Exploring the inner workings of Ibex: A deep dive into the GGR & Cost Models

Connor Coutts
April 11, 2023
Data Science
Back to blogs
In this third instalment of our series exploring the inner workings of Ibex we will be focusing on the GGR model and cost models.
With Ibex now being able to determine which customers can be contacted, the predicted deposit rate impact and the deposit amount uplift, Ibex moves onto analysing the expected revenues generated from players and the cost implications of sending recommendations.
GGR Model
The GGR Model predicts the expected Gross Gaming Revenue (GGR) for each pound deposited by a player. This is calculated by multiplying the output of the deposit rate model, deposit amount model and an estimate of the expected GGR %.
GGR is estimated using historic values for GGR / Deposits. Ibex considers a number of factors when determining the expected GGR%:
  • The type of offer (if there is one) – some offers tend to increase GGR %, while others do not (or even might lead players to decrease staking and thus GGR %)
  • The country / region
As GGR data is very noisy (big wins significantly change this analysis), Ibex is also removing outliers from this analysis by winsorising the data. The process of winsorisation removes the most extreme 5% of instances before the GGR figures are aggregated which results in smoother estimates of the expected GGR % for each action type and region.
Bonus Cost Model
The Bonus Cost Model predicts the cost of bonuses for each player based on the action type and bonus mechanics configured in the Ibex portal.
Ibex supports various types of bonuses, including match bonuses, free spins, cashback, free bets, and free credit as well as combinations of them.
When an action is configured, the user selects an “Action type” which tells Ibex the type of bonus which accompanies the communication.
Example of bonus configuration
Bonus Given
Ibex first predicts the value of bonus given to the player.
By way of an example, we’ll calculate the expected bonus given for a match bonus offer (which requires a deposit). In its most simple form we have the formula below
DA, is the predicted deposit amount for a particular player and action and factors such as minimum and maximum deposit requirements associated with the action will influence the players deposit.
Other action types are calculated in a similar manner by combining parameters with simple arithmetic operations.
Bonus Cost
Now we have an expected bonus given for a player and action combination, Ibex can calculate the expected bonus cost.
When bonuses are given to a player they are usually credited to a bonus wallet and requires the user to wager the money multiple times before it converts to cash.
The exact amount which converts to cash is a result of random gameplay, however over a large enough sample the conversion percentage can be predicted.
The two key drivers of this conversion percentage are:
  1. Return to player (RTP) – the percentage of each bet which is returned to the player on average.
  2. Playthrough requirements – the number of times the player “turns-over” their bonus money
Equipped with these two variables, Ibex can estimate the conversion percentage, mu, which allows us to predict the bonus cost.
Other cost estimates
As Ibex aims to make decisions on casino income, the income generated by the operator after all attributable costs are applied, it is necessary to estimate further costs to arrive at an expected casino income figure for each player.
When calculating the casino income for each player, Ibex considers the following factors:
These additional cost estimates will be explained in turn.
Jackpot contributions
Certain casino games have progressive contributions where a small % of a players bets are collected by the progressive slot. These type of games incur an additional cost as the contribution is deducted from revenue.
So, to account for this Ibex monitors each players jackpot behaviour by calculating what percentage of their total cash stake has historically gone towards jackpot contributions.
This historic ratio is then applied to future expected stakes to calculate the expected future jackpot contribution for each player.
Revenue share
A lot of casino operators use revenue share deals to pay for casino traffic. This is where affiliates get a percentage of NGR in order to align them with the operator: higher value traffic will result in increased revenue for both the operator and the affiliate.
This means that operators need to take this into account when calculating the expected future value from players.
Ibex keeps track of revenue share paid to affiliates as well as NGR to understand how much revenue is lost to affiliates and uses this information when calculating the expected income.
Game supplier fees
Each game comes with its own price to operators, with the more popular games coming at higher costs.
This means the players who consistently play premium brand games are incurring a much higher cost to the operator which needs to be taken into account when recommending games or picking game-based bonuses for each individual players.
Ibex keeps track of the fees associated with each game and the operator costs to take this into account.
Payment fees
Just like supplier fees, payment providers charge different amounts for their service. Usually debit card transactions cost the operator less while payment providers such as Paypal can be a lot more expensive.
Sometimes the difference in these fees can turn an action from being profitable to unprofitable so Ibex also takes this into account.
Ibex keeps track of which payment provider each player uses, as well as the payment fee for each payment channel. This is then used to calculate the fee we expect from future deposits to make the casino income calculation even more accurate.
These extra components gets Ibex ever closer to making a recommendation for a player. So, in summary Ibex can now determine:
  1. Which players can be contacted with marketing actions on a given day
  2. The predicted deposit rate impact of sending marketing actions
  3. The predicted deposit amount uplift
  4. The expected GGR generated
  5. All of the costs associated with a marketing action
In the final article of this series we will discuss the final three steps Ibex takes to arrive at a decision which are: the lifetime value model, experimentation module and control group module.

Grow your business

Use Ibex to drive customer engagement and deliver the right messages at the right time
Start Now