EXERCISE 2 OF 6

Context Layering

Exercise 2: Context Layering

Mr. DIY Thailand - Generative AI Workshop

Objective: Experience how each context layer progressively improves output quality

📖 Context & Scenario

About Mr. DIY Thailand

Mr. DIY is Thailand’s fastest-growing value retail chain specializing in home improvement, hardware, and household products. Since entering Thailand in 2016, we’ve grown to over 1,000 stores nationwide with 16.2 billion THB in annual revenue.

Core Brand Promise:

  • “Always Low Prices” - Value-focused positioning
  • 15,000+ SKUs across home improvement, hardware, household items
  • Budget-friendly alternative to premium competitors (20-30% lower prices)
  • Serving DIY homeowners, small contractors, and budget-conscious shoppers

Today’s Scenario

You’re helping Mr. DIY’s operations team create AI-powered product recommendations for customers. Your goal is to learn how adding relevant business context progressively improves the quality and usefulness of AI outputs.

🎯 Your Task

Create AI-powered cross-sell recommendations to increase basket value for Mr. DIY customers.

Scenario Context:

  • Customer status: Has already placed a power drill (THB 899) in their online shopping basket
  • Your role: You are designing product recommendations for Mr. DIY’s online recommendation system (displayed at checkout)
  • Business objective: Generate 3-5 complementary product recommendations that increase basket purchase value by THB 300-500 while maintaining Mr. DIY’s “Always Low Prices” positioning
  • Success metric: Customer adds 2-3 practical items that increase basket value by THB 300-500
  • What you’re recommending: Complementary accessories and tools that complete the customer’s project (NOT selling the drill itself)

You will build your prompt progressively through 4 context layers and observe quality improvement at each step.

🔑 Core Principle

Principle 2: Context is King

Better context = Better quality. Always.

The more relevant context you provide, the better AI understands your needs and delivers higher quality outputs.

📋 Activity Structure

  • 5 minutes: Introduction and setup
  • 30 minutes: Progressive building through 4 context layers
  • 5 minutes: Pairs sharing
  • 5 minutes: Group debrief
  • Work mode: Individual or pairs (your choice)
  • Tool choice: ChatGPT OR Claude (free tier is sufficient)

Important Notes

  • Fresh conversation recommended: Start a NEW conversation for each layer to see the isolated impact of each context addition
  • Iteration encouraged: If a layer’s output isn’t satisfactory, refine your prompt and try again
  • WiFi required: Ensure you have stable internet connection
  • AI tool access: Free tier of ChatGPT or Claude is sufficient - no paid subscription required

🧱 The 4 Context Layers

🔄 LAYER 1: Basic Prompt

Time: 5 minutes

⚠️ IMPORTANT: Start a NEW conversation

Open a completely fresh chat window in ChatGPT or Claude for this layer.

Start with the most basic prompt possible. No context, just the core request.

Suggested prompt: “Recommend complementary products for a customer who just added a power drill to their basket”

Instructions:

  1. Submit this basic prompt to your chosen AI tool
  2. Review the output
  3. Rate quality: 1-5 ⭐ (1=poor, 5=excellent)
  4. Record output in your worksheet

📝 Layer 1 Observation:

After reviewing the AI output, consider:

  • What specific product recommendations did you receive?
  • Are they generic or specific to your scenario?
  • Do they reflect any understanding of Mr. DIY’s brand positioning?

Quality Rating: ☐ 1 ☐ 2 ☐ 3 ☐ 4 ☐ 5

🔄 LAYER 2: Brand Voice + Price Positioning

Time: 8 minutes

⚠️ IMPORTANT: Start a NEW conversation (do NOT continue from Layer 1)

Open a completely fresh chat window. This is a separate conversation to show the impact of added context.

Add Mr. DIY’s brand context to your prompt.

Context to add:

  • Brand positioning: “Always Low Prices” - value-focused, practical, accessible
  • Tone: Friendly, helpful, budget-conscious (not luxury, not cheap)
  • Price positioning: Budget-friendly, typically 20-30% below premium competitors like HomePro

NEW prompt example: “As Mr. DIY (budget-friendly home improvement retailer with ‘Always Low Prices’ positioning), recommend complementary products to upsell to a customer who just added a power drill to their basket. Recommendations should emphasize value and practicality to increase basket value.”

