Case
Delivery tracking keeps working, even when partner services are down
A delivery platform with tens of thousands of business customers. It relied on outside services to find out where a shipment is and what delivery will cost.
What happened: those services were sometimes slow, and sometimes stopped responding altogether. The platform's users felt it every time.
After the fix, the platform remembers answers it has already received and doesn't ask the same thing twice. When a service is down, it shows the last known data, or a clear message instead of an error if there's none. It doesn't flood the services with repeat requests, and it writes down what went wrong so it can be fixed fast. Now users barely notice when a partner has trouble.
