Player churn is the single largest drain on casino revenue that operators can actually control. Acquiring a new depositor costs several times more than retaining an existing one, yet most operators still rely on instinct rather than data when deciding which players to target and how to measure whether their win-back efforts are working. Churn prediction models and win-back campaigns are only as valuable as the KPIs used to evaluate them.
What Churn Prediction Actually Measures
Churn in a casino context is not binary. A player who deposits once a week does not churn the same way as a daily slots grinder. Before building any predictive model, operators need to define churn thresholds by segment. A common framework uses three tiers:
- At-risk: No session in the last 7 to 14 days for high-frequency players, or 21 to 30 days for casual players.
- Dormant: No deposit in 30 to 60 days, depending on the segment baseline.
- Churned: No activity for 90 or more days, with declining probability of return.
Prediction models ingest behavioural signals, including session frequency decline, shrinking average bet size, narrowing game variety, reduced bonus uptake, and shorter session durations. Each signal carries a weighted score. When the combined score crosses a threshold, the player enters a win-back queue. The model is only useful if those thresholds are calibrated to your actual player base rather than copied from a generic benchmark.
The Core KPIs for Churn Prediction Accuracy
A churn model must be measured before any campaign spend is committed. The key model-level KPIs are:
- Precision: Of all players flagged as likely to churn, what percentage actually churned? Low precision wastes campaign budget on players who would have stayed anyway.
- Recall: Of all players who actually churned, what percentage did the model flag in advance? Low recall means you miss genuine leavers.
- F1 Score: The harmonic mean of precision and recall, providing a single accuracy benchmark. Target an F1 above 0.70 for a production model.
- Lead time: How many days before confirmed churn does the model flag a player? A lead time of seven to fourteen days gives the retention team a practical intervention window.
Win-Back Campaign KPIs That Connect to Revenue
Once a campaign is live, model accuracy becomes secondary to commercial outcomes. Track these metrics by campaign cohort, not in aggregate, so you can isolate what works by segment and offer type.
- Reactivation rate: The percentage of targeted dormant players who make at least one deposit within the campaign window, typically 7 to 30 days. Anything above 10 to 15 percent for a cold segment is a strong result.
- Cost per reactivation (CPR): Total campaign cost, including bonus value, divided by the number of reactivated players. Compare CPR against the average first-month GGR of a reactivated player to confirm positive unit economics.
- Reactivated player GGR at 30 and 90 days: A player who deposits once after a bonus and disappears is not a genuine win-back. Measure GGR at both intervals to separate genuine returners from bonus hunters.
- Retention rate post-reactivation: What percentage of reactivated players are still active at 60 and 90 days? This reveals whether the win-back offer matched the player's actual preferences.
- Bonus abuse rate: The share of reactivated players whose bonus cost exceeds their generated GGR. A rate above 20 percent signals that offer mechanics need tightening.
Structuring Campaigns for Measurable Results
A controlled test-and-learn structure is essential. Split each churn segment into a treatment group receiving the win-back offer and a holdout control group receiving nothing. The reactivation rate differential between the two groups is the true incremental lift of the campaign, removing the noise of players who would have returned organically.
Measuring win-back success without a holdout group systematically overstates campaign performance and leads operators to over-invest in bonus spend that generates no incremental revenue.
Offer sequencing also matters for KPI tracking. A three-step sequence, starting with a personalised email, followed by a free spin or matched deposit offer, and closing with a time-limited escalation, produces cleaner attribution than a single broad promotion. Each step should be tracked separately so the team knows at which point dormant players re-engage.
How OnlineShine Approaches Churn and Retention Measurement
At OnlineShine, we work with operators to define segment-specific churn thresholds, build holdout-controlled campaign structures, and establish dashboards that report CPR and 90-day GGR side by side. The goal is to replace intuition with a repeatable, auditable process that links every euro of retention spend to a measurable commercial outcome. For operators running on lean CRM teams, having a clear KPI framework is what separates sustainable retention from expensive guesswork.
