department onboarding       | Distinct user groups forming  | Warm
Feature requests for higher tier| Support/product feedback data | Warm
Usage growing 20%+ QoQ          | Trend analysis                | Warm
Integration count increasing    | Product analytics             | Warm
Admin creating new workspaces   | Product event tracking        | Hot
Export/reporting usage growing   | Feature analytics             | Warm

Qualitative Signals (relationship-driven)

Signal                          | Detection Method              | Readiness
--------------------------------|-------------------------------|----------
Customer mentions new initiative| QBR, check-in calls           | Hot
Executive champion advocates    | NPS/CSAT comments, referrals  | Hot
Customer asks about pricing     | CSM conversations, chat logs  | Hot
New stakeholders joining calls  | Meeting attendee tracking     | Warm
Customer shares positive ROI    | QBR presentations, emails     | Hot
Asks about roadmap features     | CSM notes, support tickets    | Warm
Participates in case study      | Marketing program tracking    | Hot
References you in their docs    | Web monitoring, CSM reports   | Warm

Expansion Timing Framework

When to plant seeds vs. when to propose

Phase 1 - Plant Seeds (ongoing, no ask):
  - Share relevant case studies from similar customers
  - Mention adjacent use cases during regular check-ins
  - Include benchmark comparisons in QBRs showing what peers do
  - Invite to product demos or webinars for premium features
  Duration: Continuous, part of regular CS cadence

Phase 2 - Explore Interest (when 3+ warm signals detected):
  - Ask discovery questions about their evolving needs
  - "I've noticed your team has grown - how are you managing X?"
  - "Other customers in your space have started using Y for Z"
  - Map new stakeholders and their pain points
  Duration: 2-4 weeks of exploration

Phase 3 - Propose Value (when 3+ hot signals detected):
  - Build a business case with quantified ROI
  - Connect specific expansion to their stated goals
  - Introduce AE or solutions engineer for technical scoping
  - Present pricing with anchor to value delivered
  Duration: 2-6 weeks to close

Phase 4 - Close and Onboard:
  - AE leads commercial negotiation
  - CSM ensures smooth onboarding of new capabilities
  - Set new success metrics for expanded scope
  - Update health score model with new baseline

CSM-AE Handoff Protocol

The handoff between CSM and AE for expansion deals is a common failure point.

Handoff checklist

Before introducing AE, CSM must have:
  [ ] Confirmed customer interest (not just CSM interpretation)
  [ ] Identified the economic buyer or decision maker
  [ ] Documented the business need in customer's own words
  [ ] Mapped the evaluation process and timeline
  [ ] Confirmed budget availability (at least directionally)
  [ ] Prepared a brief for AE covering:
      - Account history and relationship dynamics
      - Current product usage and health score
      - Expansion opportunity details
      - Key stakeholders and their roles
      - Potential objections and sensitivities
      - Pricing history and contract terms

Handoff meeting structure:
  1. CSM introduces AE as "specialist" (not "salesperson")
  2. CSM frames the conversation around customer's goals
  3. AE asks discovery questions (not a pitch)
  4. CSM stays involved through the process
  5. CSM owns onboarding of expanded product/seats after close

Common handoff failures

Expansion Conversation Templates

Seed Planting (Phase 1)

"I was reviewing your usage this quarter and noticed your team has been
really active with [feature]. Some of our customers who use [feature]
heavily have also found [adjacent feature/tier] helpful for [specific
outcome]. Just wanted to flag it as something to keep in mind as your
needs evolve."

Discovery (Phase 2)

"In our last QBR, you mentioned [business initiative]. I've been thinking
about how we might help with that. A few questions:
- What's the timeline for that initiative?
- Who else on your team is involved?
- Have you evaluated any solutions for that yet?
- Would it be helpful if I shared how [similar customer] approached this?"

Value Proposal (Phase 3)

"Based on our conversations, here's what I'm recommending:

Current state: [what they have today]
Desired state: [what they want to achieve]
Gap: [what's missing]
Proposed solution: [specific expansion - seats, tier, module]
Expected outcome: [quantified value - time saved, revenue generated, etc.]
Investment: [pricing, positioned against value]

I'd like to bring in [AE name], who specializes in [area], to walk
through the details and answer any technical questions. Would [date]
work for a 30-minute call?"

