Once a purchase had been authorised, there is no going back. That means that all fraud detection measures must be done during in the first step of a transaction.
Here’s how it works (in a dramatically simplified fashion).
Once companies such as Visa or Mastercard have licensed their brands to a card issuer – a lender like, say, Barclays Bank – and to the merchant’s bank, they fix the terms of the transaction agreement.
Then, the card issuer physically delivers the credit card to the consumer. To make a purchase with it, the cardholder gives his card to the vendor (or, online, manually enters the card information), who forwards data on the consumer and the desired purchase to the merchant’s bank.
The bank, in turn, routes the required information to the card issuer for analysis and approval – or rejection. The card issuer’s final decision is sent back to both the merchant’s bank and the vendor.
Rejection may be issued only in two situations: if the balance on the cardholder’s account is insufficient or if, based on the data provided by the merchant’s bank, there is suspicion of fraud.
Incorrect suspicions of fraud is inconvenient for the consumer, whose purchase has been denied and whose card may summarily be blocked by the card issuer, and poses a reputational damage to the vendor.
How to counter frauds?
Based on my research, which examines how advanced statistical and probabilistic techniques could better detect fraud, sequential analysis – coupled with new technology – holds the key.
Thanks to the continuous monitoring of cardholder expenditure and information – including the time, amount and geographical coordinates of each purchase – it should be possible to develop a computer model that would calculate the probability that a purchase is fraudulent. If the probability passes a certain threshold, the card issuer would be issued an alarm.