dojo.md 0.1.0

package/courses/ecommerce-support-tickets/scenarios/level-2/live-chat-excellence.yaml

@@ -0,0 +1,107 @@
+meta:
+  id: live-chat-excellence
+  level: 2
+  course: ecommerce-support-tickets
+  type: output
+  description: "Excel at live chat support — manage real-time conversations, handle multiple chats simultaneously, and convert chat interactions into sales"
+  tags: [support, ecommerce, live-chat, real-time, conversion, intermediate]
+
+state: {}
+
+trigger: |
+  Live chat is the highest-converting support channel in e-commerce.
+  Customers who use live chat are 3x more likely to buy. But chat
+  requires different skills than email: real-time typing, managing
+  multiple conversations, and thinking on your feet. Response time
+  under 30 seconds is the expectation.
+
+  Part 1: Live chat best practices
+  1. Chat-specific skills:
+     - Short, scannable messages (2-3 sentences max)
+     - Break long answers into multiple messages
+     - Use "typing indicator" strategically (customer knows you're there)
+     - Personal greeting (name, offer to help)
+     - Proactive chat triggers (offer help before they ask)
+  2. Managing multiple chats:
+     - Ideal: 2-3 simultaneous chats
+     - Maximum effective: 4-5 (quality drops after)
+     - Priority: active conversation > waiting customer > idle
+     - Use macros for common openings while personalizing closings
+  3. Chat conversion techniques:
+     - Answering product questions instantly (no "let me check")
+     - Reducing checkout friction in real-time
+     - Suggesting complementary products naturally
+     - Creating urgency (limited stock, sale ending)
+     - Following up on abandoned chats
+
+  Part 2: Handle live chat scenarios
+  For "UrbanStyle" — an online streetwear brand with live chat:
+
+  Simulate 3 complete chat conversations (5-8 messages each):
+
+  1. Pre-sale conversion chat:
+     Customer browsing sneakers page for 3 minutes, chat proactively
+     triggered: "Hey! Need help finding the perfect pair?"
+     Customer is deciding between two sneakers, worried about sizing,
+     and wants to know about a sale. Convert them to purchase.
+     - Write the full chat transcript
+     - Include product recommendation and sizing help
+     - Close with a purchase or clear next step
+
+  2. Post-purchase issue chat (frustrated):
+     Customer initiates chat: "Why was my order cancelled without
+     telling me??"
+     Their order was cancelled due to a payment authorization failure.
+     They weren't notified (a system gap).
+     - De-escalate in real-time (chat de-escalation is harder)
+     - Fix the issue (reprocess the order)
+     - Acknowledge the notification failure
+
+  3. Multi-question chat:
+     Customer: "Hi quick questions — do you ship to UK? How long?
+     Can I return if it doesn't fit? Also is the hoodie on your
+     homepage still in stock in XL?"
+     - Handle multiple questions efficiently
+     - Answer all questions (don't miss any)
+     - Lead toward a purchase
+
+  Then handle 2 challenging chat situations:
+
+  4. Managing 3 simultaneous chats:
+     You have 3 active chats:
+     - Chat A: Complex return issue (needs investigation)
+     - Chat B: Simple WISMO question
+     - Chat C: Pre-sale sizing question (high purchase intent)
+     Show how you prioritize and manage response timing across
+     all three. Write key messages for each.
+
+  5. Chat-to-email handoff:
+     A customer's issue requires investigation that will take
+     20+ minutes. They can't stay on chat.
