dojo.md 0.1.0

package/courses/customer-churn-prevention/scenarios/level-5/customer-success-transformation.yaml

@@ -0,0 +1,83 @@
+meta:
+  id: customer-success-transformation
+  level: 5
+  course: customer-churn-prevention
+  type: output
+  description: "Lead a company-wide transformation from reactive customer support to proactive customer success — changing the operating model, culture, and business outcomes"
+  tags: [churn, transformation, operating-model, culture, master]
+
+state: {}
+
+trigger: |
+  You're the incoming CCO at "TransformCo" ($500M ARR, 40,000
+  customers, 200 CS + Support staff). The CEO hired you because:
+  "We have a support team that puts out fires. I want a success
+  team that prevents them."
+
+  Current state assessment:
+
+  Operating model: Reactive
+  - 80% of CS time is spent on support tickets
+  - 15% on renewals (last-minute)
+  - 5% on proactive engagement
+  - No customer journey mapping
+  - No value realization tracking
+  - Customer engagement: only when customers reach out
+
+  Team structure:
+  - 150 support agents (tier 1-3)
+  - 40 CSMs (generalists, fire-fighters)
+  - 10 renewals specialists (order takers)
+  - No CS operations, no analytics, no enablement
+
+  Metrics:
+  - Support CSAT: 3.8/5.0 (declining)
+  - First response time: 4 hours (target: 1 hour)
+  - Resolution time: 3 days (target: 1 day)
+  - GRR: 82% (industry: 90%)
+  - NRR: 95% (industry: 115%)
+  - No health scores, no churn prediction, no lifecycle tracking
+
+  Target state (in 2 years):
+  - Proactive CS model with 70% of CS time spent on prevention
+  - GRR: 92%
+  - NRR: 118%
+  - Support automated with AI for tier 1 (60% deflection)
+  - Health scoring and churn prediction for all accounts
+  - Customer journey mapped and measured
+
+  Constraints:
+  - Cannot increase total headcount by more than 10%
+  - Must maintain support quality during transformation
+  - Union-like resistance from support agents ("Are you replacing us?")
+
+  Tasks:
+  1. Design the 2-year transformation roadmap with quarterly
+     milestones.
+
+  2. How do you restructure 200 people without mass layoffs? What
+     roles change, who gets retrained, and what new roles are created?
+
+  3. Address the "are you replacing us?" concern from support agents.
+     How do you bring the existing team along?
+
+  4. Measure the transformation: What metrics prove you're moving
+     from reactive to proactive, and what are the quarterly targets?
+
+assertions:
+  - type: llm_judge
+    criteria: "Roadmap is phased with clear milestones — progresses from quick wins to structural changes to culture transformation with specific quarterly deliverables"
+    weight: 0.25
+    description: "Milestone-clear transformation roadmap"
+  - type: llm_judge
+    criteria: "Restructuring preserves people — retrains support agents into proactive roles (onboarding, adoption, expansion) rather than eliminating positions"
+    weight: 0.25
+    description: "People-preserving restructuring"
+  - type: llm_judge
+    criteria: "Change management addresses fear directly — communicates the vision, shows how roles evolve upward (not downward), and provides training and support"
+    weight: 0.25
+    description: "Fear-addressing change management"
+  - type: llm_judge
+    criteria: "Transformation metrics are leading — includes proactive engagement rate, health score accuracy, time-to-intervention, and intervention save rate alongside GRR/NRR"
+    weight: 0.25
+    description: "Leading transformation metrics"

package/courses/customer-churn-prevention/scenarios/level-5/industry-specific-retention.yaml

