Modern travel is exceptional - aircraft with wifi, bullet trains that average 200+ mph, boutique hotels that cater to every whim of weary travelers.
NB: This is a guest article by Jason Tan, CEO of Sift Science.
But perhaps even more incredible is that, with the click of a mouse or the swipe of a finger, users can book flights, reserve hotel rooms and make travel arrangements effortlessly and instantly.
The convenience and near-anonymity of travel sites and apps eliminate the face-to-face elements of holding on the phone, standing in lines, and coordinating with physical travel agents.
A credit card can give customers access to world of wonderful experiences.
But this convenience and access comes with a price. The relatively low cost, lack of physical shipping address and immediate acceptance or denial of payment on travel sites and apps make these businesses a target for credit card and account fraud.
When combined with the rise in popularity of on-demand businesses, companies such as Airbnb, TravelMob, Reserbus, and HotelTonight are finding that they must learn to predict fraud before it strikes - combatting fraud after the fact is too late.
The challenge
Flight or train reservations, car rentals, last-minute hotel bookings: what these travel services share is a need for competitive prices and instant confirmations.
The real-time need to book and confirm reservations means these businesses don’t have the luxury of manual review time.
The vacation rental marketplaces face an additional fraud threat. Even more damaging than losing a sale to a chargeback is losing a good customer.
When travel marketplaces are plagued with fake listings, good users have bad experiences.
Every time a good customer pays for a vacation rental that doesn’t match the photos or reviews or – even worse – doesn’t exist at all, that customer loses trust in the travel site.
Bad experiences create their own negative momentum as, as the value of lost future bookings and requests for refunds add up.
The situation
When a reservation is made diligent travel sites get to work immediately to acquire that hotel room or whatever travel service requested.
If Freddy Fraudster takes a stolen credit card number and uses it to book a first-class flight tonight and a hotel for tomorrow, by the time the travel businesses learn through a chargeback that fraud occurred, the travel will already be over.
The site has already paid for the room or service. Not only do they lose the profits of a sale, but they’re also out to the tune of the cost of the booking. There might also be a chargeback fee or penalty from the issuing bank may levy.
There is a contradiction in the e-commerce world which applies to travel - making things simpler for customers also makes it easier for fraudsters.
Take a vacation rental site where only minimal information is collected from the user. Don’t want to log in? Okay! Prefer to not share your phone number? Fine by us.
But adding these steps into the booking process is one way of separating the customers from the fraudsters - travel is a popular vertical with criminals looking to test stolen credit card numbers or see what they can do with stolen payment information.
How to stay ahead of fraudsters
Staying ahead of fraudsters is the only way for travel businesses to protect themselves and their good customers.
No matter the specialty –flights, ride-sharing, hotel bookings, or user marketplace – today’s travel sites need the insight of large-scale machine learning.
The global machine learning model is a solution that allows fraud analysts to benefit from fraud instances and patterns observed and reported worldwide.
The real-time element ensures that travel sites, which rely on instantaneous reservations to secure bookings, have the most up-to-date data on known fraudulent attributes and connections to bad users.
Visualizing these connections allows for proactive fraud fighting, rather than waiting for the chargeback or complaint to arrive.
Easy API integration allows for instantaneous feedback into the system, ensuring that the most up-to-date information is always at your fingertips.
The moment that a user in the network is flagged as fraudulent, all analysts are in the know.
Beyond this, fraud systems need to be customizable. All travel sites have highly specialized insights into the behaviors of the good and bad users on their sites. These insights can be used to tailor fraud detection needs.
Large-scale machine learning can help travel firms reduce fraud by stopping it from happening in the first place.
NB: This is a guest article by Jason Tan, CEO of Sift Science.
NB2: Fraud image by Shutterstock