RevenueBoost AI

• 

  Landing Page

• 

  Staff counter Tab - Recommend the Best Cross sell service to the Customer which is likely to get accepted

• 

  Dashboard - Shows live track of metrics of the Salon business

• 

  Revenue Simulation - Helps the owner to understand the business impact

• 

  First section shows top Cross Sells Rule Analyzed By our ML models and other section is todays recommendation logs.

• 

  Chatbot which helps staff to chat with the Database of the Salon in natural Language

• 

  Settings tab

🚀 Project Story — RevenueBoost AI

💡 Inspiration

The idea for RevenueBoost AI came from observing a common problem in salons.

Despite having regular customers, salons often fail to maximize revenue from each visit. Staff rely on guesswork to suggest additional services, which leads to missed upsell and cross-sell opportunities.

At the same time, salons already collect valuable customer data through POS systems — such as visit history, spending patterns, and service preferences — but this data is not actively used.

This led to a key question:
What if this data could drive real-time , personalized cross sell recommendation?

---

🧠 What We Built

RevenueBoost AI is a system that converts raw POS data into real-time, intelligent sales decisions.

When a customer checks in, the system:

• Recommends the most relevant services
  
• Predicts the likelihood of accepting an offer
  
• Segments the customer for personalization
  
• Validates recommendations based on available time
  
• Generates a ready-to-use sales script for staff
  

This ensures every recommendation is data-driven, personalized, and practical.

---

⚙️ How We Built It

We designed a hybrid AI system combining multiple ML models:

• XGBoost → Predicts receptivity score
  
• FP-Growth → Finds cross-sell patterns
  
• K-Means → Segments customers
  
• Time-aware filter → Ensures feasibility
  

Tech Stack:

• Frontend: React
  
• Backend: FastAPI (Python)
  
• Database: Supabase
  
• AI: Gemini (for script generation)
  

---

⚠️ Challenges We Faced

Data Transformation
Raw POS data required feature engineering (recency, premium ratio, etc.)

Real-Time Constraints
Ensuring recommendations fit within limited appointment time

Cold Start Problem
Handling new customers with no historical data

Human Interaction Gap
Bridging the gap between AI recommendations and staff communication

---

📚 What We Learned

• AI should focus on decision-making, not just prediction
  
• Feature engineering is critical for meaningful insights
  
• Hybrid systems (ML + rules) are more practical
  
• Usability is as important as model accuracy
  

---

🎯 Impact

RevenueBoost AI turns passive data into an active revenue driver.

Extra Revenue Formula

Extra Revenue = Customers × Acceptance Rate × Avg Upsell Value

Example:
500 × 10% × $45 = $2250

Even small improvements in conversion can generate significant revenue.

---

🔥 Closing

Large companies use data to drive decisions.
RevenueBoost AI brings that same capability to small businesses — enabling smarter, real-time decisions at every customer interaction.

Built With

• fastapi
• machine-learning
• netlify
• postgresql
• python
• react
• render
• rest
• supabase