Agent utilisation rate

Agent utilisation rate is the proportion of an agent's working time spent directly handling customer conversations. If an agent is available for eight hours and spends six hours actively working on conversations (including research, composing responses, and post-conversation wrap-up), their utilisation rate is 75%.

The metric excludes scheduled breaks, training, team meetings, and other non-conversation activities. It specifically measures how much of the time designated for customer interaction is actually spent on customer interaction. The remaining time is idle time: periods where the agent is available but no conversations are assigned.

Agent utilisation rate is a double-edged metric. Too low and the team has excess capacity that could be redeployed or reduced. Too high and agents are overloaded, which leads to longer response times, lower quality interactions, higher error rates, and eventually burnout and turnover. The relationship between utilisation and service quality is non-linear: performance degrades gradually at moderate utilisation levels but collapses rapidly as utilisation approaches 100%.

This non-linear relationship is explained by queueing theory. When utilisation is high, any small spike in demand creates disproportionately long queues because there is no buffer capacity to absorb the spike. At 70% utilisation, a 10% demand increase is easily absorbed. At 95% utilisation, the same increase causes queue times to skyrocket. Support leaders must manage utilisation within a range that provides sufficient buffer for demand variability.

A metric tree breaks agent utilisation into its components, revealing what agents spend their time on and where capacity is consumed or wasted.

The tree reveals that utilisation is not simply about handling more conversations. If agents spend 25% of their active time on research because knowledge base articles are poor, fixing the knowledge base would reduce per-conversation handle time and allow the same utilisation rate to cover more conversations.

Connecting utilisation to average handle time, conversations per teammate, and first response time creates a complete capacity picture. When utilisation rises, handle time tends to stay constant but first response time degrades because the queue grows. The tree makes this relationship visible and helps leaders identify the utilisation threshold where service quality begins to deteriorate.

The acceptable range increases as the channel becomes more asynchronous. Synchronous channels (phone, live chat) require buffer capacity because customers expect immediate responses. Asynchronous channels (messaging, email) tolerate higher utilisation because response time expectations are measured in hours rather than seconds.

1. 1

   Align staffing to demand patterns

   Support demand is rarely constant throughout the day or week. Analyse historical conversation volume by hour and day to identify peaks and troughs. Schedule shifts to match demand patterns, using part-time staff or flexible scheduling to cover peaks without overstaffing during quiet periods.

2. 2

   Enable concurrent conversation handling

   For chat and messaging channels, allow agents to handle multiple conversations simultaneously. This fills the idle time within individual conversations (waiting for customer responses, loading account data) with productive work on other conversations. Increase concurrency gradually and monitor quality metrics to find the right balance.

3. 3

   Reduce handle time through better tooling

   Every minute saved per conversation frees capacity that can be applied to additional conversations or used as buffer. Invest in macros and templates for common responses, customer context panels that eliminate research time, and diagnostic tools that speed up troubleshooting.

4. 4

   Deflect simple queries to self-service

   Simple, repetitive queries consume agent time that could be spent on complex issues. Expand self-service coverage through better knowledge base content, chatbots for common questions, and automated workflows for routine requests like password resets or order status checks.

5. 5

   Monitor the utilisation-quality trade-off

   Track utilisation alongside service quality metrics (CSAT, resolution rate, first response time) to identify the threshold where quality begins to degrade. Set a utilisation ceiling slightly below that threshold and treat it as a hiring trigger rather than a target to exceed.

KPI Tree lets you model agent utilisation alongside the service quality metrics it affects, creating a balanced view of support team performance. The tree can segment utilisation by agent, team, shift, and channel to identify where capacity is tight and where there is room to absorb more volume.

Linking utilisation to first response time, conversation resolution rate, and customer satisfaction score ensures that capacity optimisation does not come at the cost of customer experience. When utilisation crosses the quality threshold, the tree makes the trade-off visible and provides the data needed to justify hiring decisions.

Support operations leaders can set utilisation targets for each channel and team, with automatic alerts when the metric moves outside the optimal range. Over time, the tree builds a historical record that connects staffing decisions to their effects on both efficiency and quality.