Last Updated: October 22, 2025
Status: Strategic Pivot - Approved
Priority: Core Product Definition
Previous Positioning: Simple menu builder for small Vietnamese businesses
New Positioning: AI-powered business intelligence platform for restaurant chains and non-technical owners
Core Insight: Menu creation and online ordering are just data collection tools. The real value is AI-driven business intelligence that helps owners make better decisions when they lack business expertise or can't manage multiple locations.
Target Customer #1: Chain Restaurant Owners
Pain Points:
- Own 3-10 locations across a city
- Each location has different customer demographics
- Each location needs different strategies (pricing, promotions, inventory)
- Cannot physically manage all locations - need data-driven insights
- Current solution: Manual spreadsheets, gut feeling, or expensive consultants
Target Customer #2: Skilled Operators Without Business Training
Pain Points:
- Expert chef/barista but no business/economics background
- Don't know how to analyze profitability, optimize pricing, or forecast demand
- Lose money on inventory waste, missed promotion opportunities
- Need AI co-pilot to make business decisions
Previous Approach:
- Compete with 100+ menu builders on features
- Commoditized product (easy to copy)
- Low switching costs
- Price-sensitive market
New Approach:
- AI recommendations are proprietary (data moat)
- High switching costs (AI improves over time with tenant's data)
- Premium pricing justified (increase revenue by 10-20%, reduce waste by 30%)
- Network effects (multi-tenant data improves all models)
Important: Both segments are equally important. They have different pain points but the same solution: AI-powered decision support.
Profile:
- Vietnamese coffee chains (Highlands Coffee style, but smaller)
- Phở/Bún chains expanding from 1 flagship to multiple locations
- Bubble tea franchises
- Bánh mì chains
Combined Market Size (Vietnam):
Segment A (Chains):
- ~50,000 restaurant chains with 2+ locations
- ~10,000 chains with 5+ locations
- Growing 15-20% annually as Vietnam urbanizes
- Willingness to pay: ₫500k-2M/month (vs ₫5-15M on POS + manual management)
Segment B (Skilled Operators):
- ~200,000 owner-operated restaurants in Vietnam
- ~30% are skilled operators without business training (60,000)
- High churn rate: 50% fail in first 2 years due to poor business decisions
- Willingness to pay: ₫200k-500k/month (avoid hiring accountant at ₫8-10M/month)
Total Addressable Market: ~110,000 potential customers
Profile:
- Master chefs opening first restaurant
- Award-winning baristas opening café
- Bakers with culinary degrees but no MBA
Pain Point:
- "I know how to make great food, but I don't know if I'm making money"
- "Should I raise prices? Add a new item? Run a promotion?"
- "I have 20kg of ingredients expiring tomorrow - what should I do?"
Market Size:
- ~200,000 owner-operated restaurants in Vietnam
- ~30% are skilled operators without business training
- High churn (50% fail in first 2 years due to poor business decisions)
"AI quản lý kinh doanh cho chủ quán không có thời gian hoặc kiến thức về kinh tế"
(AI business manager for owners without time or business knowledge)
This single value proposition serves both segments:
- Chain owners: "không có thời gian" (no time to manage all locations)
- Skilled operators: "kiến thức về kinh tế" (lack business/economics knowledge)
1. Admin Dashboard + AI Recommendations (80% of development effort)
This is the core product. Everything else exists to feed data into this.
Features:
- Real-time multi-location dashboard
- AI recommendations (pricing, promotions, inventory)
- Automated reports for owners
- Predictive analytics (demand forecasting, profit optimization)
2. Data Collection Tools (15% of development effort)
These are means to an end - collect data for AI:
- Online ordering system (track customer behavior)
- POS integration (real sales data)
- Menu management (track item performance)
- Inventory tracking (prevent waste)
3. Customer-Facing Tools (5% of development effort)
Nice-to-have for completeness:
- Public storefront/menu website
- QR code menus
- Zalo Mini App (future)
1. Multi-Tenant Data Advantage
- Each new tenant improves the model for everyone
- Single-location tools can't learn patterns across businesses
- Example: "Rainy day → soup sales up 30%" learned from 1,000 tenants
2. Localized Models
- Vietnam-specific: Understand Tết, weather patterns, local preferences
- City-specific: HCMC lunch patterns ≠ Hanoi patterns
- Business-type specific: Café AI ≠ Restaurant AI
3. Time-Based Moat
- AI improves with each tenant's historical data (6+ months)
- New competitor starts from zero
- Switching means losing personalized insights
4. Integration Depth
- Deep integration with Vietnamese payment/delivery platforms
- Training data from MoMo transactions, GrabFood orders
- Competitors need same integrations to replicate
Goal: Prove AI recommendations work and add value
Features:
- ✅ Multi-location dashboard
- Switch between locations
- Compare performance across locations
- Aggregate owner-level view
- ✅ Manual data entry tools
- Daily sales logging per location
- Inventory tracking
- Menu management
- ✅ Basic AI recommendations (rule-based + simple ML)
- "Phở bò selling 2x more at Location A than B - replicate strategy?"
