Harnessing the Power of Real-time Data Platforms to Propel Digital Payments

  • Stuart Tarmy, Global Director at Aerospike

  • 06.02.2023 03:30 pm
  • #digitalpayments

The global digital payments market is expected to grow to $19.89 trillion by 2026, a massive uptick from 2018 when revenue stood at $3.53 trillion. This expansion is also being witnessed in the instant payments area, which, while smaller, is growing at a rate of around 18%. Growth is being driven by customers worldwide who expect secure, fast, and reliable digital experiences with every payment interaction. 

It's not just the markets that are developing. Online payment fraud is escalating at a phenomenal rate, with eCommerce losses last year estimated at $41 billion globally and expectations that this could increase to $48 billion in 2023. Fraudsters are getting more sophisticated, for example, in mining customer information on the internet that in earlier years would have been unavailable.   Halting the rise in fraud, however, is a huge battle.

Preventing fraud

Part of the challenge is dealing with the amount of data surrounding every transaction. The digital nature of payments these days makes it essential that payment service providers must immediately verify the online identities of customers and stop fraudulent transactions using stolen credit cards or identities before customers and merchants are affected. Simultaneously they want to minimise false positives to provide a pleasing customer experience and keep customer inconvenience to a minimum. The more sophisticated fraud solutions use AI algorithms, while best-in-class fraud solutions are using more advanced AI algorithms such as neural nets and deep learning.  Companies using these more advanced fraud algorithms need to run much more complex algorithms together with much more data, all while maintaining the same SLA as before, e.g., approving/denying a payment within 20ms.  One technique to help meet this tight fraud SLA is using machine learning at the edge. The closer to the customer that a transaction can be processed, the faster, more friction-free, and more secure it is likely to be. This requires real-time database management platforms that ingest and act on streaming data where payments occur, allowing organisations to act in real-time across billions of database transactions. 

Modern NoSQL databases ensure that every transaction can be augmented with real-time data from systems of record, third-party sources, data lakes, hot/cold storage, and other sources to assist operational, transactional, and analytical workloads. In addition, rather than demanding that companies add to their technology assets to manage the additional data, they can instead assist in reducing server footprint. 

PayPal is a great example. The world’s largest online payment system uses a real-time database management platform to process and analyse data to identify emerging fraud patterns in under 20 milliseconds, even when volumes are enormously high (over 150 TBs). It can scale with increased volume and still meet its service level agreements over 99 percent of the time.  Remarkably, their advanced fraud algorithms reduced their fraud levels to 0.3% while reducing false positives by 30 times. 

A different challenge existed at Barclays Bank.  Barclays had numerous siloed fraud systems in different areas of the bank.  There was no single, unified customer profile store to provide a complete snapshot of the customer.  Consequently, the disparate fraud solution contained incomplete customer information and was becoming unwieldy regarding maintenance and development. To address this, the bank implemented a real-time data platform to unify customer profile data.  In addition, the bank was planning for significant growth and implemented a fraud detection system that allowed it to increase its dataset from 3TB to over 30TB in just three years, share fraud rules across multiple platforms and facilitate machine learning consistently to achieve a sub-100 millisecond response time in virtually every transaction.  

Delivering hyper-personalization

It is not only fraud prevention that is driving the move toward improved data management. Personalisation, one of the key pillars on which the success of the eCommerce industry is built, now extends into multiple other areas, not least digital payments. 

As a result, customers are selecting peer-to-peer payments such as Venmo and PayPal, which they regard as secure methods of sending money, with the additional benefit of tracking their expenditures. 

Personalisation is the reason that credit card providers offer users loyalty points on a wide variety of different purchases, from holiday bookings to food. Credit cards also offer a degree of consumer protection in the form of reverse payments and error resolutions, which cannot be found with instant payments. 

One company, Airtel, launched an initiative that provides a holistic set of historical data and real-time triggers from their siloed data across different business units. After installing a real-time database management platform, Airtel was able to take data from 350 million customers and bring it into one database. They achieved it at a rate of more than 40,000 transactions per second with sub-millisecond performance, which included both read-and-writes to the database.

Making payments instant

Many consumers expect payments that are effective immediately, and increasingly banks and payment companies are offering this option to customers. European banks can now deliver fast, reliable, and cost-effective payments services from any bank in the euro-area over their own network while ensuring full privacy and security, thanks to a real-time platform from Aerospike. This allows it to compete with established instant payments services and lay the groundwork to offer other services, such as person-to-person mobile payments. 

The payments sector is increasingly moving forward with instant payments which are enabled by real-time data management platforms. Organisations are leveraging hyper-scale speed and performance, reducing fraud, false positives and money laundering,  and enabling modern AI-based value-added solutions such as personalised customer offers that are giving them a critical edge in what has become a highly competitive market.

Other Blogs