Expansion Metrics

Track these metrics to measure expansion program effectiveness:

Metric                          | Target         | Measurement
--------------------------------|----------------|-----------------------------------
Expansion revenue rate          | >20% of total  | Expansion ARR / Total new ARR
Net Revenue Retention (NRR)     | >115%          | Including expansion, contra, churn
Expansion win rate              | >40%           | Proposals won / proposals made
Time from signal to proposal    | <30 days       | Average days from hot signal to AE intro
CSM-sourced pipeline            | >30% of expand | Pipeline originated by CS team
Expansion cycle time            | <45 days       | Average days from proposal to close
Post-expansion health score     | Stable/up      | Health score 90 days after expansion

Cross-Sell vs. Upsell Strategy

Upsell (same product, higher tier):
  - Lower friction, customer already knows the product
  - Focus on feature limits and usage growth
  - CSM can often handle without AE for small upgrades
  - Typical timing: 6-12 months into contract

Cross-Sell (new product from portfolio):
  - Higher friction, requires new evaluation and champion
  - Focus on adjacent use cases and new stakeholder needs
  - Almost always requires AE involvement
  - Typical timing: 12+ months, after strong first product adoption
  - Requires identifying new buyer/champion for the new product

Key principle: Upsell the current champion; cross-sell requires a new champion
in a different department or function.

Health Score Model

A customer health score is a weighted composite metric (0-100) that aggregates multiple behavioral and relational signals into a single risk/opportunity indicator. A well-designed model predicts renewal probability and triggers CSM actions before churn becomes irreversible.

Design principles

• Fewer inputs, higher signal. Start with 4-6 dimensions. Models with 20+ inputs dilute strong predictors and create noise. Add dimensions only when they demonstrably improve predictive accuracy.
• Automate data collection. A dimension you can't populate automatically will drift stale and corrupt the score. If a signal requires manual CSM entry, weight it lower or exclude it from the automated score.
• Every band triggers an action. If a score change doesn't trigger a defined CSM workflow, the score is decorative. Define mandatory actions per band before launch.
• Calibrate continuously. Initial weights are hypotheses. Validate against actual churn and renewal data every quarter and adjust.

Dimensions

1. Product Usage (default weight: 35%)

Measures whether the customer is actively and broadly using the product.

Normalization example (login frequency):

Composite usage sub-score: usage_score = (login_freq * 0.40) + (feature_breadth * 0.30) + (usage_depth * 0.20) + (trend * 0.10)

2. Engagement (default weight: 25%)

Measures the quality and consistency of the relationship between the customer and the CS team.

Normalization example (touchpoint cadence):

3. Outcomes (default weight: 20%)

Measures whether the customer is achieving the goals documented in the success plan. This is the hardest dimension to automate but the most predictive of genuine retention.

Scoring approach: Because outcome data is partially subjective, weight CSM-assessed milestone completion (60%) and objective NPS/CSAT (40%) to reduce gaming.

outcomes_score = (milestone_pct * 0.60) + (normalized_nps * 0.40)

4. Support (default weight: 10%)

Measures product quality experience and service burden signals.

Composite support sub-score: support_score = (100 if no_open_critical else 0) * 0.40 + (normalized_csat * 0.40) + (volume_trend_score * 0.20)

Map ticket volume trend: declining = 100, flat = 70, increasing = 40, spike = 0.

5. Relationship (default weight: 10%)

Measures stakeholder stability and strategic alignment signals.

Champion departure is a special-case trigger: regardless of composite score, a champion departure should immediately flag the account for Watch status and CSM outreach within 24 hours.

Composite score formula

health_score =
  (usage_score   * 0.35) +
  (engagement_score * 0.25) +
  (outcomes_score   * 0.20) +
  (support_score    * 0.10) +
  (relationship_score * 0.10)

Round to nearest integer. The result is a value from 0 to 100.

Thresholds and action triggers

Threshold override rules

These conditions force Red regardless of composite score:

• Any open P1 ticket unresolved for >5 business days
• Primary champion departed without replacement confirmed
• Customer has sent formal cancellation intent or legal notice
• Payment overdue >30 days

Calibration methodology

Step 1 - Collect baseline data

For each account that churned or renewed in the last 4 quarters, pull the health score from 90 days before the renewal date. This is your training set.