+     - Write the handoff message (don't lose the customer)
+     - Set clear expectations for email follow-up
+     - Capture all necessary information before they leave
+
+  Part 3: Chat optimization
+  1. Proactive chat trigger design (when to pop up, when not to)
+  2. Chat-to-conversion tracking and optimization
+  3. After-hours chat strategy (chatbot, email capture)
+  4. Measuring chat quality (response time, CSAT, conversion rate)
+
+assertions:
+  - type: llm_judge
+    criteria: "Chat conversations feel natural and real-time — messages are appropriately short, the pace feels like actual chat, not email responses in a chat window"
+    weight: 0.30
+    description: "Natural real-time chat feel"
+  - type: llm_judge
+    criteria: "The pre-sale chat successfully guides toward purchase — the conversation naturally leads to a buying decision through helpful information"
+    weight: 0.25
+    description: "Effective pre-sale chat conversion"
+  - type: llm_judge
+    criteria: "Multi-chat management demonstrates clear prioritization — the high-intent buyer gets faster responses while all three customers feel attended to"
+    weight: 0.25
+    description: "Effective multi-chat prioritization"
+  - type: llm_judge
+    criteria: "The chat-to-email handoff preserves the customer relationship — the transition feels seamless, not like being dumped"
+    weight: 0.20
+    description: "Seamless channel handoff"

package/courses/ecommerce-support-tickets/scenarios/level-2/marketplace-seller-support.yaml

@@ -0,0 +1,111 @@
+meta:
+  id: marketplace-seller-support
+  level: 2
+  course: ecommerce-support-tickets
+  type: output
+  description: "Navigate marketplace support — handle Amazon, eBay, and Etsy-specific support scenarios and understand DTC vs. marketplace support differences"
+  tags: [support, ecommerce, marketplace, amazon, etsy, dtc, intermediate]
+
+state: {}
+
+trigger: |
+  Many e-commerce businesses sell on marketplaces (Amazon, eBay,
+  Etsy) in addition to their own DTC (direct-to-consumer) store.
+  Each marketplace has its own rules, dispute resolution processes,
+  and customer expectations. Supporting marketplace customers
+  requires understanding the platform's unique constraints.
+
+  Part 1: Marketplace vs. DTC support
+  1. Key differences:
+     - DTC: Full control over policies, communication, customer data
+     - Marketplace: Platform controls policies, limited customer contact
+     - DTC: Brand voice in every interaction
+     - Marketplace: Constrained by platform messaging guidelines
+     - DTC: You handle returns/refunds directly
+     - Marketplace: Platform may handle (e.g., Amazon FBA)
+  2. Amazon seller support:
+     - A-to-Z Guarantee claims (customer protection)
+     - FBA vs. FBM handling differences
+     - Account health metrics (ODR, late shipment, cancellation rate)
+     - Restricted communication (messaging guidelines)
+     - Return policy determined by Amazon, not you
+  3. Etsy seller support:
+     - Handmade/custom product expectations
+     - Case system for disputes
+     - Star seller requirements
+     - Personalized buyer communication encouraged
+  4. eBay seller support:
+     - Money Back Guarantee disputes
+     - Seller performance standards
+     - Resolution center process
+     - Feedback management
+
+  Part 2: Handle marketplace support scenarios
+  For "CraftHaven" — selling handmade ceramics on Etsy and their
+  own Shopify store, plus some items on Amazon:
+
+  1. Amazon A-to-Z claim (unfair):
+     A customer filed an A-to-Z claim saying "item not as described"
+     for a handmade mug. Your listing clearly shows the item is
+     handmade with natural variations. The customer expected
+     machine-perfect consistency.
+     - Respond to the A-to-Z claim with evidence
+     - Explain handmade variations professionally
+     - Protect your seller metrics
+
+  2. Etsy custom order dispute:
+     "I ordered a custom name plate and you spelled the name wrong!
+     It says 'Micheal' but I wanted 'Michael'. I can't give this
+     as a gift now."
+     Check the order notes — the customer actually typed "Micheal"
+     in their custom order form.
+     - Handle diplomatically (they DID provide the wrong spelling)
+     - Offer resolution without blame
+     - Protect your Etsy seller rating
+
+  3. DTC vs. marketplace pricing complaint:
+     "I just saw the same vase on your website for $45 but I paid
+     $55 on Amazon! Why are you charging more on Amazon?!"
