Case Study

Global Payment Processor Achieves 90% Savings with Automated BI/Data Warehouse Testing

Background

A global payment processing company invested millions of dollars building applications to help business leaders better understand and leverage the massive amount of data the company was processing. However, the initial adoption rates for these applications fell short of expectations. Word got out that the resulting reports weren’t entirely accurate, and this understandably curbed adoption.  

The organization’s QA leaders were given the challenge of “fixing this problem” by transforming the end-to-end data testing process—just as they had recently transformed the organization’s end-to-end application testing process. Based on that experience, they recognized that they not only needed to introduce new technology for test automation, but also transform people and processes to ensure that the new approach was well-adopted, scalable, and sustainable.

Challenges
  • No existing tools, people, processes, or training for data warehouse testing
  • Ad hoc SQL “spot checks” barely scratched the surface of what needed to be tested
  • Massive amounts of data (from 40 billion transactions processed annually), many different sources of data, and many transformations between source and report
  • Complex technology stacks—from 40-year-old mainframes to cloud
  • Complex requirements (processing transactions across 146 counties, 126 currencies, and 300 payment types)

“Before, different teams were checking ‘their’ data…but this wasn’t effective at catching issues introduced in the many integrations and transformations. “

Head of Enterprise Quality Assurance Team

The Tricentis Solution: End-to-End BI/Data Warehouse Test Automation

The QA leaders designed and drove an initiative to automate testing across all layers of the BI flow– Ingestion, transformation, data mart, and reporting. After evaluating the available tools, they selected Tricentis Tosca, and designed a rapid rollout with minimal disruption to the many “in-flight” projects they were responsible for testing.

In just a few months, they were up and running with end-to-end regression testing that automatically tests data as it flows through the various ingestion and transformation layers—including data validation, data profiling, and checking data accuracy, integrity, and completeness.   This was accomplished by the QA team creating a core set of reusable, scalable tests …without wrestling with SQL or scripting. Using this automated test suite, billions of rows are automatically reconciled, even with constantly changing sources. Not surprisingly, this high degree of automation has dramatically increased their testing scope while slashing testing time and costs.  

Initial Results
  • 200 million values are compared in just 20 minutes
  • 90% reduction in cost
  • 90% reduction in time to market
  • Thousands of hours of manual effort saved per month
  • Improved user adoption and productivity as a result of improved data quality
  • Mindset shift that helped mature the data testing practice

A few months after our new BI/data warehouse testing approach launched, people finally started trusting the applications. They’re highly used now, and they have become a vital tool in making critical business decisions. “

Head of Enterprise Quality Assurance Team