Step 2 - Measure predictive accuracy

Calculate the confusion matrix:

True Positive  = Red score at T-90 AND customer churned
False Negative = Green/Yellow at T-90 AND customer churned  (missed churns)
True Negative  = Green score at T-90 AND customer renewed
False Positive = Red at T-90 AND customer renewed           (false alarms)

Target: False Negative rate <25% (you should catch 75%+ of churns in the Red band).

Step 3 - Adjust weights

If False Negatives are high, the model is under-weighting the signals that churned accounts showed. Analyze what those accounts had in common at T-90 and increase the weight of those dimensions.

If False Positives are high, the model is over-penalizing for signals that don't actually predict churn. Reduce weights for those dimensions or adjust the normalization thresholds.

Step 4 - Validate, then recalibrate quarterly

Never run more than one quarter on uncalibrated weights. Set a calendar reminder to re-run the confusion matrix after every renewal cycle.

Common model pitfalls

Velocity scoring (optional enhancement)

Layer a velocity multiplier onto the composite score to capture directional momentum:

30-day delta = current_score - score_30_days_ago

Velocity bucket:
  delta > +5   : improving  -> multiplier 0.85 (reduces effective risk)
  -5 to +5     : stable     -> multiplier 1.00
  -15 to -5    : declining  -> multiplier 1.25
  delta < -15  : rapid drop -> multiplier 1.50 (auto-escalate to Warning)

risk_adjusted_score = health_score / velocity_multiplier

Use risk_adjusted_score for triage prioritization but display health_score to customers and in reporting to avoid confusion.

Health Score Models - Complete Reference

Signal Selection Methodology

Start with the outcome you're predicting (churn vs. expansion vs. advocacy), then work backward to identify the signals that correlate with that outcome.

Step 1: Gather candidate signals

Audit every data source available to CS:

Product Analytics:
  - DAU/MAU ratio (stickiness)
  - Feature adoption breadth (# of paid features used / total available)
  - Core action volume (the "aha moment" action, measured per user per week)
  - Session duration and depth
  - API call volume (for developer products)
  - Integration count (connected third-party tools)

Engagement Data:
  - CSM meeting attendance rate (attended / scheduled)
  - Support ticket volume and sentiment
  - Email/message response time from customer
  - Event and webinar attendance
  - Community forum participation
  - Training/certification completion

Outcome Data:
  - Progress toward stated success criteria
  - Customer-reported satisfaction (NPS, CSAT, CES)
  - ROI metrics (if measurable and tracked)
  - Business outcome indicators (e.g., revenue influenced, time saved)

Contractual Data:
  - Days to renewal
  - Contract value trend (expanding, flat, contracting)
  - Payment history (on-time %, overdue invoices)
  - Multi-year vs. annual vs. monthly contract
  - Number of products/modules purchased

Step 2: Validate against historical outcomes

For each candidate signal, run a correlation analysis against actual churn and expansion events from the past 12-24 months.

Validation approach:

1. Pull churned accounts from past 12 months
2. Pull expanded accounts from past 12 months
3. Pull stable (retained, no change) accounts as control group
4. For each signal, compare distributions across the three groups
5. Keep signals where churned accounts show statistically different
   values from retained/expanded accounts
6. Discard signals that look the same across all groups

Step 3: Assign weights

Initial weight recommendations by business model:

B2B SaaS (Sales-Led):
  Product Adoption:  35%
  Engagement:        30%  (CSM relationship matters heavily)
  Outcomes:          20%
  Contractual:       15%

B2B SaaS (Product-Led Growth):
  Product Adoption:  45%  (usage IS the relationship)
  Engagement:        15%
  Outcomes:          25%
  Contractual:       15%

Enterprise / High ACV:
  Product Adoption:  25%
  Engagement:        35%  (multi-threading, exec sponsors critical)
  Outcomes:          25%
  Contractual:       15%

These are starting points. After 2-3 quarters of data, use logistic regression or a simple decision tree on your actual churn data to derive empirically optimal weights.

Scoring Implementation

Normalize each signal to 0-100

Every raw signal must be normalized before applying weights.

