An employee engagement and rewards company consults with clients to identify ways to drive customer and employee loyalty through planning, organizing, and executing incentive programs, travel, and events.
This company looked to expand service offerings to include performing data analysis on customers’ data. Analyzing data would provide these customers with actionable insights that they could not previously obtain, speeding up the customer’s return on investment and strengthening this company’s customer relationships.
Large Volume of Point of Sale Data With Limited Analytic Capabilities
One particular customer provided this company with a large amount of point-of-sale data. The rewards company proposed performing a market basket analysis on the data to identify purchasing trends, allowing the customer to better understand consumers and influence purchasing decisions. The rewards company’s previous attempts to analyze the 185 million rows of sales data were cumbersome and limited:
Only a subset of the data could be analyzed, which required the assumption that the sampling was a proper representation of the dataset as a whole.
It was extremely time-intensive; Even from the limited subset, an analyst could only process a further limited number of rows at a time, so the analysis required multiple iterations, reviews, and deduplication to actually derive the baskets needed to attempt to identify trends.
This employee engagement and rewards company partnered with Zirous to evaluate current limitations and determine how to quickly deliver a custom solution that would provide new analytic possibilities and operational insights.
The teams worked together to:
- Understand business goals and current architecture
- Identify all data sources and formats
- Assist and empower the company’s in-house resources
- Implement a sustainable data analytics solution
The forward-thinking development resulted in a flexible design that allows for additional sources to be added in the future with minimal impact on existing processes.
Increased Resource Availability
Data analysis previously took hours, required a lot of manual effort by data analysts, and was only performed on a subset of the data. By utilizing the processing power a Big Data platform offers, the entirety of the 185+ million rows of point-of-sale data was analyzed in under 10 minutes.
Additionally, filters now quickly recalculate market baskets on segments of the data based on date range, store number, product type, and more.
This company’s new data analytics architecture opens up new revenue streams and enables them to offer value to customers they were unable to provide before.
It also provides the company with a sustainable architecture to perform their own internal analytics that can tangibly prove the cost benefits of their services. All of this additionally contributes to strengthening their customer relationships.
The development of this platform was done in modular form to allow the addition of future data sources without disrupting the current design.
Implementation of a Hortonworks platform provided a predictive and prescriptive analytical tool with Nifi continuously ingests point-of-sale data into Hive from the company’s network file storage location, and then also stores it in Hortonworks Data File Storage (HDFS).
Once imported, the Zeppelin notebooking tool performs analysis over the entire dataset using SQL queries that run against the Hive tables.