- "Item X hasn't sold in 3 days - run 20% off promotion?"
- "You're running low on ingredient Y at Location C"
- ✅ Simple ordering system (just to collect customer behavior data)
Success Metric: 10 chain owners paying ₫500k/month, reporting 10-15% revenue increase
Goal: AI becomes indispensable - owners can't live without it
Features:
- 🤖 Demand forecasting
- Predict tomorrow's sales by item, by location
- Optimize inventory orders
- Staff scheduling recommendations
- 🤖 Dynamic pricing suggestions
- Real-time pricing optimization
- Competitive pricing analysis (scrape competitor menus)
- Promotional timing (when to discount, when to hold firm)
- 🤖 Customer segmentation per location
- "Location A customers prefer premium items"
- "Location B is price-sensitive - focus on combos"
- 🤖 Automated campaign creation
- AI writes promotion copy
- Auto-posts to Facebook/Zalo
Success Metric: 50 chains paying ₫1-2M/month, 85% retention, NPS >50
Goal: AI operates the business, owner just approves
Features:
- 🤖 Autonomous inventory management
- AI automatically orders supplies when low
- Negotiates with suppliers (via API)
- 🤖 Multi-location optimization
- Transfer inventory between locations
- Move staff based on predicted demand
- 🤖 Financial modeling
- "Should you open a 4th location? Here's the ROI model"
- "Expand Location B or renovate Location A?"
- 🤖 Voice assistant
- "How did Location 3 do today?" → AI responds with summary
Success Metric: 200+ chains, ₫50M+ MRR, expand to Thailand/Indonesia
Free Tier: Hobby/Testing
- Single location only
- Basic dashboard, no AI recommendations
- Manual data entry
- Use case: Testing product, very small operations
- Upsell path A: When they open location #2 → Growth tier
- Upsell path B: When they want AI guidance → Starter tier
Starter Tier: ₫200k/month (NEW - for Segment B: Skilled Operators)
- 1 location only (perfect for single restaurant/café)
- Basic AI recommendations (pricing guidance, waste alerts, profitability tips)
- Daily automated reports ("Hôm nay bán được gì?", "Món nào lời nhất?")
- Manual data entry (or POS integration)
- Target: Chefs, baristas who lack "kiến thức kinh tế"
- Value: Avoid hiring accountant (₫8-10M/month) or business consultant (₫20M/project)
Growth Tier: ₫500k/month (for Segment A: Small Chains)
- Up to 3 locations
- Basic AI recommendations (cross-location comparison)
- Automated reports per location
- POS integration
Pro Tier: ₫1.5M/month (for larger chains OR power users)
- Up to 10 locations (for chains) OR 1 location with advanced AI (for skilled operators)
- Advanced AI (demand forecasting, dynamic pricing, customer segmentation)
- Priority support + monthly business consultant calls
- Custom integrations
- Segment A use case: 5-10 location chains needing sophisticated analytics
- Segment B use case: High-revenue single location (₫500M+/month) wanting maximum AI optimization
Enterprise: Custom Pricing
- 10+ locations
- White-label option
- Dedicated success manager
- Custom AI models for their business type
Conservative:
- 50 chains × ₫500k/month = ₫25M/month = ₫300M/year (~$12k USD)
- 20 chains × ₫1.5M/month = ₫30M/month = ₫360M/year (~$15k USD)
- Total Year 1: ₫660M (~$27k USD)
Optimistic:
- 200 chains × ₫500k = ₫100M/month
- 100 chains × ₫1.5M = ₫150M/month
- 10 enterprise × ₫5M = ₫50M/month
- Total Year 1: ₫3.6B/year (~$150k USD)
Old: "Tạo website thực đơn trong 5 phút" (Create menu website in 5 minutes)
New: "AI giúp chủ quán kinh doanh thông minh hơn, không cần MBA" (AI helps restaurant owners run smarter businesses, no MBA required)
Target: Chain Owners (Not Solo Operators)
- Identify chains: Scrape Google Maps for businesses with 2+ locations
- Cold outreach: "We analyzed your 3 locations - Location A is underperforming by 20%. Want to see why?"