@@ -0,0 +1,77 @@
+meta:
+  id: industry-specific-retention
+  level: 5
+  course: customer-churn-prevention
+  type: output
+  description: "Design industry-specific retention strategies for healthcare, financial services, e-commerce, and education SaaS — where churn drivers and solutions differ fundamentally"
+  tags: [churn, industry-specific, vertical, strategy, master]
+
+state: {}
+
+trigger: |
+  You're a retention strategy consultant advising "VerticalSaaS"
+  ($350M ARR), which serves 4 distinct vertical markets. Each
+  vertical has unique churn dynamics.
+
+  Healthcare SaaS ($100M ARR, 2,000 customers):
+  - Annual churn: 8% (low, due to integration depth and compliance)
+  - NRR: 118%
+  - Top risk: Regulatory changes (HIPAA updates, CMS requirements)
+  - Retention driver: EHR integration depth — takes 6 months to
+    implement, creating massive switching costs
+  - Churn trigger: Government contract cycles (3-year terms)
+
+  Financial Services SaaS ($120M ARR, 3,000 customers):
+  - Annual churn: 12%
+  - NRR: 110%
+  - Top risk: Security incidents and compliance failures
+  - Retention driver: Data-in-place gravity (10 years of financial
+    records in your platform)
+  - Churn trigger: M&A consolidation (acquirer dictates tools)
+
+  E-commerce SaaS ($80M ARR, 8,000 customers):
+  - Annual churn: 22%
+  - NRR: 95%
+  - Top risk: Seasonal businesses cancel after peak season
+  - Retention driver: Revenue attribution (proving ROI)
+  - Churn trigger: Monthly billing cycle, low switching costs
+
+  Education SaaS ($50M ARR, 5,000 customers):
+  - Annual churn: 30%
+  - NRR: 85%
+  - Top risk: Budget cycles (academic year, grant expiration)
+  - Retention driver: Student/teacher adoption (hard to retrain)
+  - Churn trigger: Summer break cancellations, new administration
+
+  Tasks:
+  1. For each vertical, design the retention strategy that addresses
+     the unique churn drivers.
+
+  2. E-commerce churn is 22%. Can you get it below 15% without
+     adding annual contracts? Design the product-led retention
+     approach.
+
+  3. Education churn is 30% — the worst segment. Is it worth
+     investing in retention or should the company focus on other
+     verticals?
+
+  4. Design the universal retention framework that works across all
+     4 verticals while allowing vertical-specific customization.
+
+assertions:
+  - type: llm_judge
+    criteria: "Vertical strategies address unique drivers — healthcare gets compliance focus, fintech gets security and data gravity, e-commerce gets ROI proof, education gets budget cycle alignment"
+    weight: 0.25
+    description: "Unique vertical strategies"
+  - type: llm_judge
+    criteria: "E-commerce product-led approach is creative — uses engagement triggers, seasonal planning, and revenue attribution to keep customers engaged year-round"
+    weight: 0.25
+    description: "Creative e-commerce retention"
+  - type: llm_judge
+    criteria: "Education investment decision is analytical — evaluates the vertical's contribution margin, improvement potential, and opportunity cost vs. investing in higher-NRR verticals"
+    weight: 0.25
+    description: "Analytical education decision"
+  - type: llm_judge
+    criteria: "Universal framework has core + flex — defines common elements (health scoring, renewal process) while allowing vertical-specific customization (triggers, offers, timing)"
+    weight: 0.25
+    description: "Core-plus-flex universal framework"

package/courses/customer-churn-prevention/scenarios/level-5/negative-churn-strategy.yaml

