Back to overview
Lesson 2 of 5

Segmentation Criteria and Models

8 min read

The value of segmentation depends entirely on the criteria you use. Segment by the wrong attributes and you get groups that look different on paper but behave the same in practice. Segment by the right attributes and your tiers will naturally reflect the economic reality of your partner base.

Primary Segmentation Dimensions

There are four core dimensions that most affiliate programs should evaluate when building segments. Not every program needs all four, but ignoring any of them entirely creates blind spots.

DimensionWhat It MeasuresExample Metrics
VolumeHow much traffic or conversions a partner deliversClicks/month, signups/month, FTDs/month
QualityHow valuable the referred customers areAvg deposit, LTV, retention rate, chargeback rate
EngagementHow active the partner is in the programLogin frequency, creative usage, communication responsiveness
Strategic ValueHow important the partner is beyond raw numbersBrand alignment, content authority, geographic reach

Volume-Based Segmentation

Volume is the most intuitive dimension. An affiliate sending 500 signups per month is fundamentally different from one sending 5. Volume-based segments are easy to define, easy to measure, and easy to communicate to partners. The risk is that volume alone does not capture quality -- a high-volume affiliate sending fraudulent or low-deposit traffic can damage your program.

Never use volume as your only segmentation criterion. A partner sending 1,000 clicks with a 0.5% conversion rate and high chargeback ratio is not a top-tier affiliate -- they are a liability.

Quality-Based Segmentation

Quality segmentation focuses on the downstream value of referred customers. In iGaming, this means tracking GGR per player, deposit frequency, and bonus-to-deposit ratios. In Forex, it means lots traded, account longevity, and funding consistency. In Prop Trading, it means challenge completion rates and repeat purchase behavior.

Quality metrics take longer to mature than volume metrics. You need 30-90 days of customer behavior data before quality scores stabilize. Build this delay into your tiering logic so that partners are not promoted or demoted based on incomplete data.

Composite Scoring Models

The most effective segmentation combines multiple dimensions into a single composite score. This prevents gaming any single metric and gives you a holistic view of partner value.

  • Weighted scoring: Assign weights to each dimension (e.g., 40% quality, 30% volume, 20% engagement, 10% strategic value)
  • Threshold-based: Set minimum thresholds on each dimension, then rank by primary metric
  • Cluster analysis: Group partners by behavioral patterns rather than predefined rules
  • Hybrid: Use thresholds for tier qualification and scoring for within-tier ranking

Start with a simple two-dimensional model (volume + quality) and add dimensions as your data matures. Overcomplicating your initial segmentation model makes it harder to explain to partners and harder to maintain.

Whichever model you choose, document the criteria clearly and make them visible to your affiliate team. If the segmentation logic lives only in one person's head, it will not survive their vacation.

Key Takeaways

  • Four core segmentation dimensions: volume, quality, engagement, and strategic value
  • Volume alone is insufficient -- always pair it with quality metrics
  • Quality metrics need 30-90 days to stabilize before they are reliable
  • Composite scoring models prevent gaming of individual metrics
  • Start simple with two dimensions, then add complexity as data matures