How Do You Approach End to End Testing When Third-Party APIs Are Involved?

One of the trickiest parts of end to end testing is dealing with third-party APIs — payment gateways, email services, shipping providers, or analytics platforms. These are critical to the user flow, but they’re also unpredictable, rate-limited, or unavailable in staging environments.

Some teams use mock servers or simulate responses for common scenarios, but that comes with trade-offs. You lose visibility into real-time issues like API schema changes, response delays, or unexpected behavior from the third-party service. On the other hand, relying on live APIs in test environments can lead to flakiness and even unintended side effects (like sending real emails or charging real cards).

There’s also the question of what to record and replay. Tools like Keploy offer an interesting middle ground by capturing actual API interactions and generating mocks based on real behavior, making it easier to test flows involving third-party services without depending on live systems.

Would love to know how others in the community are handling this. Do you mock third-party APIs in your end to end testing, use contract testing, or run any tests against live services? How do you balance reliability with realism?