Data Analytics In Gambling: A Double Headed Coin

Data in gambling has a bad reputation, but is also being used for good


In 2014, British bookmaker, Ladbrokes, released its ‘Ladbrokes Life’ campaign, which portrayed five average ‘blokes’ bouncing joyfully around their average lives. They revolved primarily around playing football, drinking, and gambling, only one of which is good for you. This kind of mildly irritating advert is par for the course in the $500 billion gambling industry, which court young adult males with appearances from ‘legends’ such as Chris Kamara (former pro soccer player) and Shane Warne (former pro cricket player). As young adult males are the most likely demographic to regularly gamble, their motivation for targeting this group is understandable, if troubling. It’s estimated that 3-5% of people who gamble develop an addiction to the activity, which can lead to an array of problems, not only for gamblers and their families, but society as a whole, with the social costs thought to exceed $4 billion.

In order to ensure that people keep gambling, casinos and bookmakers, both online and in their bricks-and-mortar operations, rely heavily on personalized marketing and understanding customer behavior. Andy May, brand research and retail marketing director at Ladbrokes, said of the Ladbrokes Life campaign: ‘It’s time to make a real statement and say to customers that Ladbrokes understand you, knows what you like and how you bet.’ As in other sectors, the gambling industry relies heavily on big data analytics to understand its clientele. Casinos have long looked at big data to understand potential big customers. A 2001 article in Time magazine claimed that, in the 1990s, many would buy records from credit-card companies and mailing lists from direct-mail marketers that contained the names of people who demonstrated, as one reported titled the ‘Compulsive Gamblers Special’ put it, ‘unquenchable appetites for all forms of gambling.’ While casinos no longer need to resort to such brazenly unethical methods, it could be argued that those they do employ are not far removed. Data is now collected by the casino from loyalty cards and cameras placed artfully around the pit, watching every aspect of player behavior which is then used in everything from targeting offers to positioning games machines, with pit bosses employing similar tactics to supermarkets in putting them where they can maximize earnings.

One of the most effective casinos when it comes to big data is Caesars Entertainment, with data from the Total Rewards loyalty program it introduced 18 years ago estimated to be worth in excess of $1 billion. This dataset contains information on more than 45 million customers, with a team of 200 employed to analyze it. The scheme sees customers given rewards including meals, room upgrades, and tickets to shows, advancing through rewards tiers as they spend more. In return, Caesars gets a wealth of data around how the customer behaves while at their resorts.

The implications for analytics to be used in online gaming are even greater, with digital touchpoints across sites making it easy to monitor customer behavior for every second they are logged on, and target them for offers in real time accordingly. Equally, while this data can be used to pinpoint problem gamblers to exploit them, it can also be used to help them, as a number of bookmakers are doing, but more need to in the future.

One major project has seen five of the largest UK bookmakers - Betfred, Coral, Ladbrokes, Paddy Power, and William Hill - make their industry data available to the Responsible Gambling Trust. More than 10 billion gaming machine events were analysed over a ten month period to identify behaviors in problem gamblers, with machine learning algorithms employed to measure the predictability of each individual player’s interactions. An adaptive behavioural analytics approach was used to build statistical profiles of ‘normal’ patterns of behaviour for each anonymous customer, pinpointing anomalies that indicate precisely when a gambler’s behavior starts to change. An automated system uses this to understand, in real-time, if the change indicates a player being at risk of harmful play. When the algorithm identifies someone at risk, the operator could potentially send personalised individual interventions to reduce the effects of gambling-related harm.

The research identified 19 potential markers of harm. These included the frequency of gambling and how individuals behaved while using the machines. The resulting model was found to be 66% better at detecting players at risk of harm, compared to the industry standard. Other similar projects by companies such as BetBuddy, which was able to forecast problem gamblers to 87% accuracy. As the gambling industry grows and evolves, casinos and bookmakers have a responsibility to identify the minority of gamblers who can’t control themselves and help them, not target them for advertising and encourage them to keep gambling. Data science is vital for finding problems. Nine independent studies have shown that problem gamblers generate anywhere from 30-60% of total gambling revenues, and you’d be hard pressed to find a CEO willing to spurn that proportion of income. However, casinos need to realize the long term benefits of having a business that encourages people to treat gambling as a fun activity, not as a way of life. 

Vision small

Read next:

Big Data Forecasting In Pharma