Drops can occur at span creation, sampling, exporter queues, transport, or ingestion. Locate the stage before increasing limits.

Checks

1. Compare application request volume with created spans and forwarded spans over the same interval.
2. Review sampling decisions and confirm that the observed reduction matches policy.
3. Check exporter and collector counters or logs for queue saturation, rejected items, retries, and timeouts.
4. Look for traffic bursts, large batches, or recent increases in span size.
5. Confirm network stability and endpoint availability.
6. Review client usage notifications and configured ingestion governance.

Reduce unnecessary attributes and events before increasing queue or batch settings. Apply backpressure and capacity changes gradually, then verify memory use and drop counts. Ingestion usage limits can generate warnings without blocking traffic, so do not assume every missing span was rejected by a quota.