The Production Story Several years ago, my team made a decision that felt obviously correct: If a downstream call fails, retry it. More retries, more resilience. We set three retries on every integration touching our order-fulfillment platform, shipped it on a Thursday, and went home feeling good about our reliability posture.