- Free audit: Offer 1-month free AI analysis (requires data access)
- Land & expand: Start with 2-3 locations, expand as they grow
Channels:
- LinkedIn outreach to F&B owners
- Facebook Groups for restaurant owners
- Partnership with POS system vendors (data integration)
- Content marketing: Blog about "How to manage multiple restaurant locations"
- Free tier for solo owners
- Automatic upsell prompt when they mention opening location #2
- Referral bonus: "Bring another chain owner, get 2 months free"
1. Multi-Location Data Model
// Every tenant can have multiple locations
tenants (1) ─────< (many) locations
locations (1) ────< (many) menu_items
locations (1) ────< (many) orders
locations (1) ────< (many) inventory_items
// Analytics must be location-aware
analytics_events {
tenant_id,
location_id, // NEW: Track which location
...
}2. AI Training Pipeline
- Separate models per business type (café vs restaurant)
- Transfer learning from multi-tenant data
- Per-tenant fine-tuning after 3+ months of data
3. Data Collection Priority
Focus on collecting:
- Sales by item, by location, by hour
- Inventory levels and waste
- Customer demographics (age, order frequency)
- External data (weather, local events, competitor pricing)
Phase 1 (MVP):
- PostgreSQL: Sufficient for 50 tenants
- Basic ML: Scikit-learn for rule-based recommendations
- Manual data entry: Simple forms
Phase 2 (Scale):
- TimescaleDB: Time-series data for real-time analytics
- Cloud ML: Google Cloud AI or AWS SageMaker for training
- Real-time pipeline: Kafka + Spark for streaming data
Phase 3 (Advanced):
- Custom models: PyTorch/TensorFlow for deep learning
- GPU instances: For training large models
- CDN: Serve recommendations in <100ms
Phase 1 (MVP):
- # of chains signed up (target: 10)
- # of locations managed (target: 30-50)
- Data quality: >80% daily sales logged
- AI accuracy: >70% of recommendations accepted
Phase 2 (Growth):
- Monthly Active Chains: 50+
- Revenue per chain: ₫750k average
- NPS: >50
- Churn: <5% monthly
Phase 3 (Scale):
- Monthly Active Chains: 200+
- Revenue per chain: ₫1.2M average
- AI impact: 15-20% revenue increase proven
- Expansion: Launch in Thailand or Indonesia
Must prove AI creates value:
- Revenue increase: 10-20% within 6 months
- Waste reduction: 30% fewer expired inventory
- Time saved: 2 hours/day on manual analysis
- Better decisions: 85% of AI recommendations lead to positive outcomes
Problem: AI needs data, but new tenants have no data
Mitigation:
- Start with rule-based recommendations (works day 1)
- Use multi-tenant data for baseline predictions
- Offer 3-month onboarding with business consultant
Problem: Manual entry leads to incomplete/inaccurate data
Mitigation:
- Prioritize POS integration (automated data)
- Gamification: "Complete daily logs, unlock premium features"
- Validation rules: Flag suspicious entries
Problem: Bad advice damages trust and business
Mitigation:
- Always show confidence level (e.g., "85% confident")
- Explain reasoning (transparency)
- A/B test recommendations on subset of locations
- Insurance fund: Compensate if AI causes significant loss
Problem: Feature can be replicated
Mitigation:
- Data moat: Competitors need years of multi-tenant data
- Speed: Ship fast, accumulate data quickly
- Integrations: Deep partnerships with MoMo, GrabFood, etc.
- Switching costs: Historical data and trained models lock in customers
- Get stakeholder sign-off on refined strategy
- Update all planning documents to reflect chain-owner focus
- Revise MVP acceptance criteria (prioritize multi-location features)
- Interview 10 chain owners (3-10 locations)
- Validate willingness to pay ₫500k-1.5M/month
- Understand their current pain points and tools
- Design multi-location database schema
- Plan AI recommendation engine architecture
- Identify initial ML models to implement (Phase 1)
- Build multi-location admin dashboard
- Implement basic AI recommendations (rule-based)
- Create onboarding flow for chain owners
1. POS Systems (Sapo, KiotViet)
- Strong at transactions, weak at multi-location analytics
- No AI recommendations
- Expensive (₫5-10M setup + ₫1-2M/month)
- Our advantage: Better analytics, AI-driven, cheaper
2. Accounting Software (MISA, Fast)
- Good at finances, terrible at operations
- No real-time insights
- Complex UI (requires training)
- Our advantage: Operations-focused, real-time, simple UX
3. Foreign Tools (Toast, Square)
- Not localized (no MoMo, Zalo integration)
- US/EU pricing (too expensive)
- English-only
- Our advantage: Hyper-localized, Vietnamese-first
4. Consultants
- Expensive (₫20-50M for 3-month engagement)
- One-time advice, no ongoing support
- Our advantage: Always-on AI, learns and improves
No one offers:
- AI-powered multi-location optimization
- Vietnamese-first platform with deep local integrations
- Affordable pricing for SMB chains
We can own this space for 2-3 years before competitors catch up.
Document Owner: Product + Strategy Team
Review Cadence: Monthly (or after major pivots)
Next Review: December 2025
Feedback: Open issue with label product-strategy