API rate limiting is now adaptive instead of fixed. The system reads the actual rate limit each account gets from Amazon and adjusts its request pacing accordingly, resulting in faster syncs for accounts with higher quotas and no more unnecessary throttling.
How it worked before
Previously, all accounts were rate-limited at the same conservative pace, regardless of their actual API quota from Amazon. This meant accounts with higher quotas (which Amazon grants based on seller performance and sales volume) were artificially slowed down to the same speed as smaller accounts.
How it works now
Every time the system makes an API call to Amazon, Amazon's response includes rate limit headers that indicate how many requests are allowed per time window. The system reads these headers and adjusts its pacing in real time. If your account has a higher quota, requests are spaced more closely together, resulting in faster catalog syncs and quicker price updates.
Graceful throttle handling
If Amazon returns a 429 (rate limit exceeded) response, the system no longer retries immediately. Instead, it backs off and requeues the request with a delay. This prevents the cascading retry loops that could lock up an account's API access. The requeued request is picked up after the cooldown period, ensuring your API access recovers smoothly.
Per-account, per-operation tracking
Rate limits are tracked separately for each combination of seller account, API operation, and region. This means a catalog sync for your Amazon US account does not interfere with pricing updates for your Amazon UK account. Each operation type (catalog, pricing, orders) maintains its own token bucket, so a heavy catalog sync will not slow down time-sensitive price changes.