RFM segmentation scores iGaming players on Recency, Frequency, and Monetary value to produce a simple, durable value-and-engagement grid for CRM targeting.
RFM Segmentation (RFM)
**TL;DR:** RFM segmentation scores iGaming players on Recency, Frequency, and Monetary value to produce a simple, durable value-and-engagement grid for CRM targeting.
What it means
RFM is a 30-year-old retail framework that maps cleanly onto iGaming. Recency = days since last bet or deposit. Frequency = number of betting days or deposits in the window. Monetary = NGR or deposit total. Each dimension is bucketed (commonly into quintiles 1–5) and combined into an RFM code (e.g. 5-5-5 = best, 1-1-1 = worst).
RFM is often the first segmentation layer at smaller operators, then complemented or replaced by ML-based scores (predictive LTV, churn probability) at scale. Even when models replace RFM for decisioning, RFM remains useful as a reporting and stakeholder-communication tool.
Formula / How it's measured
Per player over a window (e.g. last 90 days):
- R = days since last activity → bucket 1–5
- F = count of active days or deposits → bucket 1–5
- M = NGR or deposit total → bucket 1–5
- RFM score = concatenated string or sum
Example: a casino player with last bet 4 days ago (R=5), 22 active days (F=5), $1,800 NGR (M=5) → 5-5-5 → "champion" segment, eligible for VIP host outreach. Another with R=1, F=2, M=2 → "about to churn low-value" → automated 50 free spins reactivation.
Why it matters for operators
RFM is cheap, explainable, and battle-tested. It gives CRM teams an immediate map of the player base without requiring data-science investment, and it's a useful sanity check on more complex models.
Common benchmarks (2026)
- 5×5×5 grid = 125 cells; most operators collapse to 8–12 working segments
- Recency windows: 7/30/90 days are standard
- Refresh: daily for active segments
- Used as primary segmentation by ~50% of small/mid operators
- Used as input feature to ML LTV models by most large operators
Common mistakes
- Equal-weighting R, F, M when product dynamics demand different weights
- Using calendar quarters instead of rolling windows — players "expire" in jumps
- Not adjusting buckets across markets (Tier 1 and Tier 3 M values differ 10×)
See also