@@ -0,0 +1,78 @@
+meta:
+  id: negative-churn-strategy
+  level: 5
+  course: customer-churn-prevention
+  type: output
+  description: "Achieve and sustain negative churn (NRR > 130%) through expansion, usage-based growth, and product-led retention at the highest level"
+  tags: [churn, negative-churn, NRR, expansion, master]
+
+state: {}
+
+trigger: |
+  You're the Chief Revenue Officer at "HyperGrowth" ($400M ARR,
+  publicly traded). The board's new mandate: Achieve 130% NRR (best
+  in class for public SaaS).
+
+  Current state:
+  - GRR: 91% (solid, but room to improve)
+  - NRR: 115% (good but not best-in-class)
+  - Expansion revenue: 24% of starting ARR
+  - Contraction: 4% of starting ARR
+
+  To reach 130% NRR, you need:
+  - Expansion to reach 39-43% (currently 24%)
+  - Churn + contraction to stay at or below 9-13%
+
+  Expansion levers:
+  1. Seat-based expansion: Add users as customers grow. Currently
+     generates 40% of expansion revenue.
+  2. Usage-based pricing: Charge by API calls/data volume. Not yet
+     implemented — would change the pricing model.
+  3. Cross-sell new products: 2 new product lines launched this year.
+     Adoption: 8% of existing customers.
+  4. Upgrade to higher tiers: Only 15% of customers are on the
+     highest tier.
+  5. Platform effect: Customers who use 3+ products have 140% NRR.
+     Only 12% of customers use 3+.
+
+  Companies with 130%+ NRR (benchmarks):
+  - Snowflake: Usage-based pricing drives natural expansion
+  - Twilio: Volume-based growth tied to customer success
+  - Datadog: Land-and-expand with 16+ products
+  - CrowdStrike: Module cross-sell and threat expansion
+
+  Constraints:
+  - Can't raise prices more than 5% annually (competitive pressure)
+  - New products aren't mature enough for aggressive cross-sell
+  - Sales team is quota'd on new logos, not expansion
+
+  Tasks:
+  1. Design the path to 130% NRR. Which expansion levers generate
+     the most value? Show the math for each lever.
+
+  2. Should you switch to usage-based pricing? Analyze the revenue
+     model change using the benchmark companies.
+
+  3. How do you get from 12% multi-product adoption to 40%? Design
+     the cross-sell program that feels helpful, not pushy.
+
+  4. The sales team isn't incentivized for expansion. Design the
+     comp restructure that aligns the entire revenue team.
+
+assertions:
+  - type: llm_judge
+    criteria: "NRR path is quantified per lever — shows how much each expansion lever contributes and which combination reaches 130%"
+    weight: 0.25
+    description: "Quantified NRR path per lever"
+  - type: llm_judge
+    criteria: "Usage-based pricing analysis is thorough — considers revenue predictability trade-offs, customer behavior changes, and lessons from benchmark companies"
+    weight: 0.25
+    description: "Thorough usage pricing analysis"
+  - type: llm_judge
+    criteria: "Cross-sell program is customer-centric — uses product-led discovery, usage signals, and value-based recommendations rather than aggressive sales pushes"
+    weight: 0.25
+    description: "Customer-centric cross-sell"
+  - type: llm_judge
+    criteria: "Comp restructure aligns on expansion — proposes specific changes (expansion quota, NRR bonuses, CS-sales co-sell) that incentivize the right behaviors"
+    weight: 0.25
+    description: "Expansion-aligned comp restructure"

package/courses/customer-churn-prevention/scenarios/level-5/retention-automation-platform.yaml

