How AI and machine learning help quick-service restaurants battle fraud
Quick-service restaurants (QSRs) have transformed their operations with contactless payments and mobile ordering. These days, around 60% of digital orders come from mobile apps, according to a 2018 NPD Group report. By the end of 2020, the NPD Group expects digital orders to triple.
But more mobile orders mean an increase in digital fraud losses, as bad actors respond with new digital attacks. These bad actors target new QSR mobile apps with account takeover (ATO) attacks. For example, bad actors may buy blocks of stolen credentials on the dark web. Then they run them through a QSR’s new app or one of the leading food delivery apps. In general, ATO victims spent, on average, $290 out-of-pocket and 15 hours resolving fraud, says a 2018 Javelin report.
Mobile devices add complexity and scale to attacks, and rules-based fraud detection can’t keep up. But artificial intelligence (AI) and machine learning can, and they’re powering new solutions that can stop ATO to thwart fraud aimed at QSRs.
1. They segment fraudulent and trustworthy transactions
AI uses both supervised and unsupervised algorithms to assess the risk of each transaction. These algorithms learn continuously to detect fraud more accurately. This unique ability to learn automatically is the future of fraud prevention.
The most effective fraud prevention systems train their algorithms on robust historical transaction data. And they can detect emerging fraud from new data. For example, Kount’s Omniscore is an AI-based risk score that produces a single transaction safety rating based on supervised and unsupervised machine learning. At-a-glance scores can help QSRs improve fraud detection to speed up checkout and improve customer relationships.
2. They prevent account takeover fraud
Protecting loyal customers is just as important as protecting new customers from fraud. Customers who use the same QSR apps often may have accumulated a high number of loyalty points. Preventing account takeover fraud means QSRs can stop bad actors from accessing and stealing those loyalty points and customer data. Account takeover can have devastating, long-term effects on businesses. Beyond lost revenue, account takeover fraud causes brand damage. It can permanently erode the trust of good customers.
3. They assess risk in real time
Fraud detection systems that use machine learning can save QSRs valuable time. They categorize transactions and assess risk in milliseconds. How does it work? Machine learning compares a transaction’s digital footprint with many others. It looks for behaviors and new patterns associated with previous fraud or trusted activity. And it all happens in real time to accelerate good orders.
Real-time decisions can help detect fraud schemes. For example, bad actors may use the same credentials across multiple user accounts. But machine learning algorithms will flag any transaction with those identifiers by comparing millions of trust and fraud signals. By detecting these attacks, QSRs can save thousands of dollars in losses.
4. They reduce false positives that block good customers
Fraud detection sometimes flags good transactions. Errors annoy customers and create negative experiences that erode loyalty. And for QSRs, speed is the essential benefit of mobile app shopping. Any friction in the mobile experience can drive customers to competitors.
Machine learning reduces these “false positives.” It helps QSRs to deliver the all-important frictionless experience that builds customer loyalty. And it removes barriers to fast order and delivery.
5. They reduce manual reviews and preserve the customer experience
Static, rules-based fraud detection systems waste fraud analysts’ time and can damage the customer experience. Customers expect to pick up their orders in minutes. Lengthy waits due to manual review drive them away. Systems that apply arbitrary transaction limits increase manual reviews. By contrast, AI-based systems are more accurate and flag only likely fraud. Fewer reviews mean less labor and lower operational costs.
AI and machine learning upgrade QSR fraud detection
QSRs often operate with low margins. Machine learning can thwart fraudulent transactions that sink profits. In the QSR space, it’s hard to exceed customer expectations. Machine learning-based fraud prevention can help reduce false positives, improve customer experiences, and protect those thin margins.
In a mobile world, rules-based fraud prevention won’t work for QSRs. Chains that use rigid criteria to identify potential fraud generate many false positives. And false positives can drive even loyal customers away. Today, digital fraud prevention systems like Kount use machine learning to reduce false positives by up to 70%.
Kount’s award-winning AI and machine learning identify and stop sophisticated fraud attacks. Its unique system analyzes user behavior and numerous types of customer data. Kount’s AI-driven platform is powered by the Identity Trust Global Network, which includes billions of fraud and trust-related identity signals to deliver accurate decisions in milliseconds.
Kount’s AI combines both supervised and unsupervised machine learning to detect and prevent both existing and emerging fraud. That level of data enables confident decisions, which is critical for QSRs that don’t have the time for manual reviews and may encounter unusual transaction patterns.