product-recommendation

name	product-recommendation
description	Personalized product recommendation strategies based on customer data
version	1.0
authors	["Commerce Team"]
tags	["recommendation","personalization","customer","shopping"]

Product Recommendation Skill

When to Use

• Customer asks "what should I buy?"
• Need to suggest products proactively
• Have customer preference data available
• Building a personalized shopping experience

Recommendation Strategies

1. Preference-Based

Use customer profile data:

- Preferred categories
- Price range (min/max)
- Favorite brands
- Past purchases

2. Purchase History Analysis

IF customer bought PROD-001 (headphones) THEN
  → Suggest PROD-005 (wireless charger) - complementary
  → Avoid PROD-007 (speaker) - similar category, redundant

IF customer bought multiple fitness items THEN
  → Prioritize Sports & Outdoors category
  → Suggest fitness accessories

3. Loyalty-Tier Optimization

| Tier     | Strategy                              |
|----------|---------------------------------------|
| Bronze   | Focus on value, build relationship    |
| Silver   | Introduce mid-range options (5% off)  |
| Gold     | Premium suggestions okay (10% off)    |
| Platinum | Lead with best options (15% off)      |

4. Collaborative Filtering (Simple)

"Customers who bought X also bought Y"

Electronics buyers → Often want: Accessories, Wearables
Fitness buyers → Often want: Apparel, Water bottles
Eco-conscious → Cross-category sustainable products

Recommendation Flow

1. GATHER context
   └─ Check customer preferences (if available)
   └─ Review purchase history
   └─ Note loyalty tier
   └─ Identify stated need

2. FILTER candidates
   └─ Match categories
   └─ Apply price range
   └─ Check brand preferences

3. RANK options
   └─ Rating (quality signal)
   └─ Relevance to stated need
   └─ Complementary to owned items
   └─ Avoid recently purchased similar items

4. PRESENT recommendations
   └─ Top 3 picks with reasons
   └─ Include one "stretch" option if appropriate
   └─ Note any loyalty discounts

Personalization Signals

Strong Signals (High Weight)

• Explicit preference (user stated)
• Recent purchase (last 30 days)
• Repeated category purchases

Moderate Signals

• Preferred brands
• Price range boundaries
• Category interests

Weak Signals (Supplementary)

• General loyalty tier
• Account age
• Review engagement

Recommendation Templates

For New Customers (No History)

Based on what you're looking for:
1. [Best rated in category] - Most popular choice
2. [Mid-range option] - Great value
3. [Budget option] - If price is key

Tell me more about your preferences to refine these!

For Returning Customers

Based on your preferences and past purchases:
1. [Category match] - Matches your interest in [category]
2. [Brand match] - From [favorite brand]
3. [Complementary] - Pairs well with your [past purchase]

Your [tier] discount: [X]% off applies!

For Specific Need

For [stated need]:
1. [Best match] - [Why it fits]
2. [Alternative] - [Trade-off explanation]
3. [Premium option] - If you want the best

My top pick: [Product] because [reason].

Cross-Sell Opportunities

Purchased	Recommend
Headphones (PROD-001)	Charging pad (PROD-005)
Fitness Watch (PROD-003)	Running shoes (PROD-006)
Running Shoes (PROD-006)	Water bottle (PROD-004)
Yoga Mat (PROD-008)	T-shirt (PROD-002)
Speaker (PROD-007)	Headphones (PROD-001)

Avoid Recommending

• Products customer already owns
• Products outside their price range (unless asked)
• Same category as recent purchase (redundant)
• Products with low ratings (< 4.0)

Quality Checks

Before presenting recommendations:

1. Does each pick have a clear reason?
2. Is the price within their range?
3. Have we avoided redundancy with owned items?
4. Is there variety in the options?
5. Have we considered their loyalty discount?