We need productivity metrics (specifically productive adherence and % productive time) to be calculated using actual productive work performed (e.g., real ticket-handling time, message-handling time, or activity-level work), not merely based on whether an agent is sitting in a “productive” AUX status. Context: Because AUX states don’t always reflect real work (agents may be marked productive but not actively working tickets, or vice-versa), the current metrics appear inflated and unreliable compared to our manual calculations. Our internal productivity models use time spent doing real ticket work, so state-based productivity does not align with operational expectations. Other notes: The Zendesk app partially solves this by feeding ticket-level activity into Assembled. However, it only works for agents who have the app installed, leaving gaps and inconsistent data during rollout. We are maintaining parallel reports (Google Sheets) because Assembled’s metrics still differ from our “true” productive hours. Current metrics can show over 100% productive adherence, which is one of the signals the calculation is misaligned with actual work. Related metric impacted Occupancy also needs to use activity-level signals (e.g., ticket work time) rather than agent state data. Current occupancy is inflated because it treats any “productive” AUX state as fully utilized time even if no ticket work is happening. We would expect occupancy to align with real work performed.
