Tracker Event Validation Framework

Problem

During a large-scale e-commerce platform migration, every change to the product and checkout pages risked breaking the integrity of tracker event data — events like product_view, product_click, widget_view, widget_click, and add_to_cart sent to the analytics system. Validating these JSON payloads by hand, across dozens of events on three platforms (Web, Android, iOS), was slow and error-prone, and it did not scale as the number of events and fields kept growing. A single missed field or wrong data type could corrupt the analytics dashboard without anyone noticing until after release.

Role & Contribution

As the QA Engineer, I designed and built this framework from the ground up using Java and TestNG. I owned the data-driven architecture and wrote the per-schema validators that check each captured payload against its contract.

Solution

The framework follows a data-driven approach: each tracker event is defined as a JSON contract kept separate from the test code. When a payload is captured from the application under test, the framework matches it using XPath expressions with conditions rather than a plain event-name lookup. This lets a single expression cover many scenarios that share a similar payload structure — including widgets that carry the same name but differ by parameters and values — so near-identical events are matched precisely without writing a separate rule for each one. Once matched, the corresponding validator checks required fields, correct data types, and allowed enum values.

Adding coverage for a new event means adding one JSON schema file — no existing Java code is touched. The full suite runs through TestNG, so tracker regressions are caught during testing — before bad data has a chance to land in the analytics dashboard.

Impact

• Replaced manual JSON checking with automated, repeatable validation across Web, Android, and iOS
• Cut new-event onboarding to a single JSON schema file, with zero changes to test code
• Caught tracking regressions during testing, before they reached the analytics dashboard
• Scaled cleanly as events and fields grew, keeping maintenance cost flat