Common normalization approaches:

Percentile-based (recommended for most signals):
  Score = Percentile rank of this account among all active accounts
  Example: If account's DAU/MAU is in the 75th percentile, score = 75

Threshold-based (good for binary or categorical signals):
  Score = 100 if condition met, 50 if partially met, 0 if not met
  Example: Executive sponsor identified and engaged = 100,
           identified but unresponsive = 50, none identified = 0

Trend-based (for signals where direction matters):
  Score = 50 (baseline) + (trend_direction * magnitude_adjustment)
  Example: Login frequency up 20% MoM = 50 + 20 = 70
           Login frequency down 30% MoM = 50 - 30 = 20

Composite score calculation

Health Score = SUM(signal_score_i * weight_i) for all signals

Example:
  Product Adoption score: 72, weight: 0.40 -> 28.8
  Engagement score:       85, weight: 0.25 -> 21.25
  Outcomes score:         60, weight: 0.20 -> 12.0
  Contractual score:      90, weight: 0.15 -> 13.5
  -----------------------------------------------
  Composite Health Score: 75.55 -> rounds to 76 (Green)

Threshold Calibration

Initial thresholds (start here, then calibrate)

Score Range | Label  | Action
------------|--------|-------------------------------------------
85-100      | Strong | Expansion opportunity - connect with AE
70-84       | Good   | Monitor, continue driving deeper adoption
50-69       | Fair   | Proactive outreach, investigate weak signals
30-49       | Poor   | Urgent intervention, CSM + manager review
0-29        | Crisis | Executive escalation, recovery playbook

Calibration process (run quarterly)

1. Pull all accounts that churned last quarter
2. Record their health score 90 days before churn
3. Calculate the distribution:
   - What % of churned accounts were scored Green?  (false negatives)
   - What % of Green accounts actually churned?     (false negative rate)
   - What % of Red accounts did NOT churn?           (false positives)

4. Adjust thresholds to minimize false negatives:
   - If many churned accounts were Green, lower the Green threshold
   - If many Red accounts are stable, raise the Red threshold
   - Target: <10% of churned accounts should have been Green 90 days prior

5. Revalidate signal weights:
   - Which signals had the strongest predictive power this quarter?
   - Which signals added noise (no correlation with outcomes)?
   - Adjust weights accordingly, maximum 10% shift per quarter

Do not change thresholds or weights more than once per quarter. Frequent changes make the score unreliable for CSMs who learn to calibrate their intuition against it. Stability builds trust.

Health Score by Customer Lifecycle Stage

Not all signals matter equally at every stage.

Onboarding (0-90 days):
  Overweight: Time-to-first-value, onboarding milestone completion,
              stakeholder identification, training attendance
  Underweight: NPS (too early), expansion signals, renewal proximity

Adoption (90 days - 1 year):
  Overweight: Feature adoption breadth, core action frequency,
              integration depth, user growth within account
  Underweight: Renewal proximity (still far away)

Maturity (1+ year):
  Overweight: Outcome achievement, executive engagement, NPS trend,
              expansion signals, renewal proximity
  Underweight: Onboarding milestones (irrelevant)

Pre-Renewal (90 days before renewal):
  Overweight: Renewal signals (budget approval, legal review started),
              executive sentiment, competitive mentions in tickets
  Underweight: Feature adoption (too late to change)

Common Health Score Pitfalls

QBR Templates - Complete Reference

QBR Program Design

Cadence by Segment

Segment      | Cadence       | Format          | Duration  | Attendees
-------------|---------------|-----------------|-----------|------------------
Enterprise   | Quarterly     | In-person/video | 45-60 min | Exec + CSM + AE
Mid-Market   | Semi-annual   | Video call      | 30-45 min | CSM + stakeholders
SMB          | Annual        | Video call      | 20-30 min | CSM (pooled)
Self-Serve   | None (auto)   | Email report    | N/A       | Automated

QBR Readiness Checklist

Before scheduling the QBR, CSM must complete:

Data Preparation:
  [ ] Pull usage analytics for the quarter (DAU/MAU, feature adoption, volume)
  [ ] Calculate value delivered (ROI, time saved, outcomes achieved)
  [ ] Prepare benchmark comparisons (anonymized peer data)
  [ ] Review support ticket history and sentiment
  [ ] Check health score trend for the quarter
  [ ] Pull any NPS/CSAT responses from the period

Stakeholder Preparation:
  [ ] Confirm attendees 2 weeks in advance
  [ ] Identify if any stakeholder changes occurred
  [ ] Pre-align with champion on any sensitive topics
  [ ] Send pre-read agenda 1 week before
  [ ] Gather customer's priorities for the quarter ahead

Internal Preparation:
  [ ] Review previous QBR action items and status
  [ ] Align with AE on any expansion/renewal topics
  [ ] Prepare product roadmap highlights relevant to this customer
  [ ] Brief exec sponsor (if attending) on account context
  [ ] Rehearse key talking points and transitions

The Look Back / Look Around / Look Forward Framework

Look Back (33% of time)

Purpose: Demonstrate value delivered and build credibility.

