Real-Time detection and alerting of unwanted credit card charges (Part 1 of 3)

This is the first part of a 3-part blog series (Part 1, Part 2, Part 3) about a real-time streaming analytics use case in financial services. This real-world use case is not only the foundation for other use cases in financial services but could also be inspiring for other use cases in other verticals such as advertising, security & intelligence, manufacturing, healthcare …

This first part is about the business problem and potential benefits of the solution to service providers and end users. The second part is about the solution architecture, implementation options and tools selection. The third part is about building the solution with the end-to-end Striim?platform: a single platform that combines real-time ingestion, stream processing and real-time delivery of streaming data while overcoming the complexity and higher cost of stitching together and maintaining disparate tools.

1. What is the business problem?

There are fraudulent credit card transactions and there are valid transactions. There are numerous systems and services that deal with both of these types of transactions. But, what about those deemed “grey charges” between fraudulent and valid ones for which the credit card holder perspective is required to confirm potentially unwanted charges or mistakes?

Such “grey charges” can be duplicate charges for a single purchase, increase in a recurring monthly utility bill, an unusually large tip left at a restaurant, … These charges are either overlooked by the credit card holders or disputed later resulting in either money loss or time loss, hassle and frustration.

In the lack of an off-the-shelf related product or service for the credit card issuing banks, how to design and build such a real-time system to deal with these “grey charges”?

2. What would be the solution?

The solution would be a streaming data application that:

• Monitors credit card transactions in real-time and, based on some rules, flags some of these transactions as potential mistakes or unwanted charges
• Joins in real-time with contact preferences and contact info of credit card holders
• Sends credit card holders alerts via text message, email or push notification and ask them about their perspective on those “grey charges” while giving them the opportunity to immediately correct those charges with the merchant as well as instructions should they decide to challenge the charges later
• Persists content related to these “grey charges” to be used in a section in the upcoming credit card holders’ monthly statements

3. What would be the benefits of such solution?

The benefits to the issuing banks would be:

• A cost saving initiative that helps the issuing banks cut down their liability for unauthorized charges, reduce calls to call centers and reduce time for resolving these mistakes and decrease customer churn
• A competitive edge to issuing banks, in a very crowded credit card market, by offering this solution as an additional attractive feature to their credit card holders
• An enhanced customer service by facilitating direct communication and personalized interactions between the banks and the credit card holders based on their transactions activities

The benefits to the credit card holders would be:

• Awareness, through real-time alerts and monthly statements, of potential mistakes or potentially unwanted charges that they might have overlooked
• Recovery of money related to mistakes or unwanted charges they would have lost
• Additional incentive and increased confidence to do business with the issuing banks
• Time saving through the possibility of resolving those “grey charges”, as soon as they happen with the merchant first, making it unnecessary to contact the issuing banks

4. Why this solution would be the foundation for other use cases?

Although the business use cases will vary, the data flow of a streaming data application remains mostly the same:

• Ingestion of event streams from a variety of data sources
• Processing of event streams such as continuous queries, windowing, filtering, transformations and aggregation, pattern matching, …
• Joining of event streams with other streams or with external context for enrichment and correlation
• Delivery and persistence of event streams
• Visualization of event streams on a dashboard

5. How this solution would be the foundation for other use cases?

This solution can be the foundation to build more sophisticated use cases in financial services such as:

• Real-Time fraud prevention: detect fraudulent transaction on the fly rather than after the transaction is approved
• Real-Time CLIP decision: Credit Limit Increase Processing on the fly when a transaction goes above the credit card limit
• Real-Time targeted offers: special offers pushed to users in real-time based on their actions and locations
• Real-Time customer assistant: detect what customers are trying to do and assist them in real-time

This solution can be an inspiration to build other use cases in other verticals such as: 

• Advertising: Real-Time Bidding (RTB) for digital advertising
• Manufacturing: Real-Time system monitoring and detection of early warnings
• Healthcare: Health monitoring from medical devices
• Internet of Things (IoT): Real-Time analysis of streaming sensor data and actions