A frequent player
A break from
the usual pattern.
An unexpected pause may deserve attention when it follows regular activity. Review the change alongside the score and the player’s recent history.
Gamblitude predictive models
Player Churn Prediction
for iGaming.
See where engagement
may be changing.
Use player behaviour and predictive churn scores to identify potential disengagement. Give retention teams a focused starting point for review, Segments and CRM planning.
Current scoreChange over timePlayer context
01 / What the model tells you
A clearer signal
for retention decisions.
Player Churn Prediction estimates the risk of disengagement against an agreed future outcome. The sportsbook configuration shown here uses a 30-day horizon and updates scores daily for eligible players.
The result brings the current score and its changes into your team’s view of player activity. The qualifying activity, outcome and eligible population are defined around your operation.
- Output
- Player-level
churn risk score - Configuration shown
- Sportsbook
30-day horizon - Scoring
- Daily, for
eligible players - Used in Gamblitude
- Attributes, Segments
and dashboards
02 / Understand the pattern behind the pause
The same gap.
A different story.
A pause in activity means more when you understand the pattern around it. How recently a player engaged, how often they usually play and how that behaviour is changing all provide context for review.
A break from
An occasional player
A gap within
the usual rhythm.
The same pause may fit a less frequent pattern of play. Looking at the wider history helps your team interpret what the gap means.
The model draws on activity history and other selected attributes to give your team a broader basis for review than elapsed time alone.
03 / Three parts of a useful review
Read the score.
Look at the change.
Add context.
Bring the assessment together with the player’s recent activity and the context of the decision.
Current score
Where to
focus now.
Use the score to identify players who meet your agreed review criteria. Set those criteria around the model’s interpretation and your team’s process.
Change over time
What deserves
a fresh look.
Compare assessments to find meaningful movement. Changes in score can support Segments alongside the current output.
Player context
What the signal
means in practice.
Review recent activity, lifecycle stage and other relevant Attributes. Combine churn risk with expected GGR when future value is useful to the decision.
04 / A broad feature store
Many signals.
A richer view of behaviour.
Gamblitude’s feature store brings together attributes describing how players arrive, engage, transact and use your products. Each model draws on a selection suited to your data and prediction target.
The families below show the range available for model development.
Sessions & engagement
Activity frequency, return patterns and changes in engagement over time.
Product usage
Product mix, content preferences and behaviour across the available offering.
Channel & acquisition
Channel, acquisition source and the context of the player’s cohort.
Transactions
Deposit and withdrawal behaviour, transaction frequency and lifecycle patterns.
Revenue & wagering
Revenue and wagering patterns that add context to player activity.
Lifecycle & behaviour
Tenure, recency, activity timing and evolving behaviour across observation periods.
05 / Individually built for every operator
Built for your players.
Trained on your history.
Every operator has a different player base, market and product mix. Each Gamblitude machine learning model is built individually and trained on that operator’s historical data.
For churn prediction, feature selection, training and evaluation reflect your player population and the outcome you want to identify. The aim is a useful signal for the decisions your retention team needs to make.
Your player base.
Learn from how your customers engage and how their activity changes over time.
Your market and products.
Build around the context in which your players use the available offering.
Your churn definition.
Agree the qualifying activity, future outcome and player population used to train and evaluate the model.
06 / A practical retention workflow
From a signal
to the right next step.
A practical sequence for turning the model’s output into a decision your team can evaluate.
Identify a signal.
Find players whose current score or change in score meets agreed review criteria.
Add context.
Review recent activity and combine the score with the relevant player Attributes.
Build eligible Segments.
Apply your contact eligibility and player protection rules before selecting an audience.
Choose a response.
Decide whether a service check, relevant communication or no contact is appropriate.
Measure the outcome.
Assess the response against a suitable comparison group and include the cost of action.
07 / Measure the signal and the response
Two questions.
Two kinds of evidence.
Churn risk helps focus attention. The effect of an intervention depends on what the team does and how players respond.
Did the model identify
the agreed outcome?
Compare scores with subsequent behaviour. Assess the quality of the selection, the review volume and whether the signal arrives in time to inform a useful decision.
Did the response
improve the outcome?
Test the campaign or service change against an appropriate control group. Review its effect alongside communication and incentive costs.
The practical details
Player Churn Prediction.
Your questions.
The target, its interpretation and how it fits your retention process.
What does the model count as churn?
The target is agreed for your operation, including the qualifying activity, outcome period and eligible population. This definition makes it possible to interpret the score and evaluate it against later outcomes.
How often are scores updated, and what period do they cover?
The sportsbook configuration shown here uses daily scoring and a 30-day horizon. The precise target and player eligibility are established as part of the model configuration.
How does this differ from a fixed inactivity rule?
An inactivity rule responds to a set gap since the last activity. Predictive modelling brings a wider activity history and other relevant attributes into the assessment, helping teams review different patterns of disengagement.
Can we use changes in score?
Yes. Scores and changes between assessments can be used in Segments. Review criteria should reflect the model’s interpretation, your team’s capacity and the intended next step.
Where can we use the result?
Scores can be used as player Attributes, combined into Segments and analysed in Gamblitude. A configured webhook can pass an eligible Segment to a connected CRM workflow.
Does high churn risk mean a player should receive a bonus?
Churn risk alone does not show how a player will respond to an incentive. Review the player context and eligibility before choosing a response. A service check, relevant information or no contact may be appropriate.
Can it be combined with Player GGR Prediction?
Yes. Churn risk and Player GGR Prediction offer different views of a player’s outlook. They can be combined in Segments to support review priorities, while the effect of any campaign is evaluated separately.
Does the model prevent churn automatically?
It predicts risk and supports decisions. Your configured workflow determines the response; a separate test establishes whether that response improves retention.
Player Churn Prediction by Gamblitude
Give your retention team
a clearer starting point.
Bring your current inactivity rules and the decisions they trigger.
Let’s discuss where a predictive signal could help your team decide what to do next.