Call centre software will offer you dozens of measurements. Most of them describe the past without suggesting an action, which makes them decoration.
Four are worth a weekly look. Each one, used carelessly, can also be gamed — so each comes with the warning.
1. Calls answered within twenty seconds
The share of calls picked up before the caller starts wondering whether anyone is there.
Twenty seconds is roughly four rings. Beyond that, people begin deciding you are not available.
What good looks like. For a small Kenyan support desk, 80% within twenty seconds is a strong number and a reasonable target. Below 60%, callers are noticing.
How it is gamed. Agents answer and immediately put the caller on hold. The number looks excellent and the experience is worse than waiting in a queue. Watch hold time alongside it.
2. Abandoned calls
The share of callers who hung up before reaching anyone.
This is the figure nobody had before they bought a system, and it is almost always the one that changes a mind. These are customers who tried to reach you and failed, and on a mobile phone they are invisible — they simply did not happen.
What good looks like. Under 5% is healthy. Above 10% you are losing real business, usually in one or two specific hours.
How it is gamed. Excluding “short abandons” — people who hung up within a few seconds. Some of those are misdials; most are people who heard a queue message and gave up. Count them and look at the trend rather than the absolute.
There is a longer piece on what these actually cost.
3. Average handling time
How long a call takes, including the note written afterwards.
Useful for two things: working out how many people you need, and spotting a change. Useless as a target to push down.
What good looks like. There is no good number — it depends entirely on what your calls are. Three minutes for order status, twelve for technical support, both fine. What matters is whether it is drifting and why.
How it is gamed. Badly, and it is the most damaging one. Pressure agents on handling time and they will end calls early, skip the note, and avoid the complicated customer. You will see the number improve and the repeat-call rate rise.
Treat it as a planning input, not a performance target.
4. First contact resolution
The share of issues sorted without the customer having to ring again.
This is the one that correlates with whether customers actually like you, and the hardest to measure honestly. The practical proxy is: how many customers rang more than once about the same thing within a week?
What good looks like. Above 70% is solid. If you cannot measure it precisely, measuring the repeat-call rate is close enough to be useful.
How it is gamed. Logging the second call as a new issue. If your outcome list makes that easy, your number will be excellent and meaningless.
What to leave off the wall
Calls per agent per day. Rewards speed and punishes the person who takes the hard calls. If you must use it, use it to spot someone struggling, never to rank people.
Occupancy pushed high. Running agents at 95% busy sounds efficient and produces burnout, then churn, then a hiring cost that dwarfs the saving.
Anything nobody acts on. If a figure has been on the report for three months and has never changed a decision, remove it.
How to use the four
Once a week, fifteen minutes, in this order:
- Abandoned calls. Which hour? That is a staffing question.
- Answered in twenty seconds. Trending which way? That is the same question, earlier.
- Handling time. Has it moved? If yes, ask why — not of the agents, but of the work. New product, new problem, new form to fill in.
- Repeat calls. What are people ringing back about? That is usually a process to fix, not a person to coach.
Then listen to three calls. Not to score them — to remember what the numbers are describing.
The numbers tell you where to look. Only the calls tell you what is actually happening. Scoring them fairly is a separate skill.