Airlines suffer massive loss due to loyalty frauds
Airline loyalty fraud has long existed, but its prevention is getting proper attention only now
A kind of fraud that was dismissed—'It’s just points, not real money’—has become digital currency now, and its scale is hitting airlines’ bottom lines. As airlines still feel the financial crunch due to the COVID impact, they are exploring all avenues to plug gaps and leaks, and tighten processes to reduce losses, even from digital currencies like airline loyalty points. The loss due to loyalty fraud is massive when fraudulent activities on all the member airlines of an alliance are considered, forcing airlines to seriously look at fraud prevention solutions.
How mileage accrues
Generally, passengers in a frequent flyer program (FFP) can accrue miles for flying on any operating airline (OPE) that is part of an alliance.
The alliance facilitates information exchange between member airlines via intermediate systems (often called ‘hubs’). Today, only the data relevant for mileage accrual is exchanged between alliance members. The accrual process is generally automated—the OPE sends the details of the travelers to the FFP. Members who don’t receive miles via the automated process post flying can log in to the FFP’s website and raise a retroactive mileage claim request, which is sent to the OPE for validation. When these options fail, helpdesks of the airlines concerned help passengers accrue miles.
Loyalty fraud can take various forms
Current airline solutions can’t fight such frauds effectively as airlines in an alliance don’t have complete information about FFP members and their travel details.
The operating airline usually has only the travel information of the passenger (from their source systems like the departure control system (DCS) or the global distribution system (GDS)) and the FFP generally has only the member-related data. Fraudsters usually exploit this gap in information to cheat airlines. For example, fraudsters may enroll in an FFP program with a name common for their geography, say ‘John Smith’, and raise retroactive mileage claim requests on long-haul flights on different OPEs. The OPEs only check if a ‘John Smith’ was present on the flight. If there was a John Smith or a Johnathon S (an actual passenger with the same or similar name), the OPE system may deduce that it is the same FFP member and approve the request. The operating airline has only the passenger’s details and travel details. Hence, validations are done based on the combination of these details (For example: passenger name + flight departure date + origin + destination + flight number). The OPE is not in a position to verify the FFP number.
There are also complexities in matching very long names (the FFP may have some part of the name but the OPE may have the ticketed name stored differently) or very short names (a fraudster using the name ‘Li Li’ may get mapped to ‘Lima Li’ or ‘Anakeli L’). In short, name match validations are tricky for airlines. While a 100% name match is impractical, a very lenient name match algorithm exposes the airline (and, in turn, the alliance) to fraudsters, making detecting and handling them difficult. It is even harder to find internal fraudsters (within the system). Airlines have to spend serious time and effort to determine and block loyalty fraud—a major reason why many airlines do not think it worthwhile to dig deeper into such cases. A mileage fraud prevention solution will, therefore, be based on a data-centric approach. The airline solution should not only be intelligent but also possess machine learning capabilities.
Fraudsters target OPEs that have weak algorithms, bombarding their systems with a huge number of requests (via scripts and BOTs) using very short passenger names (hoping that it will pass the name match validations of the OPE concerned due to a partial match with an actual passenger’s name). But, when a fraudster targets all the OPEs in an alliance, each OPE individually may not be able to see a fraudulent pattern. Also, since the OPEs compensate the FFP for miles awarded, the FFP does not see an incentive in putting in additional checks as that would make it more difficult for its genuine members to raise requests. This leaves the OPEs to fend for themselves.
There are other modes as well to hoodwink the automated accrual process. There have been cases of third-party or airline agents entering their own FFP number for bookings made with them by flyers not part of any FFP—the agent gets miles without even travelling! There could be rogue characters within the airline loyalty program or backend IT departments tweaking the FFP number before the information is transmitted from the OPE to the FFP (or in between). It is difficult to detect such loyalty frauds.
There have also been cases of double-dipping (where a person opens an account with multiple FFPs, raises retroactive mileage claims from more than one account to claim miles multiple times for the same travel), stealing or sharing FFP member account passwords, and account takeovers. But airlines have taken steps to prevent such attempts. In programs that allow pooling of miles into a single account (ideally intended for people from the same family), fraudsters collect small amounts of miles/points and then pool them all together into a single account to make a redemption.
Besides, fraudsters also claim miles by creating new FFP accounts with stolen PNR information, selling redemption tickets and airline loyalty information on the dark web.
The fraud hurts FFPs too
It is common belief that the FFP loses nothing when fraudulent miles are claimed.
However, there is the redemption aspect to be considered. Generally, most FFPs provide more redemption options on their own airlines. As many FFP and OPE entities (even when they are within the same airline) behave as separate entities, the FFP compensates the OPE monetarily for all redemptions made (the FFP deducts miles/points from FFP member’s account for services provided by the OPE, and financially compensates the OPE). The OPE, on the other hand, ends up providing services to a fraudster (instead of a genuine FFP member) and has to also bear the opportunity cost—it loses ticket revenue on the seat booked by fraudsters by redeeming miles.
Given the number of fraudulent miles being claimed and the ease with which they can be redeemed, the FFP stands to lose quite a bit by allowing fraudsters to exist in the FFP program. When the redemptions are made on other partners/airlines (other OPEs), the FFP ends up compensating the partners for the services provided as well.
The OPE pays the FFP for the miles it authorizes to be provided to an FFP member (for the member’s travel). There could also be cases where a genuine passenger requests for miles (for travel) but finds the miles already credited to a fraudster. In such cases, it is generally the FFP that absorbs the cost for mileage accrual as the OPE would have already approved the miles for the flight activity based on the first request.
Alliances can prevent such frauds
Airline alliances have a big role in preventing such practices and should adopt a data-centric approach to develop a fraud prevention solution. To begin with, an alliance should inspect the vast amount of data it has and look for patterns indicating potential loyalty fraud.
It can, for instance, check the mileage accrual requests to figure out if there are FFP members who seem to be travelling around the world and back–all in the same day.
Retro requests raised in the blink of an eye could alert the alliance to potential fraud as BOTS could be behind such requests. The alliance, at this stage, may not be able to zero in on the fraudster. It could, however, make reports available to member airlines, which can help FFPs and OPEs detect loyalty frauds and make their systems stronger to prevent them.
Alerts could be put in place to let FFPs know that something seems to be out of place (for example, over 100 requests from an FFP member in an hour, or many requests with the same ticket number from one or multiple FFP members should raise an alarm). The next step would be to put in place blocks for cases that seem to be suspicious. These could be based on configurable thresholds (for example, all retroactive mileage claim requests can be blocked if more than 10 requests have already been received from an FFP number in a day). The FFPs would then have to review the account activities for all the blocked FFP members and confirm if the members are genuine.
The alliance can also facilitate exchange of key member data (like date of birth, date of enrollment) to the FFP program, to ascertain the genuineness of a claim. The ideal step would be the creation of an automated, machine learning-based model trained on existing data within the alliance, that can be used to predict whether a mileage accrual request is genuine. In a world where fraudsters are innovative, an airline alliance, along with its members, will have to take a data-centric approach while looking for airline solutions that are unique and futuristic to prevent them.