Structure:
  1. Recap goals set in the previous QBR
  2. For each goal, show:
     - Status: Achieved / On Track / Behind / Missed
     - Evidence: data, metrics, or customer confirmation
     - If behind/missed: explain why and what was done
  3. Highlight unexpected wins (value delivered beyond stated goals)
  4. Acknowledge any service issues and remediation steps taken

Slide template:

| Goal                  | Target        | Actual        | Status    |
|-----------------------|---------------|---------------|-----------|
| Onboard 50 users      | 50 users      | 62 users      | Exceeded  |
| Reduce process time   | 30% reduction | 22% reduction | On Track  |
| Launch API integration | Q2 launch     | Delayed to Q3 | Behind    |

Key: Green = Exceeded/Achieved, Yellow = On Track, Red = Behind/Missed

Talking points:

• Lead with wins. Even if overall performance is mixed, start with what went well to establish positive momentum.
• For missed goals, own the gap honestly and focus on what was learned and what changes are being made.
• Use customer quotes or data from their own systems when possible - third-party validation is more credible than your dashboards.

Look Around (33% of time)

Purpose: Provide external context and position yourself as a strategic partner.

Structure:
  1. Industry benchmarks - how does this customer compare to peers?
  2. Best practices - what are top-performing customers doing differently?
  3. Product updates relevant to their use cases
  4. Market/industry trends that affect their business

Benchmark slide template:

| Metric              | This Customer | Peer Median | Top Quartile |
|---------------------|---------------|-------------|--------------|
| User adoption rate  | 72%           | 65%         | 85%          |
| Feature utilization | 45%           | 40%         | 70%          |
| Time to resolution  | 4.2 hours     | 6.1 hours   | 2.8 hours    |
| Support tickets/mo  | 12            | 18          | 8            |

Rules for benchmarking:

• Always anonymize peer data. Never reveal other customers' identities.
• Position benchmarks as aspirational, not critical. "Here's where you are vs. where the best performers are" not "you're below average."
• Only benchmark metrics where you have statistically meaningful data (minimum 20+ accounts in the comparison set).
• If the customer is already top-quartile, highlight it. Nothing builds confidence like knowing they're leading their peers.

Look Forward (33% of time)

Purpose: Co-create the next quarter's success plan and secure commitment.

Structure:
  1. Ask about the customer's business priorities for next quarter
     (do NOT assume - let them tell you)
  2. Propose 2-3 measurable goals tied to their priorities
  3. Define mutual commitments:
     - What your team will deliver (training, features, support)
     - What their team needs to do (adoption, feedback, stakeholder access)
  4. Set milestones and check-in cadence
  5. If appropriate: discuss renewal timeline or expansion opportunity

Success plan template:

| Goal                        | Metric          | Target     | Owner     | Deadline |
|-----------------------------|-----------------|------------|-----------|----------|
| Expand to marketing team    | Active users    | +25 users  | Customer  | End Q3   |
| Launch automated workflows  | Workflows built | 10 active  | Joint     | Mid Q3   |
| Achieve 40% time savings    | Process audit   | 40% faster | Customer  | End Q3   |

QBR Slide Deck Template

Slide-by-Slide Guide

Slide 1: Title Slide
  - Customer logo + your logo
  - "Quarterly Business Review - Q[X] [Year]"
  - Date and attendees
  - Confidential marking

Slide 2: Agenda
  - Look Back: Q[X-1] in Review
  - Look Around: Benchmarks & Best Practices
  - Look Forward: Q[X] Success Plan
  - Discussion & Next Steps
  - Time allocation for each section

Slide 3: Partnership Summary
  - Customer since: [date]
  - Current plan/tier: [details]
  - Key milestones in the relationship
  - Current health score (if shared with customer)
  - CSM and support contacts

Slide 4: Goals vs. Actuals (Look Back)
  - Table format with RAG status per goal
  - Include data visualization for key metrics
  - Narrative for each goal: what happened and why

