Electronic Consumer Payment Network Used Big Data To Achieve Highly Available System With Low Upfront Costs

Customer Profile

The Company is a technology driven and customer focused organization. It is building India’s leading electronic consumer payments network. This network of participating merchants allows service providers and businesses to collect payments from Indian consumers. The cash acceptance network allows service providers to reach the length and breadth of the nation through their 30,000 + Recharge Points.

With a vision “to be a catalyst for the conversion of the cash economy into electronic transactions”, this company is currently implementing and operating payment networks for multiple services like mobile prepaid top-up, DTH top-up and Cash acceptance for various service providers/utilities. The merchants provide these services either through their mobile phones or through the web (in case they have a PC with internet).

The Challenge

The company had conceptualized a business model towards building an analytics system that could recommend offers and discounts in real time through the mobile app and SMS notifications to the users. This allowed the retailers and manufacturers to analyze the product performance and take action towards promoting a certain product(s) in real-time.

This underlined the need to analyze the existing data that was collected from the Point of Sale (POS) systems to determine the usage/engagement pattern, and therefore provide required insights to the manufacturers.

Since they were expecting a huge amount of data for processing (around 200 Million rows or 100 GB data per day), they also wanted to achieve a highly sophisticated and secure data warehouse so that they could retain the data up to 3 years.

The Solution

BlazeClan conducted an exhaustive study to understand the workflow of the process wherein:

• The retailers would enrol the consumers with the company’s system.
• Sales would be done using POS (Point of Sales) by the Retailers.
• Regardless of the fact that the consumer is registered or not, the POS would send data to the company’s system.
• The data gathered would be mined for insights and then sent to the manufacturers.
• For registered users, each transaction based on the demographic information and past history in real time bases would be followed by an offer for the user.
• For smartphone users, it would be in the form of push notification while for other users, the mode of communication would be SMS.
• The consumer can use the discount coupons and the company would pay the relator and the financial data would be maintained in an Oracle system.

Once the workflow was understood, BlazeClan adhered to the Big Data Solution approach that is designed to help the company define a:

• Big Data Analytics Process,
• Adopt Right Tools,
• Build a Cloud-Based Analytics engine and
• Empower businesses with actionable and operational business analytics.

This 5 step solution approach included:

1. Data Discovery: Data structure, data sources, data volume were identified.
2. Analytics Discovery: The architecture was designed by understanding data correlation, frequency, and algorithm.
3. Tool and Technology Discovery: The tools and technologies required for the implementation.
4. Infrastructure Discovery: Helped in identifying the right resources for the architecture in a cost-optimized way.
5. Implementation Phase: A continuous delivery approach was taken during the implementation Phase, with a lean structure of continuous learning based iterations. This approach assured low upfront investments and adoptive actionable analytics.

BlazeClan solution ensured data security, scalability, and high availability. With the use of serverless technology like Lambda, the company could easily take action on any new transactions within a few milliseconds.

All the real-time data was also being pushed into the data warehouse for analysts and business stakeholders to analyze the performance of any campaign or product line-up.

The Benefits

1. Customized Offers: The Company was able to provide customized offers and discounts real time to their customers.
2. Real-Time Insights: The analysts and business stakeholders could easily analyze the performance of any campaign for the product(s) line-up.
3. High Availability: The Company achieved a highly available system that could meet and cater to an increasing number of customer transactions within milliseconds.
4. Scalability: They achieved the ability to scale the application as and when required by leveraging auto-scaling and load balancing features of AWS.

Tech Stack

• EC2
• Redshift
• S3
• SNS
• Dynamo DB
• Dynamo stream
• Lambda
• Kinesis Firehose