New version of DupZapper combines machine learning and behaviour analysis to predict customers' behaviour by analysing players' betting activity, website and app usage patterns, payment and withdrawal preferences, detecting duplicate registrations along with other forms of collusion and advantage betting.
This technology is a giant leap forward from the legacy industry-standard fingerprinting techniques that are based on device identification, permitting even companies that already employ device-based customer identification technologies to detect advanced fraud in the absence of useful device fingerprinting data, as well as boost marketing and CRM efforts by analysing and predicting players' preferences detecting new VIP customers and professional gamblers before they make their first bet.
Rather than merely identifying fraudulent customers, the solution boosts productivity of the CRM team enabling it to analyse many more customers in greater detail.
The system fingerprints customers by measuring tens of thousands of player traits, most of which are specific to the online gambling industry:
|Betting Behaviour||Usage Behaviour||Payment Behaviour|
|Preferred leagues, sports, teams and championships||Typing and browsing speed||Payment methods used|
|Share of sports bets vs casino||Browsing activity that takes place before placing a bet||Amount of deposits|
|Live and pre-match betting behaviour||Interest in content available on the website||Time between depositing money and placing bets|
|Preference towards market types||Operating system, browser and language||Relative amount of deposit compared to applicable limits|
|Amount of time between placing bets and game/period finish||Use of recommended bets||Withdrawal behaviour: how much money player usually withdraws and when.|
|Risk/payout preference (sure bets)||Behaviour after win/loss||Bonus offerings and coupons|
|'Shopping' behaviour: single vs multiple bets||Behaviour after rejected/unavailable bet placement attempt: choosing another bet or leaving the website||Most popular time of the day, week and month for depositing|
Independent variables from these categories are processed to form a fingerprint that is then compared with past CRM data such as player LTV and future behaviour of players that had similar fingerprints in the past to detect likely future behaviour of the given player, groups of players exhibiting likely collusion or advantage betting as well as customers that appear similar to the ones who became VIPs.
Due to the amount of insight processed by the behaviour fingerprinting solution its implementation is always tailored to take advantage of all behaviour data that is available in the clients' platform, which varies depending on the betting product offering, website and mobile platforms user interface and payment and CRM information available.