The situation I lived a few months ago was “dramatic.” Oasis announces that it returns to the stage. It is one of my favorite groups and I had very clear that I wanted to listen to them live again, so the day the tickets came up for sale was prepared in front of my PC, in the virtual tail, with hundreds of thousands of fans like me waiting for That my turn will arrive to buy the tickets. And there was luck! I managed to confirm the purchase of two tickets … until, at the time of paying … my bank rejected the transaction for no reason.
The payment platform gave me a second chance and tried Good luck with the purchase.
The reality is that the issuer of the first card recognized the transaction as a suspicious fraud, which caused unnecessary friction, and made me reconsider using it again. It is very likely that the two entities will use fraud detection models based on AI but what is clear is that only the second takes advantage of hyperpersonalization –Mayor precision in the detection of fraud or reduction to the minimum of false positives, among others – To make automated decisions.
The AI models for fraud detection can determine if a transaction is extraordinarily unusual to automatically reject it, as could be in my case the purchase of a couple of tickets for almost 1,000 euros, but only in the case of the second entity They combined detection techniques with hyperpersonalization to know my desire to buy.
Hyperpersonalization based on AI
To enter more in the detail of how hyperpersonalization can improve the user experience while increasing the detection of real fraud, it is essential to understand how AI systems have evolved from the mere fact of making decisions based on population statistics to Take them according to the user’s behaviors, taking advantage of the transactions profiles and capturing the purchase history of each client individually to understand and anticipate future transactions.
Acting in this way, with hyperpersonalized profiles, a too generic prediction can be avoided that will end up causing friction in the client. Over time, the models are learning to know which transactions can really be fraudulent and which are legitimate, concentrating the analysis only in the former.
Hyperpersonalization based on AI to reduce fraud in financial entities
In my case, even if it was my first time buying tickets of 1,000 euros, the issuer of my second card Vio that I had already bought tickets for concerts previously and that my expenses with the card are reduced to paying trips and leisure, so that the detection model decided that it was a legitimate purchase.
In any case, hyperpersonalization does not only know the past transactions of a client. It especially serves to predict how users with similar expenses behave. Collaborative profiles mark the purchase records of hundreds of thousands of customers; This information is grouped together optimally to get information from archetypes and models associate probabilities for behaviors that are repeated between users with similar behaviors.
These archetypes can provide very powerful information, for example how much a user can spend on a trip, what cash limitations a family can suffer when the end of the month approaches or how much a young man will spend on video games next month. And when a user has a new behavior, such as buying tickets for 1,000 euros, the fraud detection model takes advantage of everything he knows about past purchases and analyzes similar archetype behaviors to determine if similar profiles have made purchases of similar value With the time of time to make the payment in a virtual tail.
The models that combine hyperpersonalized information with the recurring patterns of similar profiles perform, as we see, an individualized analysis of great value for the neural networks of fraud detection systems, mechanisms that qualify transactions according to the probability of being fraudulent and They provide objective reasons behind this qualification.
As we see, then, it is vital for any financial entity to have a more intelligent artificial intelligence focused on the individual to make better optimal decisions and reduce friction.