@@ -0,0 +1,76 @@
+meta:
+  id: retention-automation-platform
+  level: 5
+  course: customer-churn-prevention
+  type: output
+  description: "Design a comprehensive retention automation platform — from data ingestion to predictive scoring to automated intervention to measurement"
+  tags: [churn, automation, platform, end-to-end, master]
+
+state: {}
+
+trigger: |
+  You're the VP of CS Operations at "AutoRetain" ($250M ARR,
+  20,000 customers). You've been tasked with building the internal
+  "Retention Engine" — a platform that automates the entire
+  retention lifecycle.
+
+  Current manual processes:
+  1. Health scoring: CSMs manually review accounts weekly (100+
+     hours/month across the team)
+  2. Risk identification: Monthly report from analytics team
+     (2-week lag between signal and action)
+  3. Intervention selection: CSMs choose their own approach (no
+     consistency, no optimization)
+  4. Outreach execution: Manual emails and calls
+  5. Offer management: CSMs negotiate individually (no standardized
+     playbooks)
+  6. Measurement: Quarterly retention review (too infrequent)
+
+  Platform requirements:
+  - Real-time health scoring (updated daily, not monthly)
+  - Predictive churn alerts (60+ day lead time)
+  - Automated intervention triggers (email, in-app, CSM assignment)
+  - Dynamic offer engine (right offer for right customer at right time)
+  - A/B testing framework for retention interventions
+  - ROI tracking per intervention type
+  - Integration with: Salesforce, Gainsight, Intercom, Stripe,
+    Snowflake data warehouse
+
+  Constraints:
+  - 20,000 customers across 4 segments (enterprise, mid-market,
+    SMB, self-serve)
+  - Budget: $1.5M Year 1, $500K/year ongoing
+  - Must show ROI within 6 months
+  - Engineering team: 4 engineers allocated
+
+  Tasks:
+  1. Design the platform architecture. What are the core systems,
+     data flows, and integration points?
+
+  2. Design the dynamic offer engine. How does it determine the
+     right retention offer for each customer automatically?
+
+  3. Build the A/B testing framework for retention. How do you
+     test interventions ethically (some customers get no
+     intervention)?
+
+  4. Calculate the expected ROI. If the platform improves save
+     rates from 20% to 35%, what's the revenue impact?
+
+assertions:
+  - type: llm_judge
+    criteria: "Platform architecture is well-designed — identifies data ingestion, scoring engine, intervention router, offer engine, and measurement system as core components"
+    weight: 0.25
+    description: "Well-designed platform architecture"
+  - type: llm_judge
+    criteria: "Offer engine uses multi-factor routing — considers customer segment, risk level, churn reason, historical response to offers, and financial constraints"
+    weight: 0.25
+    description: "Multi-factor offer engine"
+  - type: llm_judge
+    criteria: "A/B testing framework is ethical — addresses the control group concern by using holdout percentages, minimum viable sample sizes, and escalation rules"
+    weight: 0.25
+    description: "Ethical A/B testing framework"
+  - type: llm_judge
+    criteria: "ROI calculation is credible — quantifies revenue saved from improved save rates, factors in platform cost, and shows clear payback timeline"
+    weight: 0.25
+    description: "Credible ROI calculation"

package/courses/customer-churn-prevention/scenarios/level-5/retention-consulting-engagement.yaml

@@ -0,0 +1,73 @@
+meta:
+  id: retention-consulting-engagement
+  level: 5
+  course: customer-churn-prevention
+  type: output
+  description: "Act as a retention strategy consultant assessing a struggling company's churn problems and designing a transformation program"
+  tags: [churn, consulting, strategy, assessment, master]
+
+state: {}
+
+trigger: |
+  You're a world-class retention consultant. "CrisisCloud" ($300M
+  ARR, 20,000 customers) hires you after their annual churn hit 28%
+  — the worst in their industry. They've lost $84M in ARR this year.
+
+  Discovery findings:
+
+  Organizational:
+  - CS reports to Sales VP (conflict of interest)
+  - No CCO or dedicated retention leadership
+  - CS team of 15 manages 20,000 accounts (1:1,333 ratio)
+  - CS team turnover: 40% annually
+  - No career path for CS professionals
+
+  Process:
+  - No health scoring (CSMs use "gut feeling")
+  - Renewal process starts 30 days before expiry
+  - No save playbooks or authorized retention offers
+  - Exit interviews: done for <5% of churned customers
+  - No dunning optimization (5% of churn is involuntary)
+
+  Product:
+  - Time-to-value: 45 days (industry: 14 days)
+  - Feature adoption: average customer uses 18% of features
+  - Product NPS: +15 (industry: +40)
+  - Top churn reason: "Didn't realize the value" (35%)
+
+  Financial:
+  - CAC: $8,000
+  - LTV: $16,000 (should be $40,000+ at industry retention rates)
+  - LTV/CAC: 2x (should be 5x+)
+  - Total cost of churn (credits + replacement): $126M/year
+
+  Tasks:
+  1. Present your 90-day assessment. What are the top 10 findings
+     ranked by impact and urgency?
+
+  2. Design the 18-month transformation roadmap. Phase 1 quick wins,
+     Phase 2 structural changes, Phase 3 optimization.
+
+  3. The CEO says: "We can't afford to hire more CS people." Design
+     the retention improvement plan within the existing headcount.
+
+  4. Project the financial impact: If you reduce churn from 28% to
+     15% over 18 months, what's the ARR and LTV impact?
+
+assertions:
+  - type: llm_judge
+    criteria: "Assessment is prioritized by impact — ranks findings by revenue impact and urgency, identifying the highest-ROI changes first"
+    weight: 0.25
+    description: "Impact-prioritized assessment"
+  - type: llm_judge
+    criteria: "Roadmap is phased with quick wins — Phase 1 delivers measurable improvement in 90 days (dunning, renewal process, basic health scoring)"
+    weight: 0.25
+    description: "Quick-win phased roadmap"
+  - type: llm_judge
+    criteria: "Headcount-constrained plan uses automation — proposes tech-touch, automated triggers, and prioritized human intervention to maximize impact per CSM"
+    weight: 0.25
+    description: "Automation-leveraged constrained plan"
+  - type: llm_judge
+    criteria: "Financial projection is specific — calculates ARR saved, LTV improvement, and LTV/CAC ratio change with year-by-year projections"
+    weight: 0.25
+    description: "Specific financial projection"

