Reading partner callbacks as part of your own endpoint story

Many “our API is slow” tickets turn out to be a chain: your endpoint waits on a partner, retries aggressively, then holds a scarce lock while the next customer request queues.

What belongs in the review

• Timing for your endpoint separately from the partner hop
• Retry policy and whether it multiplies load during partner delay
• User-visible timeout versus internal wait budgets

Application analytics that only chart your process CPU will miss this. Endpoint performance monitoring that times the chain can still keep ownership clear: partner latency is a fact; your backoff and lock behaviour are choices you can change.

Bring partner dependencies to the intake call. They belong on the scope sheet beside your own routes.