How Anomaly Detection helps in Improving Payment Success Rate.

In an ideal world, every transaction is smooth, precise and fast.

Unfortunately, we don’t live in an ideal world. So the next time your customer places an order for their favorite brand of delectably mouth-watering waffles there is every possibility that a delayed authentication may occur or, worst-case scenario, a payment gateway failure.

Even a small misstep within your complex payment infrastructure can have your competition stepping in to snatch away your sale and possibly all future sales from that end.

Let’s simplify what happens behind the scene!

In the two to three seconds it takes to place an order,there are mammoth tasks being handled.

  • Encrypted transmission of transaction details where user card elements are encrypted for protection before being transmitted to the payment gateway
  • Authentication adds a second layer of protection guaranteeing a secure payment process for the user
  • OTP generation is a step that ensures payment has been requested by a legitimate user
  • Validation protocols establish the legitimacy of the entire process
  • Approval/Denial by issuing bank means the bank has received all relevant details and can transmit back the apt response
  • The acquiring bank receives the encrypted details of the transmission.

Of course, the end user has eyes only for the ‘Transaction is Successful’ declaration.

The sheer volume of data that is accumulated during routine transactions like these make it impossible for a traditional system to seek out anomalies. This is where advanced machine learning capabilities coupled with Fintech domain expertise come into the picture and what CrunchMetrics offers by way of AI powered payment outlier detection techniques.

Bringing your A-game to the table

Using machine learning-based and deep learning-based proprietary algorithms, the system investigates historical data to detect patterns, spot anomalies and test assumptions. Based on the nature of data and the business, the prediction window and prediction frequency are determined. CrunchMetrics has four key modules embedded within the product for end-to-end anomaly detection, investigation and providing the correct solution.

  • Anomaly Detection Module: Detects anomalies in business KPIs and metrics in real-time
  • Anomaly Investigation Module: Uses the robust Root Cause Analysis (RCA) framework to identify the cause of a detected anomaly or series of anomalies
  • Automated Insights Module: Uses Natural Language Generation (NLG) based insights module for end-to-end view of anomalies, patterns, underlying relational attributes and the potential root cause
  • Alert Module: Amalgamates rule-based and machine learning-based alerting, incorporates feedback and auto-actions triggers from user to module

In the high volume payment processing world, CrunchMetrics adeptly handles seasonal and cyclical variations. For example, it could refer to transactions during e-commerce annual sales and payment gateway experience during seasonal or other types of clearance sales.

  • CrunchMetrics evaluates the success rate of payments for combination of payment types, processors, terminals and gateways. It iterates over the true target with cyclical and random variations of data to ensure it can handle such variations in the future
  • A robust Root Cause Analysis (RCA) framework is established to offer a multivariate approach that points out anomalies, not just in one variable, but the effect they will have on other variables in the dataset
  • Auto-actioning, like switching from an inactive or bad performing (anomalous) terminal to another, can be enabled through the alert module

The system self-learns from regular patterns of data, detects malfunctions instantly and performs corrective measures for the transaction to move forward smoothly.

To say simply, by instituting real-time anomaly detection for Fintech, your system will have the capabilities to learn, explore, detect and clean-up anomalies without any predictable business loss in the future.

Summon the power of Augmented Analytics to help you identify risks and business incidents in real-time.

Sign up for a pilot run on your data


Shashank Shekhar is Data Sciences leader with diverse experience across verticals including CPG, Retail, Hitech and E-commerce domains. He joined Subex from VMware where he was heading Data Sciences practice for transformational projects. In the past, he has worked in Amazon, Flipkart and Target and has been involved in solving various complex business problems using Machine Learning and Data Sciences.

Leave a Reply

Your email address will not be published. Required fields are marked *