+     - Explain pricing differences (marketplace fees, fulfillment costs)
+     - Don't undermine your Amazon presence
+     - Consider offering a future incentive for DTC purchases
+
+  4. Etsy review response:
+     Customer leaves a 2-star review: "Nice quality but took
+     3 weeks to arrive. Way too long for the price."
+     You shipped within your stated 1-2 week processing time;
+     USPS delayed the rest.
+     - Write a professional public review response
+     - Balance defending your shop with empathy for their wait
+
+  5. Cross-platform order confusion:
+     "I have orders on both your website and Amazon. I want to
+     return the Amazon one but exchange the website one. Also,
+     can I use my website loyalty points on Amazon? This is confusing."
+     - Clarify the separate systems
+     - Help them navigate both returns
+     - Guide them toward DTC for future (loyalty benefits)
+
+  Part 3: Multi-marketplace strategy
+  1. Unified support across marketplaces
+  2. Protecting seller ratings and account health
+  3. Moving marketplace customers to DTC
+  4. Managing inventory and customer expectations across channels
+
+assertions:
+  - type: llm_judge
+    criteria: "Marketplace-specific constraints are respected — Amazon responses follow messaging guidelines, Etsy responses align with platform norms"
+    weight: 0.30
+    description: "Platform-appropriate responses"
+  - type: llm_judge
+    criteria: "The custom order spelling dispute is handled diplomatically — the customer's mistake is addressed without blame while still offering resolution"
+    weight: 0.25
+    description: "Diplomatic blame-free resolution"
+  - type: llm_judge
+    criteria: "The pricing complaint explains marketplace economics honestly — it justifies the price difference without making the customer feel cheated"
+    weight: 0.25
+    description: "Honest pricing explanation"
+  - type: llm_judge
+    criteria: "The review response is professional and effective — it provides context without being defensive or argumentative"
+    weight: 0.20
+    description: "Professional review response"

package/courses/ecommerce-support-tickets/scenarios/level-2/post-purchase-experience.yaml

@@ -0,0 +1,103 @@
+meta:
+  id: post-purchase-experience
+  level: 2
+  course: ecommerce-support-tickets
+  type: output
+  description: "Design the post-purchase support experience — create order confirmations, shipping updates, delivery follow-ups, and review request sequences"
+  tags: [support, ecommerce, post-purchase, notifications, follow-up, intermediate]
+
+state: {}
+
+trigger: |
+  The post-purchase experience determines whether a customer
+  buys again. Most e-commerce brands focus on acquisition but
+  neglect the journey after the order is placed. Proactive
+  post-purchase communication reduces support tickets AND
+  increases repeat purchases.
+
+  Part 1: Post-purchase communication timeline
+  1. Immediate (0-1 hour):
+     - Order confirmation email
+     - Payment confirmation
+     - Expected delivery date
+  2. Processing (1-48 hours):
+     - Order processing confirmation
+     - Any delays communicated proactively
+  3. Shipping (2-7 days):
+     - Shipping confirmation with tracking
+     - In-transit updates (major milestones)
+     - Out for delivery notification
+     - Delivered confirmation
+  4. Post-delivery (1-14 days):
+     - "How's your purchase?" check-in
+     - Product care/usage tips
+     - Review request
+     - Cross-sell recommendations
+  5. Ongoing (30+ days):
+     - Replenishment reminder (consumable products)
+     - Loyalty program updates
+     - Win-back if no repeat purchase
+
+  Part 2: Design the post-purchase experience
+  For "GreenLeaf" — an organic skincare e-commerce brand:
+
+  1. Write the complete email sequence:
+     Design 8 emails from order to repeat purchase:
+     a) Order confirmation (immediate)
+     b) Shipping confirmation (with branded tracking page concept)
+     c) Out for delivery notification
+     d) Delivery confirmation + "What's in your box" guide
+     e) Day 3: Product usage tips and routine suggestions
+     f) Day 7: "How's it going?" check-in + review request
+     g) Day 14: Complementary product recommendation
+     h) Day 30: Replenishment reminder
+
+     For each email: subject line, preview text, key content,
+     CTA, and tone guidance.