package/courses/customer-churn-prevention/scenarios/level-5/retention-economics-masterclass.yaml

@@ -0,0 +1,75 @@
+meta:
+  id: retention-economics-masterclass
+  level: 5
+  course: customer-churn-prevention
+  type: output
+  description: "Master the economics of customer retention — modeling LTV, quantifying retention investments, and connecting churn to company valuation"
+  tags: [churn, economics, LTV, valuation, master]
+
+state: {}
+
+trigger: |
+  You're the Chief Strategy Officer at "ValueSaaS" (pre-IPO, $500M
+  ARR, 30,000 customers). The board is preparing for an IPO and
+  retention economics are under intense scrutiny from investment
+  bankers.
+
+  Current metrics:
+  - GRR: 88% (public SaaS median: 90%)
+  - NRR: 112% (public SaaS median: 115%)
+  - Logo retention: 82%
+  - Average contract value: $16.7K
+  - CAC: $22K
+  - Payback period: 18 months
+  - Gross margin: 75%
+  - LTV (current): $52K
+  - LTV/CAC: 2.4x (target for IPO: 4x+)
+
+  Investment banker questions:
+  1. "Your GRR is below median. How does that affect your valuation
+     multiple?"
+  2. "What's the sensitivity analysis? If GRR drops to 85%, what
+     happens to your LTV and growth model?"
+  3. "Your NRR of 112% masks underlying churn. What's your gross
+     churn rate and how do you plan to improve it?"
+  4. "Companies with 120%+ NRR trade at 15x revenue. You're at 112%.
+     What's the valuation gap?"
+
+  Modeling challenge:
+  The board wants you to model 3 scenarios over 5 years:
+  - Scenario A (Status quo): GRR stays at 88%, NRR at 112%
+  - Scenario B (Moderate improvement): GRR to 92%, NRR to 118%
+  - Scenario C (Best in class): GRR to 95%, NRR to 125%
+
+  Assume: 25% annual new ARR growth, same CAC, 75% margin.
+
+  Tasks:
+  1. Answer each investment banker question with data-backed
+     responses.
+
+  2. Model the 3 scenarios over 5 years. Show year-by-year ARR,
+     LTV, LTV/CAC, and estimated valuation multiple.
+
+  3. What investments are needed to move from Scenario A to Scenario
+     B? From B to C? What's the marginal cost of each improvement?
+
+  4. Make the case for investing $20M in retention before the IPO.
+     Connect it directly to IPO valuation.
+
+assertions:
+  - type: llm_judge
+    criteria: "Investment banker answers are financially sophisticated — uses correct SaaS metrics language, cites comparables, and addresses concerns directly"
+    weight: 0.25
+    description: "Financially sophisticated answers"
+  - type: llm_judge
+    criteria: "5-year model is coherent — shows compounding effects of retention improvement on ARR, LTV, and unit economics with consistent assumptions"
+    weight: 0.25
+    description: "Coherent 5-year model"
+  - type: llm_judge
+    criteria: "Marginal cost analysis is realistic — shows diminishing returns as you move from moderate to best-in-class, identifying the efficient frontier"
+    weight: 0.25
+    description: "Realistic marginal cost analysis"
+  - type: llm_judge
+    criteria: "Pre-IPO investment case is compelling — quantifies the valuation uplift from improved NRR in terms of revenue multiple expansion"
+    weight: 0.25
+    description: "Compelling pre-IPO investment case"

