rhordoan/sb-mcp-server

Enhanced Recommendation System for Sally Beauty

1. Introduction

This project is an MCP (Model Context Protocol) server designed to enhance the e-commerce experience and in-store customer assistance for Sally Beauty. The system provides a personalized shopping journey by analyzing customer behavior, professional needs, seasonal beauty trends, and product preferences. It leverages data from customer interactions across the Sally Beauty website, app, and loyalty programs to generate tailored product recommendations and targeted marketing campaigns, helping to better meet customer needs and boost sales for both consumer and professional segments.

2. Functional Requirements

2.1 User Tracking and Data Collection

• Track Customer Searches: Monitors and stores search terms, applied filters (e.g., "for professionals," "sulfate-free"), and time spent on products.
• Track Customer Purchases: Records product details, quantity, price, purchase channel (online/in-store), and date of purchase.
• Maintain Customer Profiles: Creates and updates real-time profiles based on all interactions, distinguishing between general consumers and licensed professionals.

2.2 Trend Analysis and Recommendations

• Integrate Trend Data: Pulls in seasonal beauty and hair trends (e.g., "glass hair," "balayage maintenance") to inform recommendations.
• Trend-Based Recommendations: Suggests products that align with current trends and a customer's specific profile and hair/beauty type.
• Behavior Analysis: Recommends new or complementary products (e.g., toner for a customer who buys bleach) based on search and purchase history.

2.3 Personalized Marketing & In-Store Assistance

• Dynamic Customer Journeys: Generates personalized product carousels, email content, and targeted promotions for each customer.
• Targeted Marketing Campaigns: Creates audience segments for specific campaigns (e.g., "customers interested in vegan hair care").
• In-Store Associate Tools: Provides store associates with a tablet interface to look up a customer's profile (with consent) and offer data-driven, personalized recommendations on the spot.

2.4 User Interface and Integration

• Business Intelligence Dashboard: An easy-to-use interface for marketing and merchandising teams to view customer segments, trend performance, and campaign analytics.
• Personalized Recommendations on Site/App: Dedicated, dynamic sections on the website and mobile app for each customer showing recommended products and content.
• Associate Assist Interface: A simplified view for in-store tablets designed for quick customer look-up and product recommendations.

2.5 Reporting and Analytics

• Campaign Performance Tracking: Measures the effectiveness of personalized recommendations and marketing campaigns through metrics like conversion rates, AOV, and engagement.
• Trend Performance Analytics: Analyzes the sales performance and adoption rate of trend-aligned products.

3. Technical Architecture

3.1 MCP (Model Context Protocol) Architecture

• Contextual Engine: Uses MCP to maintain a dynamic context for each customer interaction, ensuring that recommendations are always informed by the latest behavior, inventory levels, and trends.
• Data Ingestion: Supports real-time data ingestion from customer interactions (web, app, POS) and external trend sources.
• Multi-Channel Processing: Integrates data from various channels (web, mobile, in-store, email) to create a holistic customer view.

3.2 Data and AI

• Data Storage: Utilizes a secure, cloud-based solution (e.g., AWS, Azure) for storing all relevant customer and transaction data.
• AI & Machine Learning: Leverages AI models with contextual memory and recommendation algorithms (e.g., collaborative filtering, content-based filtering) to personalize the shopping experience.
• Product Catalog Integration: Syncs with the Sally Beauty product catalog in real-time via APIs to ensure data accuracy, including stock levels and professional-only product flags.

4. MCP Server Implementation

This server exposes a set of tools and resources that AI models can use to fulfill the functional requirements.

Potential Tools:

• get_customer_profile(customer_id): Retrieves the complete profile for a customer, including their segment (e.g., consumer, professional).
• get_customer_recommendations(customer_id, context): Generates personalized product recommendations for a specific context (e.g., 'homepage_widget', 'email_campaign', 'in_store_assist').
• get_trend_recommendations(customer_id, trend_name): Suggests products based on a specific trend tailored to the customer.
• update_customer_profile(customer_id, new_data): Updates a customer's profile with new interaction data from any channel.
• get_targeted_audience(segment_criteria): Generates a list of customer IDs for a marketing campaign based on specified criteria.

Potential Resources:

• resource://trends/seasonal: A resource providing the latest beauty and hair trends.
• resource://products/catalog: The complete, up-to-date Sally Beauty product catalog.
• resource://customers/{customer_id}/purchase_history: A dynamic resource to view a customer's purchase history.
• resource://customers/{customer_id}/profile_type: A resource indicating if the customer is a general consumer or a licensed professional.

5. Running the Server

To start the MCP server, execute the main Python script from your terminal. The server will start and be ready to accept connections.

python main.py