Instructions:

  1. Reformulate your prompt with brand context
  2. Submit to AI
  3. Compare to Layer 1 output
  4. Rate quality: 1-5 ⭐
  5. Note what improved

📝 Layer 2 Observation:

After reviewing the AI output, consider:

  • How did adding brand context change the recommendations?
  • Do the suggestions now reflect Mr. DIY’s “Always Low Prices” positioning?
  • Is the tone more aligned with budget-conscious, practical values?

Quality Rating: ☐ 1 ☐ 2 ☐ 3 ☐ 4 ☐ 5

Reflection: What improved with this context layer?

  • Think about specific differences you noticed compared to Layer 1

🔄 LAYER 3: Target Customer + Seasonal Context

Time: 8 minutes

⚠️ IMPORTANT: Start a NEW conversation (do NOT continue from Layer 2)

Open a completely fresh chat window. This is a separate conversation to show the impact of added context.

Add customer profile and seasonal context to make recommendations more relevant.

Context to add:

Target Customer Profile:

  • DIY homeowners (age 30-50)
  • Small contractors and handymen
  • Budget-conscious shoppers looking for good value
  • Often need complementary products for home projects

Seasonal Context:

  • Current season: Rainy season (July-October)
  • Impact: Home repair needs increase (leaks, mold, water damage)
  • Customer mindset: Fixing problems, preventing water damage
  • Project types: Roof repairs, drainage, waterproofing, mold treatment

NEW prompt example: “As Mr. DIY, recommend complementary products to upsell to a customer who just added a power drill to their basket. Target customers are DIY homeowners (30-50 years old) and small contractors who are budget-conscious. It’s currently rainy season (July-Oct), so home repairs are common - focus on complementary products useful for rainy season home projects like roof repairs and waterproofing to increase basket value.”

Instructions:

  1. Refine your prompt with customer and seasonal context
  2. Submit to AI
  3. Compare to Layer 2 output
  4. Rate quality: 1-5 ⭐
  5. Note what improved

📝 Layer 3 Observation:

After reviewing the AI output, consider:

  • Are the recommendations now more relevant to the rainy season context?
  • Do they address specific customer needs (DIY homeowners, small contractors)?
  • How does understanding the seasonal context improve product relevance?

Quality Rating: ☐ 1 ☐ 2 ☐ 3 ☐ 4 ☐ 5

Reflection: What improved with this context layer?

  • Think about how customer profile and seasonal factors changed the recommendations

🔄 LAYER 4: Inventory Data + Competitive Intelligence

Time: 9 minutes

⚠️ IMPORTANT: Start a NEW conversation (do NOT continue from Layer 3)

Open a completely fresh chat window. This is a separate conversation to show the impact of added context.

Add real-world business constraints: current inventory levels and competitive positioning.

Context to add:

Current Inventory Levels:

  • Power Drills: 50 units (good stock) - can confidently recommend
  • Sanders: 30 units (moderate stock) - can recommend but note availability
  • Power Saws: Limited stock (only 8 units left) - create urgency
  • Drill Bits & Accessories: High stock (200+ units) - strongly push
  • Safety Equipment: High stock (gloves, goggles, masks)

Competitive Intelligence:

  • HomePro: Focuses on premium brands (Bosch, Makita) at higher prices
  • Mr. DIY advantage: Value brands with good quality at 20-30% lower prices
  • Customer switching reason: Price-conscious buyers choose Mr. DIY for budget-friendly options
  • Recommendation strategy: Emphasize value, bundle deals, practical use cases

NEW prompt example: “As Mr. DIY, recommend complementary products to upsell to a customer who just added a power drill to their basket (we have 50 drill units in stock). Target customers are budget-conscious DIY homeowners and small contractors during rainy season. Consider our inventory to maximize cross-sell: Sanders (30 units - moderate stock), Power Saws (8 units - LIMITED stock, create urgency!), Drill Bits & Accessories (200+ units - push hard as high-margin add-ons), Safety gear (high stock). Our advantage over HomePro is 20-30% lower prices on good quality tools. Emphasize value bundles and practical rainy season applications to increase average basket value.”

