Processing...
Δ
Retail & eCommerce
Data Engineering
Scala, Apache Spark, Snowflake
Our client is a grocery store chain that has locations across several countries. As the business grew and the number of outlets increased, it became necessary to automate the analytics system for margin indicators, expenses, and stock balances, since manual processing has become bulky and inaccurate. Furthermore, the client aimed to leverage remarketing campaigns, offering customers with discount cards relevant products based on an analysis of their purchase history.
The client was looking for an experienced and proven software engineering vendor experienced in implementing business automation solutions and boosting the entire digital transformation of the nationwide retail chain. At the recommendation of their partners, the client turned to NIX as a company who has extensive experience migrating business infrastructure into the cloud and implementing data management, analytics, and business intelligence solutions.
We were tasked with setting up a data pipeline that would take statistical data from point-of-sale (POS) systems and transform it into dashboards and reports. This includes:
Setup process of transforming statistical data from POS systems to the data stream for real-time data analysis
Building a data pipeline for automatic generation of complex reports covering many crucial KPIs
Generating detailed reports for different stakeholders and storing them for easy access and reuse
Our cloud specialists analyzed different cloud providers during the project’s discovery phase. As a result, they determined that Amazon AWS would be the most cost-efficient platform for the client’s configuration. Due to a large pool of technical expertise, we quickly formed a team that covered every aspect of solutions.
Our data engineers build a pipeline in which data from POS systems are refined and processed until transformed into reports and dashboards.
It all starts with a seller closing the sale. Then, a generated payment transaction containing the product name, price, count, customer ID, shop ID, and purchase time enters the pipeline through Kinesis Data Stream (KDS) service. Finally, received data is processed in the Elastic MapReduce cluster (EMR) using Spark. Processing here means grouping payment transactions by store, date, product type, and other parameters.
Then, the grouped transactions are stored on another S3 using Spark Job. If sales of any product exceed a certain threshold within 5 minutes in one store, Simple Notification Service (SNS) generates and sends the appropriate email to the purchasing department.
As soon as the files get to S3, their contents are copied to the Simple Queue Service (SQS) queue for Snowpipe mapping. The data is then stored in the DWH table. Our engineers have also synchronized data about regular customers, network stock status, and product data with the DWH tables. Thus, our solution provides capabilities for further data analysis, identifying patterns and regularities that help to grow the business.
After the end of each working shift, the system runs an EMR cluster job, which generates a report for the day, and if it’s the last day of the month, then the monthly report as well. This report is mailed to SNS topic subscribers—in our case, the head of marketing, the administrator of each store, and the COO. In addition, the data is saved on S3 and can be accessed through the API gateway to connect the graphical interface and convenient presentation of information.
This report includes tons of analytical metrics, such as:
There are only seven items, but in fact, reports consist of much more insights about the performance and profitability of each store and its customers.
The client received valuable insights about its business, giving him plenty of room for improvement, and information about customer’s preferences in each region, that can be monetized as well through the developed cloud analytic solution. In addition, the client’s marketing team obtains an accurate portrait of their audience, which allows personalization of offers and increasing customer satisfaction rate, income, and results.
This solution was a significant step in digital transformation for the client’s business. Based on this foundation, it’s possible to implement more advanced analytics using AI and Business intelligence solutions to track users’ interactions more precisely and get more opportunities for further business growth.
7 experts (Project Manager, Business Analyst, 4 Data Engineers, QA Engineer)
Snowflake, Scala, Apache Spark, Snowflake Data Platform
Schedule Meeting
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.