+
+  2. Design the branded tracking page:
+     Instead of sending customers to UPS.com, design a branded
+     tracking experience:
+     - Order status timeline (visual)
+     - Product tips while they wait
+     - FAQ for delivery issues
+     - Contact support option
+     - Cross-sell recommendations
+     - Social media engagement
+
+  3. Create the review request strategy:
+     - Optimal timing (how many days after delivery)
+     - Email template with incentive (discount for photo review)
+     - Follow-up if no review after first request
+     - Different approach for high-value vs. low-value orders
+     - Handling negative feedback within the review request flow
+
+  4. Design a proactive issue detection system:
+     - Monitor for delivery exceptions → auto-communicate
+     - Detect long processing time → auto-update customer
+     - Identify abandoned delivery → proactive reshipment offer
+     - Track return patterns → product improvement signal
+
+  Part 3: Measuring post-purchase impact
+  1. Post-purchase email engagement metrics
+  2. Tracking attribution (post-purchase to repeat purchase)
+  3. Support ticket reduction from proactive communication
+  4. Optimizing the sequence through A/B testing
+
+assertions:
+  - type: llm_judge
+    criteria: "The 8-email sequence is complete and well-timed — each email serves a clear purpose with appropriate spacing and content"
+    weight: 0.30
+    description: "Complete well-timed email sequence"
+  - type: llm_judge
+    criteria: "The branded tracking page adds genuine value — it goes beyond status updates to include product tips, FAQ, and support access"
+    weight: 0.25
+    description: "Value-adding tracking page"
+  - type: llm_judge
+    criteria: "The review request strategy is effective without being annoying — timing, incentives, and follow-up are balanced"
+    weight: 0.25
+    description: "Effective non-annoying review requests"
+  - type: llm_judge
+    criteria: "Proactive issue detection genuinely prevents support tickets — each automated trigger addresses a real customer pain point before they contact support"
+    weight: 0.20
+    description: "Genuinely preventive detection"

package/courses/ecommerce-support-tickets/scenarios/level-2/returns-logistics-optimization.yaml

@@ -0,0 +1,103 @@
+meta:
+  id: returns-logistics-optimization
+  level: 2
+  course: ecommerce-support-tickets
+  type: output
+  description: "Optimize the returns process — reduce return rates, streamline return logistics, and turn returns into exchange and retention opportunities"
+  tags: [support, ecommerce, returns, logistics, optimization, intermediate]
+
+state: {}
+
+trigger: |
+  E-commerce return rates average 20-30% (higher for fashion at
+  30-40%). Returns cost the industry $816 billion annually. But
+  returns aren't just costs — they're opportunities to learn
+  about product issues, convert to exchanges, and build loyalty
+  through easy processes. 92% of customers buy again from stores
+  with easy returns.
+
+  Part 1: Returns optimization strategy
+  1. Reducing return rates:
+     - Better product descriptions and photos
+     - Size guides and fit technology
+     - Customer reviews with fit information
+     - AR/virtual try-on (where applicable)
+     - Video demonstrations
+     - Pre-purchase chat support
+  2. Streamlining the return process:
+     - Self-service return portals (Loop Returns, Returnly)
+     - Prepaid return labels (when to offer free vs. paid)
+     - Drop-off locations vs. carrier pickup
+     - Instant exchange programs (ship replacement before return)
+     - Refund method options (original payment vs. store credit)
+  3. Return analytics:
+     - Return rate by product (which items have the most returns?)
