Average Speed of Answer

How ASA is measured

The clock starts when the call enters the agent queue, usually after the IVR menu, and stops when an agent answers. Time spent listening to IVR prompts is normally excluded, which is why a caller can feel they waited three minutes while the report shows 40 seconds.

Most ACD platforms calculate ASA per interval (15 or 30 minutes), then roll it up to the day. When rolling up, weight by calls answered. Averaging 48 interval ASAs equally gives the quiet 6 a.m. interval the same influence as the 11 a.m. peak, and the result looks better than callers experienced.

Typical ASA targets

There is no universal standard, but common targets are:

• General customer service: 20 to 60 seconds.
• Sales and order lines: under 20 seconds, because waiting callers buy elsewhere.
• Technical support and complex service: 60 to 120 seconds is often accepted.
• Emergency and urgent health lines: a few seconds, governed by their own standards.

Many organizations set ASA through the contract or service level agreement with a client, often paired with an abandonment ceiling. In regulated industries such as Medicare plans or utilities, regulators may also set answer-time rules, so check what applies to your line.

ASA vs service level

Service level is the percentage of calls answered within a threshold, for example 80% in 20 seconds. ASA is the mean wait. They move together but tell different stories.

Service level shows how many callers got a good experience. ASA shows the average, which long waits pull up sharply. A queue at 80/20 can have an ASA of 15 seconds or 60 seconds depending on how badly the remaining 20% of callers wait. Track both, and add the longest wait or the 90th percentile wait if your platform reports it. That number is what angry customers remember.

What drives ASA up

ASA is a staffing result. It rises when workload outruns the agents available:

• Forecast misses on volume or handle time, especially Monday mornings and after billing runs or outages.
• Shrinkage above plan: absences, long meetings or training pulled into peak intervals.
• Poor schedule fit: enough total hours but not in the right intervals.
• Rising handle time, for instance after a new product launch or with a large class of new agents.

Because queueing is non-linear, small gaps matter a lot. With an Erlang C model, removing one or two agents from a 20-agent group at high occupancy can double ASA. That is why real-time analysts watch ASA by interval and move breaks or pull agents off back-office work as soon as it climbs.

ASA and abandonment

ASA only counts callers who stayed. When waits get long, impatient callers hang up, which removes their long waits from the formula and can make ASA look better on a bad day. Always read ASA next to call abandonment rate. A drop in ASA with a jump in abandonment is not an improvement.

A practical rule: report ASA, service level and abandonment together on one line of the daily report, by queue and by interval. Any one of them alone can be gamed or misread.