Loyalty Campaign: Performance Tracking with Amazon QuickSight.



A leading financial services company embarked on a project to develop a workflow enabling individuals to customize their dashboard. The goal was to create a dashboard facilitating anomaly detection and model mapping, empowering users to efficiently identify the most suitable model.


  • Accurate performance summaries needed precise financial data, particularly due to the unique calculations required for each user.
  • The definition and collection of metrics were hindered by limited available information. This limitation affected the project's progress, particularly in understanding user requirements and optimizing the workflow.
  • The volume of data involved reached terabytes, necessitating optimizations to ensure efficient data access within the required latency. This challenge added complexity to the development process, requiring careful consideration of data management strategies.


  • Several AWS services were utilized to create new databases with tables tailored to aggregate information based on specific parameters for individual customer dashboard requirements. These services included Amazon Step Functions, Amazon Lambda, Amazon S3, Amazon Athena, AWS CloudWatch, and AWS Glue.
  • Following complex calculations on the aggregated data, a comprehensive view was formulated and utilized to populate the dataset in Amazon QuickSight, enabling the creation of the requisite dashboards.
  • A programmatic approach to dashboard creation was adopted to provide users with a personalized experience. Visuals, filters, parameters, etc., were predefined and saved. An automated workflow, powered by Amazon Lambdas in a step function, executed the automatic creation of multiple customer dashboards based on user selections and predefined definitions.
  • To ensure data privacy between users, each user was allocated its own database in Amazon Glue, along with access policies.
  • Dashboards were initially developed as exploratory tools for users to detect anomalies. Later stages involved integrating outputs from ML models into these dashboards.


Through data analysis, the team pinpointed spikes, trends, and weaknesses across various stages of the business process, enabling them to propose precise improvements. Also, they streamlined operations by decreasing SPICE capacity and associated costs by 20%, optimizing efficiency without compromising analytical capabilities.

Key Data Points

  • Detection of spikes, trends, and weak points at various business process stages
  • Reduction of SPICE capacity and associated costs by 20%.