+     - Return reasons (sizing, quality, expectations, shipping damage)
+     - Return-to-exchange conversion rate
+     - Cost per return
+     - Return customer future purchase rate
+
+  Part 2: Build a returns optimization program
+  For "FashionForward" — an online women's clothing retailer
+  (30% average return rate, $45 average order value):
+
+  1. Analyze return data and create an action plan:
+     Recent return data (last quarter, 10,000 returns):
+     - Size/fit issue: 42% (mainly dresses and jeans)
+     - Color different than expected: 18% (photos vs. reality)
+     - Quality not as expected: 15% (material feel, construction)
+     - Changed mind: 12% (impulse purchases, buyer's remorse)
+     - Wrong item received: 8% (fulfillment errors)
+     - Arrived too late: 5% (missed the occasion)
+
+     For each category:
+     - Root cause analysis
+     - Specific action to reduce return rate
+     - Expected impact (% reduction)
+     - Implementation difficulty
+
+  2. Design a self-service return portal:
+     - Customer-facing flow (step by step)
+     - Return reason capture (structured data, not free text)
+     - Exchange offer before refund (show available options)
+     - Return shipping options (free label, drop-off location)
+     - Refund method selection (original payment vs. 110% store credit)
+     - Estimated refund timeline communication
+
+  3. Create the exchange-first strategy:
+     - When to offer exchange before refund
+     - How to present exchange as more attractive
+     - Instant exchange logistics (ship new before receiving return)
+     - Store credit with bonus (incentivize keeping the revenue)
+     - A/B test: exchange offer vs. direct refund
+
+  4. Design the return feedback loop to product team:
+     - Weekly return reason report
+     - Product-specific quality alerts (return rate > 30%)
+     - Customer verbatim feedback summary
+     - Photo evidence from returns (quality issues)
+     - Sizing data for size chart improvements
+
+  Part 3: Returns economics
+  1. Calculating the true cost of a return
+  2. Free returns vs. paid returns (impact on conversion and returns)
+  3. Return policy as competitive advantage
+  4. Sustainability considerations in returns
+
+assertions:
+  - type: llm_judge
+    criteria: "Return data analysis connects each category to specific, actionable improvements — size/fit issues lead to better size guides, color issues to better photography"
+    weight: 0.30
+    description: "Data-driven return reduction"
+  - type: llm_judge
+    criteria: "The self-service portal design puts exchange options front and center — the flow naturally guides toward exchanges before offering refunds"
+    weight: 0.25
+    description: "Exchange-forward portal design"
+  - type: llm_judge
+    criteria: "The 110% store credit option is well-positioned — it's presented as a genuine value proposition, not a trick to avoid refunding money"
+    weight: 0.25
+    description: "Well-positioned store credit"
+  - type: llm_judge
+    criteria: "The product team feedback loop is structured and actionable — weekly reports with specific quality alerts enable real product improvements"
+    weight: 0.20
+    description: "Actionable product feedback"

package/courses/ecommerce-support-tickets/scenarios/level-2/review-reputation-management.yaml

@@ -0,0 +1,110 @@
+meta:
+  id: review-reputation-management
+  level: 2
+  course: ecommerce-support-tickets
+  type: output
+  description: "Manage reviews and reputation — respond to positive and negative reviews, handle review disputes, and build trust through social proof"
+  tags: [support, ecommerce, reviews, reputation, social-proof, intermediate]
+
+state: {}
+
+trigger: |
+  Reviews are the lifeblood of e-commerce. 93% of consumers read
+  reviews before buying, and a one-star increase can drive 5-9%
+  more revenue. Support's role in review management is critical —
+  responding to reviews, resolving issues that cause negative
+  reviews, and encouraging satisfied customers to leave reviews.
+
+  Part 1: Review management fundamentals
+  1. The review ecosystem:
+     - Product reviews (on-site: Yotpo, Judge.me, Stamped)
+     - Marketplace reviews (Amazon, Etsy, eBay)
+     - Third-party reviews (Google, Trustpilot, BBB)
+     - Social proof (Instagram, TikTok, YouTube)
+  2. Review response principles:
+     - Respond to ALL negative reviews (shows you care)
+     - Respond to positive reviews selectively (top reviews, detailed ones)
+     - Never argue with reviewers publicly
+     - Take resolution offline (DM, email) when possible
+     - Thank for feedback, even when negative
+  3. Review impact metrics:
+     - Average star rating (4.0+ is baseline, 4.5+ is excellent)
+     - Review volume (more reviews = more trust)
+     - Recency (recent reviews matter more)
+     - Photo/video reviews (highest trust)
+     - Response rate to reviews (shows active engagement)
+
+  Part 2: Handle review scenarios
+  For "KitchenCraft" — an online kitchenware store:
+
+  Write review responses for 6 scenarios:
+
+  1. Negative product review (legitimate issue):
+     ★★☆☆☆ "Non-stick pan started peeling after 2 months of
+     normal use. Very disappointed for a $65 pan."
