BetBuddy chief executive Simo Dragicevic explains how operators can develop a proactive responsible gambling strategy in the face of increasing regulatory scrutiny.

The recent Remote gambling and software technical standards consultation published in June 2017 shed further light on the expectations of the Gambling Commission, the UK regulator, on what is expected from industry in evidencing a proactive Responsible Gambling (RG) strategy.

Whilst some in the gambling industry questioned whether the recent tightening of regulations to protect consumers were justifiable, proportionate, and evidence based, the Commission's response was clear – data needs to be used for non-commercial reasons as effectively as it is for commercial reasons. The Commission stated again that it expects the industry to draw on the significant data stores used to measure and evaluate the effectiveness of gambling management controls, explore potential enhancements, and to share good practice. Indeed, the Commission stated that in interpreting the available evidence, it will take a precautionary approach. For example, caution may be justified where evidence is mixed or inconclusive, and the Commission would not want to restrict its discretion by requiring conclusive evidence that something was unsafe before taking measures to restrict it.

This clearly puts the onus on the industry to proactively use player data to better protect vulnerable or at-risk consumers. This regulatory requirement is not as black and white as previous ones, such as self-exclusion, as there are different methods and approaches that can be adopted to better utilise data for RG. To some in the industry, this may be seen as an opportunity to stall on any significant RG improvements and to keep to the status quo, in the expectation that it will be difficult for the regulator to define what good looks like in this space. However, this could turn out to be a precarious strategy as competitive dynamics will undoubtedly see operators investing significantly to raise the RG standards bar, exposing those who do not to potential fines and sanctions.

What is proactive RG?

So what are some of the things a regulator will look for in terms of evidencing a proactive RG strategy? Basically, this means evidencing a real commitment to integrating RG within all aspects of the operator's business. This means committing to undertaking a long-term approach to RG and not implementing knee-jerk or sporadic efforts, or focusing solely on above the line RG advertising. An operator needs to be ready and willing to invest R&D efforts into RG and to ultimately be willing to modify practices according to any new evidence that may be obtained from their efforts. Evidencing this will be critical in giving regulators the confidence that the industry can be trusted to lead genuine efforts to raise standards in RG.

The risk continuum and widening the net

At the European Lotteries Congress in June 2017, Playtech and BetBuddy presented some lessons learned from a pilot project undertaken this year using data for understanding at-risk behaviours. One of the key takeaways was that to enable operators to better target moderate risk players, operators need to test and adopt a variety of techniques to assess player behaviour, in advancing the current traditional, rules-based systems that many operators use to flag player risk. Whilst these approaches are effective at flagging players at the extreme of the risk continuum, they are not as effective in enabling operators to broaden the net to identify players in the moderate and high risk categories (whilst at the same time not increasing the false positive rate too high). To get over this hurdle, operators need to adopt more advanced analytical techniques that leverage machine learning technology, which enable more complex, non-linear behavioural patterns to be identified.

Why is this important? Let's take a look at Sweden, which is expected to regulate online gaming in 2018. Results from the 2015 Swelogs (Swedish Longitudinal Gambling Study) study into problem gambling found that one in four (~27%) of online casino players suffer from problem gambling (using PGSI 8+ as the threshold for defining problem gambling). If the regulator demands operators put into place proactive and integrated RG, which it is expected to do, then operators will need to evidence they are working to detect and interact with those players who are at risk (as well as those very high risk cases).

At this point we think that there are excellent opportunities in combining both traditional rules-based systems with machine learning models to provide the most effective solutions to help operators identify those players at risk of harm. Developing and deploying intelligent risk detection algorithms is a first step in evidencing a proactive RG approach to using data to regulators. Just as challenging as detection is knowing how best to use these insights to better manage the player journey.

Interacting with your players

One of the main ways in which operators can evidence proactive RG is through targeted communications with players. There is a growing body of research that suggests that personalized messages have shown to change behaviour in several areas such as smoking cessation, physical fitness, reduction in alcohol consumption, and diabetes management. Also, the use of self-guided online intervention, comparing and contrasting peoples' actual gambling behaviour (for example, frequency, expenditures, time on machine) with their perceptions of others' gambling behaviour and most importantly others' actual, monitored playing behaviour, can help individuals modify their behaviours through their cognitive restructuring.

Operators who have effective risk assessment analytics in place are now well placed to start deploying and trialling these best practices in the field and to start evidencing a proactive RG strategy. There are a number of considerations that will inform the operators' RG interaction approach – what message type should be adopted (short, long, personalised, normative), when and what frequency should messages be sent, which channel should be adopted (such as SMS, email, the gaming portal, telephone), and how should the operator fine-tune their marketing strategies given that RG behavioural insights are now available?

