Service level vs ASA: what each measures

Two numbers describe how quickly a call center picks up: service level and average speed of answer (ASA). They sound alike and they are often confused, but they answer different questions. This guide explains each one, shows how they can disagree, and says which one to set targets on.

Service level

Service level is the percentage of calls answered within a target time. The most quoted version is 80/20: 80% of calls answered within 20 seconds. It is a promise about most callers: how many get a fast answer.

The formula looks simple:

Service level % = calls answered within the target time ÷ calls offered × 100

The catch is the denominator. Some centers divide by calls offered (answered plus abandoned), others by calls answered, and some leave out calls that were abandoned in the first few seconds. None is wrong, but they give different figures from the same day, so choose a rule, write it down and keep to it.

ASA

ASA is the average time callers wait before an agent answers. Add up the waiting times of all answered calls and divide by the number of answered calls.

ASA = total wait time of answered calls ÷ number of answered calls

It describes the typical wait, which makes it easy to explain to people outside the center.

Why they can disagree

Take ten calls. Eight are answered in 5 seconds. Two wait 100 seconds.

• ASA = (8 × 5 + 2 × 100) ÷ 10 = 24 seconds.
• Service level (within 20 seconds) = 8 ÷ 10 = 80%.

By service level you hit an 80/20 target. By ASA you are above 20 seconds. Both are true: most callers were answered fast, and two had a long wait. Neither number tells you about those two callers, which is why you also watch abandonment and the longest wait.

Abandon rate and longest wait

Callers who give up before an agent answers are the real cost of slow answers.

• Abandon rate is the share of callers who hang up while waiting. Track it on its own, and decide whether very short abandons (a few seconds) count.
• Longest wait is the longest time any caller waited. It shows the tail that an average smooths over.

A good service level usually comes with a low abandon rate, but not always: if your threshold is 20 seconds and the rest wait two minutes, the service level can look fine while a small group of callers have a bad experience.

Which one to set targets on

• Service level is the better target. It speaks to the experience of most callers, and staffing formulas are built around it.
• ASA is useful for trends and for explaining results to people outside the center.
• Abandon rate and longest wait tell you how much damage slow answers are doing.

Use two or three together rather than one on its own.

How staffing moves them

Service level and ASA are both outputs of the Erlang C calculation. Given calls per hour, average handle time and the number of agents, it predicts both. For 120 calls an hour at 240 seconds each (8 erlangs of traffic), with a 20-second target:

Agents on the phones	Service level	ASA	Calls that wait	Occupancy
9	40%	157 seconds	65%	89%
10	65%	49 seconds	41%	80%
11	81%	20 seconds	25%	73%
12	90%	8 seconds	14%	67%
13	95%	4 seconds	8%	62%

Two things stand out. First, the curve is steep around the point where agents barely exceed the traffic: going from 9 to 11 agents lifts service level from 40% to 81%. Second, the returns shrink: the last agent adds only five points of service level, but lowers occupancy by another five. That is the trade-off between answering quickly and keeping agents busy. Use the Erlang C calculator to see the same table for your own numbers.

Setting a target

There is no universal right answer, but these questions help:

• What will your callers tolerate? Emergency and sales lines need quick answers. A back-office line may be fine with a longer wait.
• What does a missed call cost? If a lost call is a lost sale, a higher service level pays for itself.
• What can you staff? Each extra point costs agents, and occupancy falls as service level climbs.
• Which interval do you measure? Service level by half hour is far more honest than a daily average, which lets quiet hours hide busy ones.

What helps when service level slips

• More agents on the phones, after allowing for shrinkage.
• Shorter menus. Time spent in an IVR doesn't count as waiting for an agent, but callers feel all of it.
• Better routing so calls reach agents who can resolve them first time.
• Queue callbacks, so callers keep their place without holding the line.
• Smarter breaks and training schedules, so they avoid your peaks.

Don't average the percentages

When you report service level for a day or a week, weight it by the number of calls. Suppose one interval has 100 calls with a 90% service level, and another has 20 calls at 50%. The weighted service level is (90 + 10) ÷ 120 = 83.3%. Averaging the two percentages gives 70%, which gives a quiet interval the same weight as a busy one and badly understates how most callers were treated. Always add up calls answered within the target and divide by total calls offered.

Service level vs an SLA

An SLA (service level agreement) is a contract or internal commitment, such as "answer 80% of calls within 20 seconds, measured monthly". Service level is the measurement you use to check it. The agreement should say how the figure is calculated, over what period, and how abandoned calls are treated, so that nobody argues about the numbers later.

A simple reporting habit

For each half hour, report service level, ASA, abandon rate and the longest wait, together with calls offered. Review the worst intervals each week. Patterns show up quickly: always the same hour, always the same queue, always the day after a campaign.

Pick a target, measure it by interval, and revisit it when the numbers say callers are hanging up.