Fraud is going off the scale. PCM outlines how well current solutions are working, and what might work better.
Even before the holidays have begun, the signs are not good: a recent report from online processor Vesta reports fraud attempts are currently running between 11 and 13% of all card-not-present (CNP) transactions.
CNP transactions mostly means e- and m-commerce these days: digital business now accounts for one dollar in six spent in the US, and is growing at 13% year on year worldwide. With each Christmas comes an increase in e-commerce spending of around 30% – and, sadly, further increases in attempted fraud.
Looking beyond Christmas 2021, fraud could become such a problem it puts people off buying goods and services over the internet.
If that sounds laughable, don’t assume it can’t happen: though e-commerce is popular in Latin America, processors in markets like Colombia and Chile have resorted to draconian measures (such as refusing half of all internet transactions) to stop fraud losses.
And that’s having a knock-on effect for the region’s growth in digital business more generally. As we’ll see, one company may have a solution to this dilemma.
Hard to find, impossible to stop?
With each successful fraud costing between $108 and $155 for the bank or merchant in question, e-commerce fraud gets very expensive, very quickly. As Vesta point out, a mid-market firm doing five million transactions each year could experience 650,000 fraud events at a cost – including chargeback fees of $25 per event – of $16,250,000 every year.
What’s worse, it feels like current anti-fraud measures don’t work. Vesta explains that almost two-thirds (57%) of fraud events are hard to connect to a particular fraud method (such as account takeover) which makes them tough to stop.
And the measures taken by some firms aren’t helping: given rising fraud, some companies are turning down as many as one in three transactions, two-thirds of which might be genuine. Vesta estimate such practices could cost these firms as much as $102 million in lost revenue opportunities from legitimate transactions.
What might work better
A separate survey from payments specialists Outseer says the situation is even worse than Vesta claim, with CNP fraud in 34 major markets rocketing by 29% to reach $17.2 billion in 2023. Outseer surveyed 750 merchants and found 69% currently use 3D Secure (3DS), with most planning to increase deployments of 3DS in the next two years.
3DS uses more than 100 transaction data points, risk scoring and elevated authentication to protect card transactions online. It seems to work reasonably well, reducing fraud by two-thirds – but there’s a catch. Where 3DS is not deployed consistently across a market, fraud migrates to those issuers and merchants that don’t use the system, driving up losses.
In the US, where there’s no 3DS requirement, fraud runs at around 17 basis points (bps). By contrast, the EU’s Strong Customer Authentication (SCA) mandate has led to most issuers and merchants adopting 3DS as a baseline anti-fraud measure and reductions in the fraud rate from 12 bps to 4 bps.
That said, even 4 bps is a whopping figure, given e-commerce is worth €757 billion across the European trade bloc according to Eurostat. That means the EU is losing around €30 billion to fraud annually – a higher figure than the US, albeit with 25% higher population.
Grabbing that fraud Grinch
There seems to be no end of AI companies touting new anti-fraud techniques. Examples include Independent Device Verification, which works by confirming the geolocation and registration number of a mobile device independent of registered user verification.
Then there are schemes that add cookies as a device layer on top of merchant security, enabling users to verify their identity one time across multiple merchants, and finally AI and machine learning itself, pushing claims of ever-stronger algorithmical power.
One intriguing possibility is the growing use of neural networks in AI. Neural networks effectively supercharge AI, able to access up to 20 million different information sources to complement transaction data, and to parse huge amounts of data in near-real time to uncover anomalies and more accurately identify potential fraud.
AI experts Aerospike recently published a white paper claiming their AI, powered by neural networks, is able to parse billions of dollars of transactions rapidly and save millions in fraud losses by the day.
With results like this, it could be that the new wave of AI will help fight the fraud grinch, saving not just Christmas, but huge amounts of revenue, blocked transactions and customer friction caused by escalated authentication.