Operators will also need to demonstrate to the regulator the effectiveness of RG initiatives. These can be relatively simple analytical assessments that identify underlying behavioural changes following a series of RG interactions with the player. More advanced Randomised Control Trials (RCT) can be designed to provide more compelling evidence of RG effects, and also have the advantage of being publishable in peer-reviewed journals if the operator wishes to do so (the stand-out gambling suppliers and operators will do this). RCTs generally require large sample sizes to produce statistically significant results, and are more suited to medium/larger operators, especially if multiple hypotheses are to be tested e.g., the impact of personalised v. normative messaging, the impact of RG interactions on VIP v. non-VIP at risk play, the impact of message channel on changing behaviour, etc.. Also, the assumptions underpinning the RCT have a major impact on sample sizes required e.g., cautious assumptions for how many people would both read an RG message and be influenced by it tend to increase the sample sizes.

It is working? A case study from OLG

At the Responsible Gambling Council Discovery conference in April 2017, we presented some early results of the effects of customer RG interactions on underlying behaviour using anonymized data from Ontario and Lottery Gaming Corporation's (OLG) PlayOLG site, which offers internet casino and lottery. PlayOLG provide mandatory self-assessments and behavioural risk profiles for their players.

During 2015/2016 we recorded around 97,000 Responsible Gambling Interactions (RGIs) on PlayOLG; players who either viewed their risk profile or self-test result. We wanted to know if these RGIs had a positive impact on player behaviour and initially focused on a small subset of RGIs that met certain criteria such as players who used RGIs regularly (at least quarterly) and who were rated either moderate or high risk at the time of the RGI. We tested the underlying behaviour (bet amount) in a period leading up to the RGI and a period after it.

The results showed that 24 per cent of RGIs in these conditions led to a moderation in underlying behaviour (i.e., improvement), 64 per cent had minimal impact, whilst 12 per cent led to an increase in underlying behaviour. Whilst we cannot generalize from these results, they evidence indicates that RGIs can potentially help some players. Also, this data can be used to inform further analyses, experimentation, and testing e.g., understanding if there is there something distinct about the 24 per cent of players who moderated their behaviour after the RGI v. those that increased behaviour, whether more effective channels, or combinations of channels, can improve on the 24 per cent of players who moderated their behaviour.

Putting the customer at the heart of business

Whilst industry initiatives such as NOSES will help in targeting those players who are already experiencing harm from gambling, they can’t help those showing the early signs of risk, a much larger population segment. The fate of the UK Fixed Odds Betting Terminal (FOBT) business is awaiting the conclusions from a government review into stake size. Despite significant industry investment in various initiatives, including implementing generic RG warning messages, implementing industry-wide messaging trials, and investing in RG social marketing campaigns, they arguably fall short in proactively addressing the vast majority of FOBT players who gamble anonymously. The industry appears to accept this has now become not a question of whether stake size will be reduced, but rather by how much.

Despite these genuine industry efforts to improve RG standards, the fact that a stake reduction is imminent is evidence of either i) FOBTs are a high risk product that need to be restricted or ii) the UK bookmakers haven't gone far enough in convincing the government and stakeholders that they can be trusted to implement measures that proactively help the vast majority of their customers. The use of anonymous play data and artificial intelligence methods could go a long way in building intelligence in FOBTs that can better help those intensive gamblers to manage and moderate their play as they get exposed to risk throughout the customer journey, compared with the current suite of RG features which treat all gamblers as the same.

In the eyes of many stakeholders the industry will never be able to do enough to protect consumers. The days of a tick box compliance approach and for the industry arguing that there isn't sufficient evidence to prove their products are harmful are clearly over. The effective use of data to inform RG strategy, processes, and interventions will become increasingly the most effective lever for industry to argue for greater freedoms to innovate in product development and marketing.

At the City, University of London and BetBuddy Responsible Gambling Algorithms roundtable event in July 2016, we heard from a former Barclays CEO that culture is the key to integrating a consumer first approach and that banking today is moving away from the mind-set of individual transactions to customer journeys and 'life events', with the aim of understanding and helping customers through their life journeys (as opposed to selling a product). The future gambling industry winners will be those that are not only allowed to continue to build and market great and compelling gambling products that customers love, but also those that evidence a serious, genuine, and long-term commitment to putting the consumer interest at the heart of everything they do. 

Simo Dragicevic is chief executive of Bet Buddy, developer of a patent-pending behavioural identification and modification platform that provides unparalleled direct marketing and responsible gaming features for lotteries and operators.

 

Related