package/courses/customer-onboarding-emails/course.yaml

@@ -0,0 +1,12 @@
+id: customer-onboarding-emails
+name: "Customer Onboarding Email Sequences"
+description: >
+  Master customer onboarding email sequences — from writing compelling
+  welcome emails and activation nudges to designing behavior-triggered
+  drip campaigns, segmented sequences, trial-to-paid conversion flows,
+  re-engagement series, multi-channel onboarding orchestration, and
+  measuring the impact of onboarding communication on activation,
+  retention, and lifetime value.
+levels: 5
+scenarios_per_level: 10
+tags: [customer-success, onboarding, email, sequences, activation, retention]

package/courses/customer-onboarding-emails/scenarios/level-1/activation-emails.yaml

@@ -0,0 +1,87 @@
+meta:
+  id: activation-emails
+  level: 1
+  course: customer-onboarding-emails
+  type: output
+  description: "Write activation emails that drive users to complete key setup steps — guide new users through account configuration, first actions, and initial value delivery"
+  tags: [onboarding, email, activation, setup, beginner]
+
+state: {}
+
+trigger: |
+  Activation emails are the bridge between signup and value. If
+  users don't complete key setup steps, they'll never experience
+  your product's value and will churn.
+
+  Part 1: Activation email fundamentals
+
+  Activation emails should:
+  1. Focus on ONE setup step per email
+  2. Show the benefit of completing the step (not just the how)
+  3. Make the step feel quick and easy
+  4. Include visual guidance if possible
+  5. Create urgency without being pushy
+
+  Part 2: Write activation emails
+
+  "DataDash" — an analytics dashboard tool:
+
+  Setup steps (in order of importance):
+  1. Connect a data source (most critical — no data = no value)
+  2. Create first dashboard
+  3. Invite a teammate
+  4. Set up automated reports
+
+  Write activation emails for each setup step:
+
+  Email A: "Connect your first data source"
+  - User signed up yesterday but hasn't connected any data
+  - Subject line
+  - Body explaining WHY this step matters (not just how)
+  - Step-by-step guidance (keep it to 3 steps max)
+  - CTA: "Connect Your Data"
+  - Offer help if they're stuck
+
+  Email B: "Create your first dashboard"
+  - User connected data 2 days ago but hasn't made a dashboard
+  - Subject line
+  - Show what they're missing (preview of insights)
+  - Quick-start option (template dashboards)
+  - CTA: "Choose a Template"
+
+  Email C: "Invite your team"
+  - User has been active for 5 days, solo user
+  - Subject line
+  - Social proof: "Teams who collaborate see 3x more insights"
+  - Make it easy: just enter email addresses
+  - CTA: "Invite Teammates"
+
+  Part 3: Fix these activation emails
+
+  Bad activation email:
+  "Subject: Action Required: Complete Your Setup
+  Dear User, You have not yet completed the required setup steps
+  for your DataDash account. Please log in and complete the
+  following mandatory configuration: [list of 8 steps]. Failure
+  to complete setup within 7 days may result in limited
+  functionality. — The DataDash Team"
+
+  Identify every problem and rewrite it.
+
+assertions:
+  - type: llm_judge
+    criteria: "Activation emails lead with benefits — each email explains WHY the step matters (what value it unlocks) before HOW to do it"
+    weight: 0.30
+    description: "Benefit-led activation"
+  - type: llm_judge
+    criteria: "Each email focuses on exactly one action — the emails don't list multiple setup steps but guide toward a single next step"
+    weight: 0.25
+    description: "Single action focus"
+  - type: llm_judge
+    criteria: "The bad example rewrite addresses the threatening tone, overwhelming list, and missing value proposition"
+    weight: 0.25
+    description: "Correct bad example fix"
+  - type: llm_judge
+    criteria: "Emails offer help — each activation email acknowledges that users might be stuck and provides a path to support"
+    weight: 0.20
+    description: "Help availability"

