AI-suggested Deposit Amounts

The AI-suggested Deposit Amounts feature predicts and displays tailored deposit amounts, including the default amount and alternative suggestions, based on historical data and the user’s behaviour. It is enabled per operator in the Back Office Settings.

Note

Historical data is required before the AI features can be enabled, to train the models and analyse the data.

What the predictions need

Data	Where it comes from
Country	The country attribute
Currency	The currency attribute
Previous deposit	previousDeposit in the user-data attribute

If any of these is missing, or there isn’t enough data, the cashier falls back to the predefined amounts configured in the Back Office.

widget.setAttribute('user-data', JSON.stringify({
  previousDeposit: 100,
  age: 30
}));

AI model

When the feature is enabled, the AI model option sets how the suggested amounts relate to the predictions:

Option	Suggested amounts
Low	Align closely with the predictions
Medium	Moderately higher than the predictions
High	Significantly higher than the predictions

A/B testing

A/B testing compares the AI models by splitting players between them. It overrides the Predictions for Suggested Amounts setting while enabled.

Give each model a share of the traffic: Low, Medium, High and No AI. The values don’t need to add up to 100; Fluid converts them into proportions. For example, Low 4, Medium 6, High 5 and No AI 5 add up to 20, which gives 20%, 30%, 25% and 25% of players.

Setting the A/B testing traffic split in the Back Office

For a balanced test, split the traffic evenly between the models you compare, and leave the test running unchanged for at least four weeks to collect enough data.