Safety net: How GiG is using AI to predict self-exclusion and what this means for the treatment of at-risk gamblers

How do you effectively identify a problem gambler? The question which is on the lips of responsible gambling professionals from Australia to the US. Every market, operator and treatment professional have their own opinion on how best to achieve this, all with varying success.

However, for operators the trick is cutting through the hyperbole and technology to find a solution that works and can be integrated quickly and effectively with your existing software. Much of this involves developing new techniques, such as the ones currently being trialled by Gaming Innovation Group (GiG) to identify players who may be at risk of gambling related harm.

Martin Pullen, head of player safety at GiG, discusses its approach to identifying problem gamblers and how insights into self-exclusion might light the way for operators.

EGR Compliance: Can you tell me about the work that the in-house player safety team does on a day to day basis?

Martin Pullen (MP): The player safety team (PS) are our frontline entity on responsible gambling. Their remit is to proactively monitor player accounts in real time, to highlight possible areas of concern and sudden spikes of player behaviour and betting patterns. Our strategy is to interact as early as possible, to ensure our customers are being given an informed choice of the various RG tools available and to safeguard against an escalation of potential harm. Once highlighted, players are interacted with, using a variety of communication channels such as phone, email and personalised on-site messaging.

EGR Compliance: What training do the in-house player safety team receive in respect of spotting problem gamblers?

MP: All our player safety advisors receive an in-depth and comprehensive training programme, concentrating not only on analysing and acting on financial triggers, but also the behaviour and psychology behind the customers at risk of developing harmful gambling patterns. In addition to an ongoing training schedule in-house, the PS advisors are given accredited training from GamCare on responsible gambling and motivating behavioural change. We are well aware of the diversity of potential gambling harm, not only in terms of resources but also the effect on relationships and health. Therefore, we regularly train our staff on supporting customers that may be experiencing a potential crisis and proceed with referrals to charities and support organisations when needed. We are always looking at avenues to expand our knowledge and understanding of player behaviour and pattern to provide a safe and secure gaming environment.

EGR Compliance: How important is it that every member of the player safety team understands their organisational responsibility when it comes to responsible gambling?

MP: We have been very successful in our recruitment of the team, with members who are fully dedicated and personally committed to providing a safer environment for our players. The motivation behind each of our advisors is to enhance the protection levels for our players, coupled with a reduction in regulatory risk. As individuals, we are very proud of GIG’s reputation across the industry in terms of working environment, product range and customer experience.

EGR Compliance: At what stage is the trial and when do you expect to complete it?

MP: The trial is based around an ML and AI solution that we developed in-house and will consist of three stages, with analysis and study of data at each stage. The first stage was to test the effectiveness of the model itself, without any interaction performed by the player safety team. We were very pleased with the results, with a high percentage of accounts indicating problem gambling, flagged by the model very early in the customer lifecycle.

The second stage consisted of personal interactions conducted on highlighted accounts on the third depositing day. We are currently collating the results of this second stage to assess the effectiveness of the interaction methods used. The final stage is to implement automated on-site messaging prior to any required interaction. After all stages of testing have been completed, we will be in the best position to ascertain which interaction methods have produced the best performance, in terms of use of RG tools and more sustainable accounts.

EGR Compliance: How does this feed into GiG’s wider sustainability strategy?

MP: GiG has a specified intent to make online gambling fun and fair and one key strand of that strategy is to create a safe, enjoyable and sustainable gaming service to our customers. Our approach to player safety not only heightens player protection, it also promotes a more sustainable customer base, through increased use of responsible gambling tools and customer self-awareness. As acquisition of players becomes more difficult due to the regulatory climate, it is increasingly important to manage player accounts more effectively. Pedro Romero, a well known and established expert in this field, started working with us recently and will be supporting the business on our sustainability and safer gambling strategy.

EGR Compliance: Why does the algorithm target the probability of self-exclusion specifically?

MP: The aim is to highlight and interact with players that have the potential to develop issues with gambling in general and self-exclusion is purely used as a parameter to measure that group. Although, in line with standard industry practice, we have stringent monitoring processes and self-exclusion facilities available for all our players who feel their gambling has become an instrument of self-harm. We feel that a successful RG strategy should also have the ambition of motivating behavioural change within players. There is plenty of published research which confirms that players who have set deposit/wagering limits, for instance, are less likely to engage in harmful betting activity that eventually leads to problem gambling and a need for self-exclusion. By incorporating data from previously self-excluded accounts into the model, we can highlight patterns that may lead a player down the road to problem gambling at a later stage.

EGR Compliance: What factors does the predictive model look for and how were these identified?

MP: The model looks at three initial areas of data: the person behind the player, value segmentation and spending metrics and rational payments and casino activity metrics. Each area is then broken down into hundreds of predictors. As a simplistic guide, we look at who the player is, age, sex, area, registration details, device and timing etc. We also take into account deposit amounts and regularity, payment methods and patterns plus time of gameplay and casino patterns. All these predictors have been measured against our historic database of self-excluded accounts from all our core markets over the last five years, and corresponding trends and patterns that are associated with self-excluded accounts are identified and measured. To give one small example, a predictor which featured on self-excluded accounts we encountered on our initial model research, was live casino players where around were one third more likely to self-exclude than the base line.

EGR Compliance: Will the predictive model eventually be rolled out on GiG’s B2B brands?

MP: While the model is still in continuous development and testing, we are always looking for potential opportunities to support our B2B partners’ product and gaming experience.

EGR Compliance: What actions are performed in respect of the accounts highlighted for a responsible gaming review by the in-house player safety team?

MP: The model flags players per day over the first three days. For day one and two, account holders will see an automated on-site message, promoting our various RG tools, populated on the website. The message will be play interrupting, with the user having to “acknowledge” the message before continuing. For day three, our player safety advisors will fully review the account and personally interact by various channels, even if high spending parameters haven’t been triggered. The UKGC, in particular, states that RG review and interaction should not be conducted solely on financial boundaries and we agree that there are various indicators of problem gambling which must be considered.

EGR Compliance: Why is it important that the calculation regarding the probability for self-exclusion be made within the first three depositing days of the customer lifecycle?

MP: During our preliminary research for the model’s concept, we found that historically over 30% of self-exclusions occurred within the first four depositing days from registration. This is line with a pattern of players opening accounts with a quick velocity of deposits and then closing or self-excluding, possibly to repeat the process with an alternative casino operator. The model’s concept is to be able to interact with these players at a crucial early stage to connect and influence player behaviour and pattern. Again, player protection and sustainability are at the forefront of the model’s remit.

EGR Compliance: Do you plan any enhancements to the predictive model during the rest of 2019? If yes, in which areas?

MP: Even though we are still trialling the original model, we are already looking to enhance the scope and performance of this tool. We have an excellent team of data scientists at GiG, who are already working on expanding the timeline for predictions, with a view to highlighting accounts further down the lifecycle. Spikes or changes in gameplay activity and deposit patterns can indicate potential problems or changes in personal circumstances which may affect the chances of harmful behaviour.

EGR Compliance:  Would GiG consider rolling this service out to external firms, in the style of GiG Comply?  

MP: One of the core strengths of GiG is that we can develop innovative products that we are proud to use and develop on our own B2C brands, with the potential to improve the experience of customers from other operators. We understand creating a safer gambling environment is at the heart of all responsible gambling operators. We hope that we can develop a leading product that can facilitate this complex task and provide a better and safer experience across the industry.