Instructions:

  1. Add inventory and competitive context to your prompt
  2. Submit to AI
  3. Compare to Layer 3 output - notice strategic recommendations
  4. Rate quality: 1-5 ⭐
  5. Note what improved - look for urgency, bundling, strategic suggestions

📝 Layer 4 Observation:

After reviewing the AI output, consider:

  • Do the recommendations show strategic thinking about inventory management?
  • Is there urgency created for limited stock items?
  • Are high-stock accessories being pushed more prominently?
  • Do recommendations leverage Mr. DIY’s competitive advantage versus HomePro?

Quality Rating: ☐ 1 ☐ 2 ☐ 3 ☐ 4 ☐ 5

Reflection: What improved with this context layer?

  • Notice how business constraints drive more strategic, actionable recommendations

📊 Quality Progression Summary

Review your quality ratings across all 4 layers. Did quality improve progressively?

Layer Your Rating
Layer 1 (Basic) ⭐ ___
Layer 2 (Brand) ⭐ ___
Layer 3 (Customer + Season) ⭐ ___
Layer 4 (Inventory + Competition) ⭐ ___

💬 Pairs Sharing

Time: 5 minutes

Find another pair (or individual) and discuss:

  1. Which context layer had the BIGGEST impact on quality?
  2. Compare quality scores: Did they improve consistently?
  3. What specific improvements did you see at each layer?
  4. Which contexts are most important for YOUR work at Mr. DIY?

🎤 Group Debrief

Time: 5 minutes

Be prepared to share insights with the full group. Instructor will ask:

  • “Show of hands: Whose output got significantly better with each layer?”
  • “Which context layer had the BIGGEST impact? Why?”
  • “In your work at Mr. DIY, what context would be most valuable to include?”

2 pairs will be selected to share examples with the full group.

📋 Reusable Prompts

4-Layer Context Template

Here’s a reusable template for building context-rich prompts in your work. Copy and customize for your specific needs.

LAYER 1: Basic Task
[State your core request clearly]

LAYER 2: Brand Voice & Positioning
- Brand: [Your company/department name]
- Positioning: [Value proposition, key differentiators]
- Tone: [Desired communication style]
- Price/Quality level: [How you position vs. competitors]

LAYER 3: Target Audience & Context
- Target audience: [Who will receive/use this?]
- Current context: [Seasonal, temporal, situational factors]
- Customer mindset: [What are they thinking/feeling?]
- Relevant use cases: [Specific scenarios where this applies]

LAYER 4: Business Constraints & Intelligence
- Resource constraints: [Inventory, budget, time, personnel]
- Competitive landscape: [Key competitors, your advantages]
- Strategic priorities: [What matters most right now?]
- Success metrics: [How will you measure success?]

Request: Combine all layers above to [YOUR SPECIFIC TASK]

Example: Mr. DIY Product Recommendation

LAYER 1: Basic Task
Recommend products for a customer buying a power drill.

LAYER 2: Brand Voice & Positioning
- Brand: Mr. DIY
- Positioning: "Always Low Prices" - value-focused, practical, accessible
- Tone: Friendly, helpful, budget-conscious
- Price level: 20-30% below premium competitors like HomePro

LAYER 3: Target Audience & Context
- Target audience: DIY homeowners (30-50 years), small contractors
- Current context: Rainy season (July-October)
- Customer mindset: Fixing water damage, preventing leaks
- Relevant use cases: Roof repairs, waterproofing, drainage projects

LAYER 4: Business Constraints & Intelligence
- Inventory: Drills (50 units), Sanders (30), Saws (8 - limited!),
  Drill bits (200+), Safety gear (high stock)
- Competition: HomePro focuses on premium brands at higher prices
- Strategic priorities: Push high-stock accessories, create urgency on
  limited items
- Success metrics: Bundle attachment rate, average transaction value

Request: Create product recommendations that emphasize value bundles,
create urgency for limited stock, and align with rainy season needs.

✅ Quality Checklist

Use this checklist to evaluate if your context-layered prompt is ready to use in real work.

