When One Failed API Affects the Entire Product

One failed integration can make an otherwise well-functioning product feel completely broken.

A payment provider goes down, an authentication service times out, or a third-party API returns unexpected data. Suddenly, a problem outside your codebase becomes your product’s problem.

Each integration adds useful functionality, but it also introduces a dependency you don’t fully control. You can’t prevent every service from failing. What you can control is how your product responds when it happens.

Can it safely retry the request? Is there a fallback? Does it fail without losing data or creating duplicate actions? Will your team know about the issue before customers start reporting it?

Retries, fallbacks, clear error handling, and continuous monitoring won’t prevent every  incident. But they can contain the impact, make recovery faster, and prevent one failed integration from taking down an entire workflow.

Reliable products aren’t built on the assumption that everything will always work. They’re built to handle the moments when it doesn’t.

What has caused more production incidents for your team: your own code or third-party integrations?