package/courses/customer-onboarding-emails/scenarios/level-1/check-in-emails.yaml

@@ -0,0 +1,85 @@
+meta:
+  id: check-in-emails
+  level: 1
+  course: customer-onboarding-emails
+  type: output
+  description: "Write onboarding check-in emails — craft emails that genuinely ask how users are doing, surface blockers, and offer help without being pushy or automated-feeling"
+  tags: [onboarding, email, check-in, engagement, beginner]
+
+state: {}
+
+trigger: |
+  Check-in emails (typically Day 5-7) serve a dual purpose: they
+  make users feel cared for, and they surface problems before
+  users quietly churn. The best check-in emails feel genuinely
+  personal, not like automated messages.
+
+  Part 1: Check-in email approaches
+
+  Approach A: The helpful friend
+  "How's everything going with [Product]? I noticed you set up
+  [thing they did]. If you need any help with [next step], I'm
+  here."
+
+  Approach B: The quick question
+  "Quick question — what made you sign up for [Product]? Knowing
+  your goal helps me point you to the right features."
+
+  Approach C: The value reminder
+  "You've been using [Product] for a week. Here's what you've
+  accomplished so far: [summary]. Here's what's next."
+
+  Part 2: Write check-in emails
+
+  Scenario 1: Active user check-in
+  "AnalyticsPro" user signed up 7 days ago. They've:
+  - Connected 2 data sources
+  - Created 1 dashboard
+  - Logged in 4 times
+  Write a check-in email that celebrates their progress and
+  suggests the next level (sharing dashboards with team).
+
+  Scenario 2: Stalled user check-in
+  "AnalyticsPro" user signed up 7 days ago. They've:
+  - Logged in once (Day 1)
+  - Started data source connection but didn't complete it
+  - Haven't returned since
+  Write a check-in email that surfaces the blocker without
+  making the user feel guilty for not using the product.
+
+  Scenario 3: Power user check-in
+  "AnalyticsPro" user signed up 7 days ago. They've:
+  - Connected 5 data sources
+  - Created 8 dashboards
+  - Invited 3 teammates
+  - Logged in every day
+  Write a check-in email that matches their enthusiasm and
+  introduces advanced features they might not have found.
+
+  Part 3: From vs. reply-to
+
+  Should check-in emails come from:
+  a) "The [Product] Team" (generic)
+  b) "Sarah from [Product]" (named person)
+  c) "[CEO Name], Founder of [Product]" (founder)
+
+  For each option, explain when it works best and write the
+  appropriate sign-off.
+
+assertions:
+  - type: llm_judge
+    criteria: "Stalled user email is empathetic, not guilt-inducing — it acknowledges the incomplete setup without phrases like 'you haven't been back' and offers specific help for the blocker"
+    weight: 0.30
+    description: "Empathetic stalled user email"
+  - type: llm_judge
+    criteria: "Active and power user emails are differentiated — the active user gets encouragement to go deeper, while the power user gets advanced features and recognition"
+    weight: 0.25
+    description: "Differentiated user-level emails"
+  - type: llm_judge
+    criteria: "Check-in emails feel personal — they reference specific user actions and feel like they were written by a human who looked at the user's account"
+    weight: 0.25
+    description: "Genuinely personal check-ins"
+  - type: llm_judge
+    criteria: "From/reply-to analysis gives practical guidance — each option has a clear use case with reasoning, not just 'it depends'"
+    weight: 0.20
+    description: "Practical sender guidance"