Self-Assessment Criteria

For each layer, check if your prompt includes:

LAYER 1 - Basic Task:

  • ☐ Clear, specific request (not vague)
  • ☐ Understandable by someone unfamiliar with your work

LAYER 2 - Brand Voice:

  • ☐ Brand identity clearly stated
  • ☐ Tone/voice specified (formal, casual, friendly, etc.)
  • ☐ Value positioning mentioned (premium, budget, mid-range)
  • ☐ Key differentiators included

LAYER 3 - Target Audience:

  • ☐ Specific audience defined (not “everyone”)
  • ☐ Relevant temporal context (season, time, event)
  • ☐ Audience mindset/needs described
  • ☐ Use cases or scenarios provided

LAYER 4 - Business Constraints:

  • ☐ Resource limitations specified (inventory, budget, time)
  • ☐ Competitive landscape mentioned
  • ☐ Strategic priorities stated
  • ☐ Success criteria defined

Quality Red Flags

Your output may need revision if:

  • ❌ Generic recommendations that could apply to any company
  • ❌ No mention of your brand identity or values
  • ❌ Ignores resource constraints (suggests out-of-stock items)
  • ❌ Doesn’t reflect target audience needs
  • ❌ Misses seasonal or temporal relevance
  • ❌ Fails to differentiate from competitors
  • ❌ No actionable business strategy

Improvement Needed?

If you checked fewer than 3 items per layer: → Add more context to that layer and regenerate

If you see 2+ red flags: → Review your context layers and add missing business details

If quality rating is below 4 stars: → Identify which context layer is weakest and strengthen it

❓ Need Help?

Common Issues & Solutions

Problem: AI output is still generic even with all 4 layers

  • ✓ Solution: Make your context MORE specific (use actual product names, specific numbers, real competitor names)
  • ✓ Try: Add one more example of what “good” looks like

Problem: Output doesn’t reflect Mr. DIY brand voice

  • ✓ Solution: Explicitly state “Use friendly, budget-conscious tone” in your prompt
  • ✓ Try: Add specific phrases Mr. DIY uses (e.g., “Always Low Prices”)

Problem: Recommendations ignore inventory constraints

  • ✓ Solution: Put inventory numbers in CAPS or bold to emphasize
  • ✓ Try: Explicitly request “Prioritize high-stock items”

Problem: Not seeing much difference between layers

  • ✓ Solution: You may be starting too context-rich in Layer 1. Try a MORE basic Layer 1.
  • ✓ Try: Start Layer 1 with just 5-7 words

Problem: AI suggests products Mr. DIY doesn’t sell

  • ✓ Solution: Add “Only recommend products typically sold at hardware/home improvement stores”
  • ✓ Try: List specific product categories available

Problem: Running out of time

  • ✓ Solution: Focus on quality over quantity. Complete 2-3 layers well rather than rushing through all 4.
  • ✓ Skip: Detailed note-taking; just rate and observe key differences

🆘 Still Stuck?

  • Raise your hand and an instructor will assist

✅ Reflection Questions

After completing all 4 layers, reflect on your experience:

1. Which context layer surprised you the most?

  • Think about which layer created the biggest improvement in output quality

2. How will you apply context layering to your work at Mr. DIY?

  • Consider specific tasks where adding context layers could improve AI outputs
  • What types of context are most relevant to your role?

3. If you could only add ONE context layer due to time constraints, which would you choose and why?

  • Which context layer provides the most value for the effort invested?
  • How does your choice depend on the task at hand?

🔑 Key Takeaway

Context is King

Better Context = Better Quality. Always.

The more relevant context you provide, the better AI understands what you need.

This principle applies to:

  • ✓ Marketing content creation
  • ✓ Operations planning and optimization
  • ✓ Customer communications
  • ✓ Product recommendations
  • ✓ ANY business task involving AI

Context layers you can apply in your work:

  • Brand voice and positioning
  • Target audience profile
  • Seasonal or temporal context
  • Business constraints (inventory, budget, resources)
  • Competitive landscape
  • Strategic objectives

🚀 Next Steps

Next, we’ll learn how to TUNE AI outputs for tone, detail, and accuracy through conversational refinement in Exercise 3.

Workshop Materials Mr. DIY Thailand Generative AI Workshop Exercise 2: Context Layering

← Previous Exercise Back to Home Next Exercise →