Discrepancy in Quality Monitoring scores after migration to remote agents

Team we are currently observing a discrepancy in Quality Monitoring scores following our migration to remote workstations two weeks ago. We are running Genesys Cloud version 1095 on AWS US-East-1 and the Historical Reporting shows significantly lower MOS values compared to Real-time Analytics dashboards. Specifically looking at /api/v2/analytics/measures/quality for the last seven days. The average MOS in HR is 3.2 but RTA shows 4.1 consistently. We have verified network bandwidth on the agent side and traceroute results show stable paths with under 5ms jitter. However data packets from the quality monitoring service seem to be dropping during export windows. Error logs do not show explicit failures but null values appear in the Quality Score field for calls lasting over three minutes. Has anyone seen this behavior after a remote migration? We require accurate historical data for compliance audits.

This occurs when streaming quality metrics exceed the ingestion buffer limits during peak volume. Check the Analytics Data Stream configuration in Settings under Reporting. Ensure Quality Monitoring is enabled for both Real-time and Historical streams separately.

Sometimes the Historian cache clears older quality records if latency exceeds 60 seconds. Verify the qualityScore field population time versus call end time in the export logs.

Adjusting the sampling rate to 100% often resolves the gap between RTA and HR.

It is important to understand how these scores aggregate before assuming a system defect. Historical Reporting calculates quality metrics over defined reporting intervals which smooths out momentary spikes found in Real-time data. If network jitter causes packet loss during the call, the score drops immediately but may recover if the connection stabilizes before the session ends.

For WFM planning we typically rely on the HR version for long-term trends rather than RTA snapshots. The discrepancy you see likely represents a difference in calculation windows rather than missing data.

We suggest exporting the raw quality scores via the reporting engine to verify the actual timestamps against your network logs. This ensures compliance requirements match the specific definition used by the reporting engine.