+     - Acknowledge the quality issue
+     - Offer resolution (replacement under warranty)
+     - Show future customers you stand behind your products
+
+  2. Negative review (unreasonable expectations):
+     ★☆☆☆☆ "This cast iron pan is SO HEAVY. I can barely lift it
+     with food in it. Terrible design. Would not recommend."
+     - Cast iron IS heavy — that's the nature of the product
+     - Respond without making the customer feel foolish
+     - Help future buyers set correct expectations
+
+  3. Negative review (shipping, not product):
+     ★★☆☆☆ "Product seems great but it arrived 2 weeks late and
+     the box was crushed. Lucky the pan wasn't damaged."
+     - Address the shipping issue (which you can fix)
+     - Don't invalidate their frustration
+     - Show you're improving shipping
+
+  4. Positive detailed review:
+     ★★★★★ "Best chef's knife I've ever owned. Holds its edge
+     beautifully after 6 months of daily use. Worth every penny
+     of the $89 price tag."
+     - Engage meaningfully (not just "thanks!")
+     - Highlight the review for other shoppers
+     - Build the relationship
+
+  5. Fake/competitor review:
+     ★☆☆☆☆ "Worst company ever. Products are cheap Chinese
+     knockoffs. Buy from [competitor name] instead."
+     No order history found for this reviewer.
+     - Handle suspected fake review
+     - Report through platform's dispute process
+     - Write a professional public response in the meantime
+
+  6. Review with specific support complaint:
+     ★★☆☆☆ "The blender is fine but customer service was awful.
+     I emailed about a missing lid and got a robotic template
+     response after 5 days. Had to email 3 times before they
+     actually helped."
+     - Own the support failure
+     - Offer to make it right
+     - Use this as internal feedback
+
+  Part 3: Review strategy
+  1. Encouraging reviews (post-purchase email timing and content)
+  2. Handling review disputes with platforms
+  3. Using review data for product improvement
+  4. Building a review response playbook
+
+assertions:
+  - type: llm_judge
+    criteria: "Negative review responses are professional and resolution-oriented — each turns a negative into an opportunity to show customer care publicly"
+    weight: 0.30
+    description: "Professional resolution-focused responses"
+  - type: llm_judge
+    criteria: "The unreasonable review response is handled gracefully — cast iron weight is addressed without making the customer feel foolish while helping future buyers"
+    weight: 0.25
+    description: "Graceful unreasonable review handling"
+  - type: llm_judge
+    criteria: "The support complaint review is owned honestly — the response acknowledges the failure rather than making excuses"
+    weight: 0.25
+    description: "Honest support failure acknowledgment"
+  - type: llm_judge
+    criteria: "The suspected fake review is handled properly — the public response is professional while the private dispute process is initiated"
+    weight: 0.20
+    description: "Proper fake review handling"

package/courses/ecommerce-support-tickets/scenarios/level-2/seasonal-surge-management.yaml

@@ -0,0 +1,100 @@
+meta:
+  id: seasonal-surge-management
+  level: 2
+  course: ecommerce-support-tickets
+  type: output
+  description: "Manage seasonal support surges — prepare for Black Friday, holiday season, flash sales, and product launches with staffing, automation, and process plans"
+  tags: [support, ecommerce, seasonal, black-friday, holiday, surge, intermediate]
+
+state: {}
+
+trigger: |
+  E-commerce support volume can spike 200-300% during seasonal
+  events. Black Friday/Cyber Monday, holiday shipping deadlines,
+  flash sales, and product launches create predictable but intense
+  surges. Preparation is everything — you can't hire and train
+  agents the week before Black Friday.
