Small call centers usually measure too much and act on too little. This is the shortlist that connects to money, and the list of numbers that feel productive to track and are not.
The metrics that matter
1. Utilisation
Utilisation = productive hours / paid hoursThe number that determines whether your pricing works. Target 75-85%. Above 85% is usually a measurement error or a pace that produces attrition. Below 70% means you are paying for time that does not turn into work.
This is the input to every quote you issue. See pricing and unit economics.
2. Occupancy
Occupancy = (talk time + wrap time) / logged-in timeHow busy an agent is while at their desk. Target 70-85% for outbound. Sustained occupancy above 90% burns people out and shows up as attrition two months later.
Occupancy and utilisation are frequently confused. Occupancy is about the shift; utilisation is about the payroll.
3. Contact rate (outbound)
Contact rate = live human answers / dial attemptsDepends on list quality, time of day, and caller ID reputation far more than on agent skill. A collapsing contact rate on a previously good list usually means your numbers are being spam-labelled — see DIDs and caller ID reputation.
Typical B2C cold ranges run 8-20%. B2B direct dials run higher.
4. Conversion per productive hour
Conversions per productive hour = outcomes / productive hoursThe single most useful outbound number. It combines contact rate, agent skill, and offer strength into one figure that maps directly onto revenue, and it is hard to game — an agent cannot inflate it by dialing faster.
5. Average handle time (AHT)
AHT = (total talk + hold + wrap) / calls handledUseful as a trend within one campaign; meaningless across campaigns. Do not manage AHT in isolation. Pushing it down while first call resolution falls creates repeat calls, which raises total cost and damages the client relationship. If you reduce AHT, watch FCR and conversion at the same time.
6. First call resolution (inbound)
FCR = issues resolved on first contact / total issuesThe best predictor of customer satisfaction and of total cost to serve. Target 70-80%. Every unresolved call generates another call.
7. Abandon rate
For outbound predictive, this is a compliance ceiling before it is a performance metric — commonly capped at 3% of answered calls. See compliance.
For inbound, under 5% is healthy; above 10% means understaffing.
8. Attrition
Monthly attrition = leavers / average headcountAnnualised industry rates commonly run 30-45%. Every point matters: replacement costs roughly one to two months of salary in recruitment and lost productivity.
9. Quality score
Your QA scorecard average, per agent, per week. Only meaningful if compliance items are pass/fail and the score is discussed with the agent — a number nobody reviews changes nothing.
10. Revenue per productive hour
RPPH = client revenue / productive hours deliveredCompare against fully-loaded cost per productive hour. The gap is your margin, expressed in the only unit that scales cleanly.
11. Schedule adherence
Adherence = time worked as scheduled / scheduled timeTarget above 90%. For inbound especially, an agent taking a break at the wrong time is as damaging as an absent one.
12. Answer seizure ratio (ASR) and average call duration (ACD)
Telecom-side health. A sudden ASR drop on one destination usually means a carrier route problem rather than anything your agents did. Worth watching because it distinguishes "our people got worse" from "the route got worse."
The vanity metrics
Total calls made. Rewards dialing fast and hanging up. An agent who makes 400 calls and books nothing outperforms nobody. If you must track volume, track it alongside conversion per hour, never alone.
Total talk time. Longer is not better. Without conversion attached it rewards rambling.
Login hours. Measures attendance, not work.
Leads touched. Encourages burning through a list rather than working it.
Raw appointment count without show rate. Booked appointments that do not attend are worse than useless — they cost the client's time and destroy trust. Always pair with show rate.
A minimal dashboard
For a center under 20 seats, this fits on one screen and is enough:
Daily
| Metric | Why |
|---|---|
| Productive hours by agent | Utilisation input |
| Contact rate | List and caller ID health |
| Conversions | Revenue |
| Conversions per productive hour | The real performance number |
| Abandon rate | Compliance ceiling |
Weekly
| Metric | Why |
|---|---|
| Utilisation | Pricing validity |
| QA score by agent | Quality trend |
| Show rate (if appointments) | Real vs booked |
| Revenue per productive hour | Margin |
Monthly
| Metric | Why |
|---|---|
| Attrition | Cost and culture |
| Gross margin per seat | Business health |
| Client-reported outcomes | The number that renews the contract |
Reading the numbers together
Individual metrics mislead. Combinations diagnose:
| Pattern | Likely meaning |
|---|---|
| High contact rate, low conversion | List is fine; script, offer, or agent skill is not |
| Low contact rate, high conversion | Good agents on a bad or over-worked list |
| High occupancy, rising attrition | Pace is unsustainable |
| Falling AHT, falling FCR | Agents rushing; costs will rise next month |
| High appointments, low show rate | Agents over-promising or soft-booking |
| Contact rate falling week over week | Caller ID being spam-labelled |
| Good QA scores, poor conversion | Scorecard measures compliance, not selling |
That last row is common and worth watching for. A QA rubric that only checks disclosure and politeness will happily give full marks to an agent who never asks for the business.
Where to go next
- Call center pricing and unit economics — turning these metrics into a defensible rate.
- Hiring and training agents — the QA scorecard in detail.
- Scaling from 10 to 50 agents — which metrics change with size.
Frequently asked questions
What is the difference between occupancy and utilisation?
Occupancy is the share of logged-in time an agent spends handling calls, including wrap-up. Utilisation is the share of paid time that is productive, which also accounts for breaks, training and meetings. Occupancy measures how busy an agent is while working; utilisation measures how much of what you pay for turns into work.
What is a good average handle time?
There is no universal figure — it depends entirely on the campaign. Simple order taking may run 90 seconds while technical support runs 12 minutes. What matters is the trend within one campaign and whether shorter handle times are damaging resolution or conversion. Cutting AHT while first call resolution falls makes the operation worse, not better.
Which single metric should a small call center track first?
Conversion per productive hour for outbound, or first call resolution for inbound. Both connect directly to revenue and both resist gaming in ways that call-count metrics do not.
What abandon rate is acceptable?
For outbound predictive dialing, regulatory ceilings commonly cap it at 3 percent of answered calls, and that is the operative limit rather than a performance target. For inbound queues, under 5 percent is generally healthy and above 10 percent indicates understaffing.