When every mortgage application runs through the same engine, a defect is not a bug. It is a wrong financial decision reaching a real borrower.
Executive Summary
“When the decisioning engine is wrong, no one sees a software error. They see an incorrect mortgage decision.”
A mortgage platform enabling thousands of brokers to submit applications and underwrite mortgages had reached a testing inflection point. From the decision engine to complex REST and SOAP APIs and enhanced broker frontends, every component was business-critical, and manual testing across frequent builds and multiple regression cycles could not keep up.
ML arteka delivered Microsoft Azure-based DevOps automation, building custom test cases in JavaScript to handle the API complexity and a data-driven API testing framework that reads parameters from CSV and XML and validates each endpoint. The mortgage business rules were implemented on the testing side, so the decision engine’s result was compared on the fly against the expected result.
The result: 50% fewer production defects. Five times greater test coverage through CI/CD. Over 100 issues resolved before they could reach a broker.
Business Outcomes
These are not test metrics. They are business outcomes measured in decisioning accuracy and release confidence.
Compared to manual testing.
Through CI/CD automation.
Found before reaching a broker.
The Transformation
From weekly regression to continuous, business-logic-aware validation.
- 1
DiscoveryModel the Decision, Not Just the APIMapped the decision engine, the REST and SOAP microservices, and the mortgage business rules that determine a correct outcome. - 2
FrameworkData-Driven API TestingBuilt a framework that reads parameters from CSV and XML files and sends requests to each endpoint using Java libraries. - 3
AutomationAzure DevOps and CI/CDAutomated the suite on Microsoft Azure so tests run on every build, validating API contracts on the fly. - 4
ValidationCompare Against the RulesImplemented the mortgage rules on the testing side and compared the engine’s result against the expected result in real time.
The Business Challenge
When a Rules-Heavy Engine Meets Manual Testing
Ensuring the smooth operation of every component, and testing all of them by hand, had become overwhelming.
Human-intensive effort
Managing resources, budget, and on-time delivery while testing manually.
Frequent builds
Multiple regression cycles on a fast release cadence.
Complex APIs
Nested structures across REST and SOAP microservices.
Business Outcomes in Detail
What the Numbers Mean
Compared to manual testing.
Through CI/CD automation.
Found before reaching a broker.
Technology Snapshot
Purpose-Built for Mortgage Decisioning
Executive Questions and Answers
The questions leadership tends to ask when evaluating an approach like this.
Accuracy
How is decisioning testing different from standard API testing?
Standard API testing checks that an endpoint returns a valid response. Here the framework re-implements the mortgage rules and compares the engine’s decision against the expected decision, so it validates correctness, not just connectivity.
Speed
How does continuous testing change the release process?
With CI/CD on Azure DevOps, tests run on every build and API contracts are validated on the fly, giving constant feedback instead of a slow, manual regression cycle.
Risk
What is the value of catching defects before release?
More than 100 issues were resolved before reaching a broker. In mortgage decisioning, a defect that reaches production is a wrong financial decision for a real borrower.