+
+  Part 1: Seasonal surge fundamentals
+  1. Major e-commerce surge events:
+     - Black Friday / Cyber Monday (highest volume)
+     - Holiday season (Nov-Dec, up to 40% of annual revenue)
+     - Valentine's Day, Mother's Day, Father's Day
+     - Back-to-school season
+     - Flash sales and product launches
+     - Amazon Prime Day (affects all e-commerce)
+  2. What surges during peaks:
+     - WISMO tickets (3-4x normal volume)
+     - Payment failures (higher transaction volume = more errors)
+     - Discount code issues (expired, wrong code, stacking)
+     - Shipping deadline anxiety ("Will it arrive by Christmas?")
+     - Returns spike (January = "returns season")
+     - Stock issues (sold out, backorder confusion)
+  3. Preparation timeline:
+     - 12 weeks before: Hiring seasonal agents
+     - 8 weeks: Training program
+     - 4 weeks: System stress testing
+     - 2 weeks: Final preparation and rehearsal
+     - During: Real-time monitoring and adjustment
+     - After: Debrief and lessons learned
+
+  Part 2: Plan a Black Friday support operation
+  For "TrendMart" — a fashion and lifestyle e-commerce store:
+  Normal volume: 200 tickets/day, 5 agents
+  Expected BFCM volume: 800+ tickets/day for 5 days
+
+  1. Create the staffing plan:
+     - How many seasonal agents to hire (calculate)
+     - Training curriculum for seasonal agents (2-week program)
+     - Shift schedules for extended hours (6 AM - midnight)
+     - Agent specialization during surge (WISMO team, returns team,
+       payment team, general)
+     - Escalation protocols (when to involve permanent agents)
+
+  2. Design the automation strategy:
+     - Auto-replies for top 5 surge ticket types
+     - Self-service flows (order tracking, return initiation)
+     - Chatbot handling for common questions
+     - Proactive communication (shipping cutoff dates, sale details)
+     - Target: 40% deflection of surge tickets
+
+  3. Create the BFCM support playbook:
+     - Day-by-day plan (Wednesday pre-sale through Cyber Monday)
+     - Hour-by-hour coverage model for Black Friday
+     - Real-time monitoring dashboard (what to watch)
+     - Escalation matrix (when surge becomes crisis)
+     - Communication templates for surge-specific issues:
+       a) "Item sold out after you ordered" communication
+       b) "Shipping delays due to high volume" notification
+       c) "Payment processing delays" explanation
+       d) "Discount code expired/sold-out item" resolution
+
+  4. Plan the January returns surge:
+     - Expected return volume (30-40% of holiday orders)
+     - Processing capacity and timeline
+     - Gift return handling (no receipt, exchange preferences)
+     - Refund processing timeline communication
+
+  Part 3: Continuous improvement
+  1. Post-surge retrospective (what worked, what didn't)
+  2. Year-over-year surge comparison metrics
+  3. Building institutional knowledge for future surges
+  4. Technology investments to reduce surge impact
+
+assertions:
+  - type: llm_judge
+    criteria: "The staffing plan is mathematically sound — agent count, shift coverage, and ticket capacity calculations add up for 800+ tickets/day"
+    weight: 0.30
+    description: "Mathematically sound staffing"
+  - type: llm_judge
+    criteria: "The automation strategy targets the right ticket types — WISMO and self-service returns deflect the highest volume surge categories"
+    weight: 0.25
+    description: "High-impact automation targeting"
+  - type: llm_judge
+    criteria: "The BFCM playbook is operationally detailed — it covers specific hours, specific scenarios, and specific communication templates"
+    weight: 0.25
+    description: "Operationally detailed playbook"
+  - type: llm_judge
+    criteria: "The January returns plan anticipates the post-holiday reality — gift returns, refund timelines, and processing capacity are addressed"
+    weight: 0.20
+    description: "Realistic post-holiday planning"