Slide 5: Usage & Adoption Highlights (Look Back)
  - 2-3 charts showing usage trends
  - Active users over time
  - Feature adoption heatmap
  - Key workflow completion rates

Slide 6: Value Delivered (Look Back)
  - Quantified ROI if available
  - Time saved, efficiency gained, revenue impacted
  - Customer quotes or testimonials if available
  - Before/after comparison

Slide 7: Benchmarks (Look Around)
  - Comparison to anonymized peer group
  - 3-4 metrics where comparison is meaningful
  - Highlight areas of strength and opportunity

Slide 8: Best Practices & Product Updates (Look Around)
  - 2-3 tips from top-performing customers
  - Relevant product releases from the quarter
  - Upcoming features on the roadmap (relevant ones only)

Slide 9: Proposed Q[X] Goals (Look Forward)
  - 2-3 measurable goals
  - Tied to customer's stated business priorities
  - Include metrics, targets, and owners

Slide 10: Success Plan & Action Items (Look Forward)
  - Detailed action items with owners and deadlines
  - Both customer and CS team commitments
  - Check-in schedule for the quarter
  - Escalation path if goals are at risk

Slide 11 (optional): Renewal / Expansion Discussion
  - Only include if timing is appropriate
  - Frame as "how to get more value" not "how to spend more"
  - Connect expansion to goals discussed in Look Forward

Executive QBR Variant

For C-level attendees, modify the standard QBR:

Changes from standard QBR:
  - Cut to 30 minutes maximum (executives have less patience)
  - Lead with business outcomes, not product usage
  - One slide on ROI/value delivered (the only slide that matters to them)
  - Strategic alignment: "Here's how our partnership connects to your
    company's top 3 priorities this year"
  - Skip feature-level details entirely
  - End with one clear ask or decision point

Executive QBR Slide Structure (5 slides max):
  1. Executive Summary (1 slide, 3-4 bullets)
  2. Business Outcomes Delivered (1 slide, quantified)
  3. Strategic Alignment (1 slide, connecting product to company priorities)
  4. Next Quarter Plan (1 slide, 2-3 goals)
  5. Discussion / Decision Point (1 slide, one clear ask)

Post-QBR Follow-Up

Timeline | Action
---------|----------------------------------------------------------
Day 0    | Send thank-you email with slide deck attached
Day 1    | Send QBR summary email with action items, owners, deadlines
Day 1    | Update CRM: log QBR, update health score, note next steps
Day 7    | Begin executing on your committed action items
Day 14   | First progress check-in on action items
Day 30   | Mid-quarter check: are we on track for QBR goals?
Day 60   | Pre-QBR data collection begins for next quarter
Day 75   | Schedule next QBR, send pre-read materials

QBR Summary Email Template

Subject: [Customer Name] - Q[X] QBR Summary & Action Items

Hi [Name],

Thank you for a productive QBR today. Here's a summary of what we covered
and our commitments for Q[X]:

**Key Highlights from Q[X-1]:**
- [Achievement 1 with metric]
- [Achievement 2 with metric]
- [Area for improvement and plan]

**Q[X] Goals:**
1. [Goal] - Target: [metric] - Owner: [name] - By: [date]
2. [Goal] - Target: [metric] - Owner: [name] - By: [date]
3. [Goal] - Target: [metric] - Owner: [name] - By: [date]

**Action Items:**
| Action | Owner | Due Date |
|--------|-------|----------|
| [item] | [who] | [when]   |

Our next check-in is scheduled for [date]. Please let me know if anything
changes on your end or if you'd like to adjust our goals.

Best,
[CSM Name]

QBR Effectiveness Metrics

Metric                          | Target    | Measurement
--------------------------------|-----------|----------------------------------
QBR completion rate             | >85%      | QBRs held / QBRs scheduled
Executive attendance rate       | >50%      | QBRs with exec present / total
Action item completion rate     | >75%      | Items completed by next QBR
Goal achievement rate           | >60%      | Goals achieved / goals set
Post-QBR health score change    | Positive  | Health score 30 days post-QBR
Post-QBR NPS/CSAT               | >8        | Survey sent within 48 hours
Expansion sourced from QBRs     | >25%      | Expansion deals with QBR origin
Renewal rate for QBR accounts   | >90%      | Renewal rate for accounts with QBRs