Contact centre metrics explained: what each one makes people do

Contact centres measure almost everything, and that is a strength: few parts of a business can see their own performance as clearly, interval by interval. The risk is that a number put on a wallboard becomes the job. Tell agents that handle time is what matters and calls get shorter, whether or not the customer's problem is solved. Tell planners that occupancy must be high and schedules get tighter, whether or not people can keep it up.

So the first question about any metric is not how to improve it, but what it will make people do. The table below sets out the core measures, what each one tells you, and what happens when it is pushed too hard. The rest of the guide explains how we at QVCCS think each one should be used, drawing on our leadership's experience in contact centre technology since 1984.

Genesys Cloud defines average speed of answer as the average time an interaction waits in the queue before an agent answers; interactions that are never answered are left out. Service level is the percentage of conversations answered within a target time you set, and the abandon rate is the share of offered interactions where the customer disconnects before reaching an agent. Together they describe the wait, which is the first part of every assisted contact.

Our view is that average speed to answer is the metric that matters most to customers, because it is the clearest sign of how seriously an organisation takes them. While customers wait, their attention moves from the problem they were going to explain to the delay itself. They arrive irritated, the effort they have spent is already high, and the agent who answers has more work to do than if the call had been answered promptly. In Dave Tidwell's words: "We must stop measuring 'How bad is good enough?'"

An average can hide a long tail, so read speed to answer alongside service level and abandons for each queue, and set targets from what your customers find acceptable rather than from what last year's staffing allowed. Callbacks, messaging and self-service can take pressure off the queue, but they are not a substitute for answering people who chose to call.

Average handle time in Genesys Cloud covers talk time, hold time and after-call work, and for outbound calls the dialling and contacting time too, divided by the number of interactions handled. At its simplest it is the time it takes to deal with an enquiry fully and completely, and it is one of the most useful numbers a contact centre has, provided it is read rather than squeezed.

Used well, handle time points to problems elsewhere. If track-and-trace enquiries on an order line usually take about three minutes and twenty seconds and suddenly take six, the cause is rarely the agents, their behaviour or their scripts. Something has changed: a slow system, a broken integration, a carrier update that customers do not understand, or a new policy that needs explaining. Handle time tells you where to look.

Used badly, it becomes a target, and pressing agents to cut handle time while the queue is under pressure is another way of asking them to give customers a poorer experience. Contacts get rushed, problems are half-solved and customers call back. Look instead at the outliers on both sides of the curve: agents who are consistently far quicker than their colleagues may be cutting corners, and agents who are far slower may be going well beyond the process, perhaps because the process does not work. Both deserve a conversation, and neither should be fixed by a number on a wallboard.

Occupancy measures how much of an agent's on-queue time is spent handling interactions; Genesys Cloud calculates it as interacting time divided by on-queue time, for the plan and for what actually happened. Shrinkage is paid time that is not available for contacts: annual leave, sickness, training, meetings and coaching. Schedule adherence, which Genesys describes as a common contact centre metric, compares what each agent is doing now with what their schedule says.

Be realistic about shrinkage. Sickness, holiday pressure at peak times, team meetings and training all take people off the phones, and plans that assume otherwise fall short every day. In our experience, measured occupancy usually comes in below the planned figure for exactly that reason: shrinkage was underestimated. Genesys Cloud's historical shrinkage view compares planned and actual non-interacting time, which makes the gap visible so the next plan can close it.

High occupancy looks efficient on paper, but it leaves no time to recover after a difficult call, to learn or to help a colleague, and sustained pressure shows up in sickness and people leaving. Adherence is best used to improve the plan and make intraday decisions, such as moving a break or offering overtime, rather than to watch every minute. If agents are regularly out of adherence for the same reason, the schedule is wrong, not the people.

First contact resolution asks whether the customer's need was met without them having to come back. It is harder to measure than the time-based metrics: most contact centres use repeat contacts from the same customer within a set window, the outcome agents record, and what customers say in surveys. It is worth the effort, because every repeat contact is demand the organisation created for itself, and the patterns behind repeat contacts usually point to broken processes rather than to individuals.

Customer satisfaction, customer effort and Net Promoter Score all ask customers directly. Net Promoter Score asks how likely the customer is to recommend you on a scale of 0 to 10, and subtracts the percentage of detractors, who score 0 to 6, from the percentage of promoters, who score 9 or 10. Genesys Cloud survey forms can include one NPS question about a specific interaction, scored separately from the rest of the survey. Whichever measure you use, read the comments as well as the scores, watch who is and is not being surveyed, and never let agents choose which customers are asked.

No single metric tells the truth on its own. Read them in pairs: speed to answer with abandons, handle time with repeat contacts, occupancy with adherence and sickness, satisfaction with resolution. Read them by queue, because a figure averaged across sales, service and complaints describes no real customer. Watch the trend across days and weeks rather than reacting to a single interval, and always ask what changed before asking who is to blame.

Share numbers with agents in a way that helps them improve. Agents who can see their own results alongside their team's tend to adjust without being told, which is one reason gamification works. Peer comparison should be handled with care: reward improvement and team goals as well as top scores, and make sure nobody is shamed at the bottom of a board. Coaching should reinforce the right behaviours, not threaten, because agents who are coached through fear express it in their next customer conversation.