Overview
Not every application has a UI. And even when it does, the UI is often the last thing built. So how do you test when there’s nothing to click?
This presentation made the case for API testing as a first-class testing discipline — not a workaround for missing screens, but a fundamentally faster and more reliable approach. API tests execute in milliseconds, don’t depend on browser rendering or CSS selectors, and give you deterministic results. When the same assertion takes 3ms at the API layer or 30 seconds through the UI, the choice is obvious.
This was also the early days of service-oriented architecture evolving toward what we’d soon call microservices. As monoliths broke apart into independently deployable services communicating over HTTP, testing at the API layer wasn’t just convenient — it became essential. You couldn’t click through a UI to test a service that had no UI. The API was the interface.
What We Covered
- Why API testing gives you better signal than UI testing for most validation scenarios
- The speed and reliability advantage: milliseconds vs seconds, deterministic vs flaky
- Setting up API test tooling from scratch (the landscape in 2010)
- Structuring API tests around user workflows, not individual endpoints
- Validating response contracts: status codes, headers, body schema, error formats
- Testing authentication and authorization at the API layer
- Managing test data and environment state for repeatable API tests
- How the rise of service-oriented architecture made API testing a necessity, not a nice-to-have
Context
This was early — REST APIs were becoming the standard architecture pattern, and the industry was beginning the shift from monoliths toward service-oriented and eventually microservices architectures. Most quality engineering teams were still UI-first in their testing approach. The idea that you could (and should) test the API layer directly, independently of any frontend, was not yet conventional wisdom.
The speed argument alone was compelling: a test that validates a response payload in 5ms is inherently more maintainable than one that launches a browser, navigates three pages, fills a form, and checks a div. But the architectural argument was stronger — as services multiplied and UIs became optional, API testing stopped being a shortcut and became the primary testing strategy.
The principles from this talk became the foundation for the later Automated Functional API Testing workshop with Postman and Newman, and eventually fed into the contract testing approach with Pact.