Skip to main content

User Loyalty Model

A framework for classifying loyal customers and analyzing brand engagement over time.

What Is It?​

The User Loyalty Model identifies which users are consistently engaged with a brand, retailer, or service using real transaction behavior. By tracking both frequency and spend, the model enables structured, repeatable measurement of:

  • Brand health
  • Customer retention
  • Lifetime value (LTV)

Why Loyalty Matters​

Loyal customers are:

  • More profitable – higher LTV and transaction value
  • Less likely to churn – better retention and advocacy
  • More valuable – stronger influence on brand performance

Carbon Arc’s model helps clients:

  • Benchmark loyalty share across competitors
  • Detect early churn signals
  • Identify high-value segments
  • Power recommendation engines and shopper overlap models

The model uses user-level transaction data to compute:

  • Cadence — how often a user transacts
  • Quantity — how much they spend or consume

These metrics are normalized and fed into an ML-based clustering algorithm to segment users into natural groupings. Each cluster is labeled as:

  • Loyal
  • Non-loyal

The model runs monthly to track trends in:

  • Loyalty growth
  • Churned loyalists
  • New loyalist acquisition

Use Case Examples​

Use CaseValue
Segment AnalysisCompare loyal vs. non-loyal spend and retention
Competitive BenchmarkingRank brands by loyalty share and stability
Customer LifecycleMonitor loyalty movement across time
Cross-Shopper ModelsIdentify adjacent brands trusted by loyal users
Recommendation SystemsRemove one-time shoppers for precision targeting

Core Output Metrics​

  • Share of Loyal Customers
  • Churned Loyalists (MoM, YoY)
  • LTV Split – loyal vs. non-loyal
  • Cross-brand loyalty overlap

Output Structure​

Each user is labeled monthly with:

  • user_id
  • date
  • brand
  • representation_type (e.g., Retailer or Service)
  • cadence
  • quantity
  • is_loyal (true/false)

Key Loyalty Metrics​

  • Retained Loyal Customers – loyal in both current and previous periods
  • Churned Loyal Customers – loyal last period, not this period
  • New Loyal Customers – first-time loyal this period
  • Total Customers – all users transacting with the brand in period
  • Churn Rate – (Churned Loyal / Loyal Last Period) × 100

Availability​

  • Monthly granularity
  • Available in select brand, service, and retailer panels
  • Delivered as certified, production-ready output for dashboards or models

Sample Output​

user_idyearbrandrepresentation_typeis_loyal
A123452024NikeRetailertrue