Metric guide

Average speed of answer (ASA)

Q: What is average speed of answer (ASA) in a contact centre?

Average speed of answer (ASA) is the mean wait time across all answered calls, calculated as total queue time divided by the number of calls answered. It includes calls answered immediately (zero wait time), which pulls the mean down relative to the actual wait experienced by callers who did queue. ASA is an output metric from Erlang C modelling and from your ACD/telephony system reports.

Q: What is a good ASA for a contact centre?

Typical ASA benchmarks for inbound voice: 15–30 seconds is typical, under 15 seconds is tight performance, 30–60 seconds is acceptable for lower-intensity operations, above 90 seconds is consistently problematic and indicates chronic understaffing or forecasting failures. ASA benchmarks vary by industry: financial services and healthcare emergency lines typically target under 20 seconds, while back-office support lines may accept 45–60 seconds. Your own operation's trend over time is a more useful signal than a cross-industry benchmark.

Q: Why does ASA fall when calls are abandoned?

ASA is calculated only on answered calls. When a long queue forms and callers abandon after waiting 2–5 minutes, those calls are excluded from the ASA calculation. This means ASA can improve, or stay flat, during periods of high abandonment. If your ASA looks good but abandonment rate is rising, ASA is being distorted by caller self-selection: only the most patient callers remain in queue, so the mean wait of answered calls appears reasonable. Always report ASA and abandonment rate together, not ASA alone.

ASA is the mean wait time across all answered calls, but it is not the same as service level, and it can be misleading when abandonment rates are high. This guide covers what ASA is, how it differs from service level, and when to use each metric.

ASA formula

Calculation

ASA = Total queue time ÷ Calls answered

In seconds. Only answered calls are included; abandoned calls are excluded.

Worked example

Calls answered

320

Total queue time

4,800s

ASA

15s

4,800 ÷ 320

ASA includes callers who were answered immediately (zero wait). If 200 of your 320 answered calls were picked up immediately, those 200 zeros pull the average down significantly. This is why ASA can look good even when queuing callers have long waits.

ASA vs service level: the key difference

Service level

% of calls answered within X seconds

✓Directly tied to customer waiting experience
✓Not distorted by zero-wait calls
✓Better for regulatory reporting
→Does not show the actual wait time, only whether the threshold was met

ASA

Mean wait time across all answered calls

✓Easier to trend and explain to non-technical stakeholders
✓Useful for benchmarking cost efficiency
✓Outputs directly from Erlang C modelling
✗Distorted by high abandonment; can improve when service worsens

The abandonment distortion problem

When queues build and callers abandon after 3–5 minutes, those calls are excluded from the ASA calculation. Only patient callers who stay in queue are counted, which can make ASA look reasonable even while service level craters. Always report ASA alongside abandonment rate. If abandonment rises while ASA stays flat or falls, ASA is being distorted.

ASA benchmarks

ASA benchmarks for inbound voice. These are attainable ranges, not optimal targets.

ASA rangeInterpretationTypical operations

< 10sExcellent

Well-staffed voice operations; small team advantage (high agent availability)

10–20sGood

Typical for regulated operations (FCA, NHS); leisure lines with clear SLA targets

20–40sAcceptable

Most mid-size inbound operations; typical industry benchmark

40–90sNeeds attention

Common in understaffed or poorly-forecast operations; acceptable for low-stakes queues

> 90sProblematic

Chronic staffing or forecasting problem; high abandonment likely

ASA as a staffing output metric

Erlang C calculates predicted ASA as a direct output from your volume, AHT, and agent count inputs. This makes ASA useful for staffing decisions: you can model the ASA impact of adding or removing an agent.

agents

Service level

71%

Predicted ASA

52s

Understaffed: a significant queue forms

agents

Service level

80%

Predicted ASA

28s

Target staffing: meets 80/20 SL

agents

Service level

89%

Predicted ASA

12s

Well-staffed: minimal queue, low ASA

Example: 80 calls/hour, 5 min AHT, no shrinkage (for illustration). Run your own numbers in the Erlang C calculator.

ASA questions

What is average speed of answer (ASA) in a contact centre?

Average speed of answer is the mean wait time across all answered calls, calculated as total queue time ÷ calls answered. It includes calls answered immediately (zero wait), which pulls the average down. ASA is an output metric from Erlang C and from your ACD reports.

What is the difference between ASA and service level?

Service level measures what percentage of calls were answered within a threshold (e.g. 80% in 20 seconds). ASA measures the average wait time across all answered calls. They move in the same direction but respond differently to queue shape. ASA is more useful for trending and benchmarking; service level is more directly tied to customer experience and regulatory reporting.

What is a good ASA for a contact centre?

Under 20 seconds is good for regulated or high-intent voice operations; 20–40 seconds is typical for most mid-size inbound operations; 40–90 seconds needs attention and indicates staffing or forecasting problems; over 90 seconds consistently signals chronic understaffing. Your own trend over time is a more actionable signal than a cross-industry benchmark.

Why does ASA fall when calls are abandoned?

ASA is calculated only on answered calls. When long queues form and callers abandon after 2–5 minutes, those calls are excluded. This means ASA can look stable or even improve during periods of high abandonment, because only the most patient callers remain, pulling down the mean. Always report ASA alongside abandonment rate to catch this distortion.