expanly-scoring-model

# Expanly Scoring Model Reference

This skill provides definitive knowledge on Expanly scoring models - the business logic, technical architecture, and implementation patterns for product prioritization in dynamic remarketing campaigns.

---

## What is an Expanly Scoring Model?

An Expanly scoring model categorizes products into **priority tiers** for advertising campaigns. Products are assigned a **score (0, 1, or 2)** based on their performance metrics, which determines their campaign allocation and bid strategy.

### Business Purpose

**Problem**: Advertisers waste money showing ads for:
- Out-of-stock products
- Poor-performing products
- Products with negative ROAS

**Solution**: Score products based on performance signals and segment them into campaigns with appropriate bid strategies.

### Score Tiers

| Score | Priority | Campaign | Bid Strategy |
|-------|----------|----------|--------------|
| **0** | Low | Low-priority | Minimize spend |
| **1** | Standard | Default | Target ROAS |
| **2** | High | High-priority | Maximize conversions |

**Important**: Score = 0 products are NOT excluded from advertising. They go to a low-priority campaign with reduced bids. This ensures coverage while minimizing waste.

---

## Rule System

### Rule Categories

#### Exclusion Rules (Score = 0)

These rules identify products that should receive minimal ad spend:

| Rule | Name | Condition | Rationale |
|------|------|-----------|-----------|
| 0 | Negative ROAS | `gads_cost > 10 AND gads_revenue < gads_cost` | Losing money on ads |
| 1 | Zombie | `gads_cost > 5 AND revenue = 0 AND views = 0` | No engagement at all |
| 2 | Zero Revenue | `gads_cost > 15 AND revenue = 0` | High spend, no sales |
| 3 | Out of Stock | `is_in_stock = FALSE` | Can't fulfill orders |
| 4 | Low Availability | `variant_availability_rate < 0.5` | Most sizes unavailable |

#### Default Rule (Score = 1)

| Rule | Name | Condition |
|------|------|-----------|
| 10 | Unclassified | No other rule matc