As the series of payCLT events in Charlotte continue to roll out, another one was held on April 15 with Todd Clark, General Manager of First Data’s STAR Network, as the guest speaker at this event. Todd Clark leads First Data’s STAR Network, the world’s largest independent debit network, which was acquired by First Data in 2003.
Mr. Clark has shared valuable insights on the debit cards industry and fraud in emerging payments.
Contactless payments are not picking up: EMV requires development
One of the insights shared by Mr. Clark was regarding the prioritization of payments methods: EMV and contactless payments at POS. According to Mr. Clark, the merchants following customer preferences have encountered a low adoption of contactless payments among the customer base. Even though EMV requires a lot of development, it is the main focus of processors and merchants.
Another interesting data point was related to Apple Pay, which is powered by First Data. As Mr. Clark shared, Apple Pay's adoption hasn’t been picking up much, which corresponds with the overall low contactless payments adoption.
Tuning the Durbin Amendment for small merchants
One of the topics of discussion with Mr. Clark was the Durbin Amendment. Passed as a part of the Dodd-Frank financial reform legislation in 2010, the amendment requires the Federal Reserve to limit fees charged to retailers for debit card processing. In a nutshell, the Durbin Amendment has granted merchants a choice of networks to route the transaction over. The merchant was powered to use the cheapest option available regardless of the processor he has a contract with.
However, it is not all that smooth with the Durbin Amendment. While large merchants have the capacity to use sophisticated systems to route transactions on an individual basis to the cheapest option, small merchants don’t get to enjoy the luxury. However, Mr. Clark has shared that First Data is working on the solution for second-tier merchants that would allow them to route transactions on the individual basis through the cheapest network for a certain fee that is bearable for small merchants.
Machine learning in fraud scoring
Aside from the insights on payments methods and notes on the Durbin Amendment, one of the main focuses of the presentation was around fraud scoring in payments and application of machine learning. As Mr. Clark shared, the company is currently using machine learning for internal purposes and testing in order to find ways to apply it to data analysis and risk assessment. Those areas are at importance now and involve investments and development.
According to the presentation, debit transactions represent 65% of all card transactions. It is as much a powerful vehicle in the payments industry as it is a soft spot for fraudsters to take advantage of. Fraud in emerging payments is an important issue that was addressed at the event by the guest speaker.
Fraud in payments is largely dependent on the processor's ability to gather and analyze data in a sophisticated and timely manner. As shared by Mr. Clark, there are four types of data sources determined by a range of use cases.
Following the modern range of data sources, the presenter has outlined three main fraud score differentiators: speed of learning (when machine learning is applied to fraud detection), breadth of data and feedback loop.
There are four important hallmarks defining effective speed of learning:
- Models updated immediately with new intelligence – self-discovery patterns
- Continuous machine learning based on behavioral patterns
- Ability to implement new data sources to enhance the model
- Real-time data used to actively retrain models after each transaction.
Speaking of common vulnerabilities of global networks, the guest speaker has listed five important areas requiring improvement:
- Ability to handle different input types
- Ability to handle missing values
- Speed of new model learning
- Tuning necessity
The next payCLT event in Charlotte will be devoted to QCFintech Power Pitches and will be held on May 6, 2016, at Packard Place, Charlotte, NC.