Ask most casinos in the U.S. and Canada if they have used or are considering the use of facial recognition and the answer is often an enthusiastic “yes.” But ask the properties that have implemented some form of biometrics how the technology has performed, and the answer isn’t as clear—the surveillance team response is often mixed at best and non-security executives will answer with a shrug of the shoulders and an admittance that they do not benefit from the use.
Facial recognition has been used to detect undesirable and self-excluded customers for years, but not without some issues—the number of false positives and negatives can often outweigh the benefits of trying to catch someone who looks a bit like someone nefarious. It may also be said that the technology was just not justifiable due to cost and the results—it was a million-dollar solution to a hundred-thousand-dollar problem.
However, with technology in cameras quickly improving and artificial intelligence (AI) driving the ability to process faces more accurately, biometric identification will be the norm very soon. So that leads to the question, what is your strategy to implement this technology? And I am not just talking about the casinos, this is a question plaguing gaming manufactures as well, since slot manufacturers have already started testing cameras in their slot machines, and at least one table game manufacturer has started testing facial recognition at the table.
All in all, everyone wants to be able to recognize and pull more information about the player. That said, when faced with the question of what they want to accomplish with facial recognition, most people cannot give a definitive answer. Some say, “We want to be able to measure the mood of our players.” And others, “We want to collect demographic information.” And finally, “We want to be able to accurately rate the players.”
Truth is, most of them have no idea what to do with the customer’s face once they have captured it; they just know it is cool technology and no one wants to be left behind. You might call this the Field of Dreams belief: “If you build it they will come,” what they might do when they come is nowhere explained, or well understood.
So, while the manufactures try to figure out how to make use of the new toy in a slot machine, let’s explore how to really use facial recognition to make a difference.
First, let’s ditch the concept that you can only use facial recognition to compare against national databases, “black-lists,” “white-lists” and whatever list has been made. There is plenty that can be accomplished with just the faces and data within your casino. Don’t worry about the other 8 billion people in the world that have never been in your property. If you limit your sample to just the people you have on the property over the last year or two, you will find out that you get far fewer “false-positives” and “false-negatives.” This is not to say you can’t run faces of interest against a database to identify them, but this is not necessary for every patron and gets into privacy issues.
Second, facial recognition accomplishes nothing if you don’t have data to tell a story. OK, maybe once in while it will get a hit against someone on the most wanted list as they walk in the door. But for the most part, that sort of activity is limited, and the reason facial recognition was not successful in the past. Data is what makes the face interesting. If I walk into a restaurant and facial recognition sees me walk in and walk out, the only thing that was learned is that I was there. However, if facial recognition sees me walk into a restaurant and ties my face to a $500 POS transaction with details, you now know vastly more about me and my behaviors.
Third, facial recognition data needs to be real-time and actionable. This means that it doesn’t just talk with surveillance and security; it must be tailored to assist every department it touches and alert when thresholds are met within the data. This is supporting a “false” theory that there is some law against anyone seeing video other than the surveillance team. So not to get anybody’s knickers in a twist, let’s just assume that only surveillance has access to the faces. But a proper system should know when “customer X” has cashed out more than $10,000 in 12 hours. Instantly triggering an alert to compliance and to surveillance.
For example, Casino X places facial recognition cameras in each of their TITO redemption kiosks and at all cashier cages, then integrates with the TITO system. This allows them to receive information in real time as to the cash-out amounts of each TITO voucher. It would be the data from the amount of the TITO voucher that would indicate whether the face is worth pursuing. If that particular face only had transactions for $120 for that day, the face could be forgotten after a relatively short period of time. This assumes there was no other transaction on previous days that would be suspicious. However, if the transaction was one of multiple transactions that day in excess of $10,000 by that one customer, immediate action can be taken by multiple departments at once.
Not only would surveillance want to be alerted immediately, but compliance as well. Suspicious Activity Reports could be completed while the patron was still on the casino floor. Further research could be done with the individual’s face as the identifier that ties all transactions together. Even if this person were not carded, the face would become the key element for all transactions and data associated.
This is the biggest shortcoming of Title 31 reporting tools today: they lack the data forensics to research beyond what is already known or recorded in systems. Real-time understanding of behaviors and transactions requires not only the monitoring of the transactions, but the ability to tie multiple transactions together. This is true regardless of the distance or time between transactions. With facial recognition, we now have the key element to associate any transaction within the casino.
The issue of privacy is always a concern when contemplating biometrics and facial recognition. In the case of using facial recognition for an AML forensics tool as outlined above, it is no more intrusive that using the current surveillance system at every casino. If the person does not have activities within the data that would suggest something nefarious, the casino would discard the face and no research would be done. However, if someone has data that could arouse reasonable suspicion or proscribed activity within AML legislation, it is not only permissible to breach that person’s privacy, it is the casinos obligation.
One of the side effects of having a successful KYC facial recognition strategy is that you build up a picture of customers that aren’t currently loyalty members. This is not to say that you will be obnoxious or intrusive or even approach that customer who wants to remain anonymous, merely that they can be offered a positive opt-in.
This brings us to the next level of facial recognition—the system’s ability to help slot operators reach the 30-60 percent of their customers that remain uncarded. With a proper facial recognition strategy, properties can offer an opt-in to market to those people based on transactional history. Also, a person does not need to be carded for the casino to understand and associate the patterns of data with a face. Marketing campaigns can be tailored to monitor repeat customers that are uncarded and offer incentives to join membership programs, for returning or reinvesting larger wins, eventually leading to increased loyalty and measurable marketing success. The same facial recognition that determined that the person deserved 50 percent off at the steak house can tell an operator when that person redeemed the coupon. With the proper POS integration, it could even tell what the person ate.The old belief in facial recognition—that it is only good for various “lists”—is actually the minimum entry to such technology. Far more important is all the other things that facial recognition can enable an operator to do. By integrating with the major data sources throughout the casino, the face can create a histogram of the customers experience and tie any type of transactions together. Thus, helping all departments to benefit from the data identity of casino patrons.