Call center service level is the share of calls answered within a target time, written as two numbers: 80/20 means 80% of calls answered within 20 seconds. That pairing is the long-standing industry default (Call Centre Helper).
It is the only contact-centre metric that is simultaneously a promise to callers, a staffing input, and a budget.
Reading the two numbers
Both halves matter, and they are not interchangeable.
The percentage is the share of calls you commit to answering in time. The threshold is how long "in time" is. 80/20 and 90/60 are different operations: the first answers most calls fast and lets a minority wait; the second answers almost everyone, but nobody especially quickly.
Neither is more correct. An emergency line should not run 80/20 — it should answer nearly everything nearly immediately, and pay for the agents that requires. A billing queue where callers expect to wait can run 70/60 and free budget for something else.
The failure is inheriting 80/20 without asking whether it fits, then staffing to a promise nobody chose.
What service level does not tell you
Service level says nothing about the 20%. In an 80/20 operation, one in five callers waits longer than 20 seconds — and the metric is silent on whether that is 25 seconds or 11 minutes.
This is why service level should be read next to two other figures. Average speed of answer shows the typical wait, though it hides the tail. Longest wait or the 95th percentile shows what the worst-served callers actually experience. A queue at 80/20 with a 14-minute longest wait is failing people the headline number says nothing about.
Abandonment is the honest check. If callers are hanging up, service level can look fine — an abandoned call is often not counted as a missed answer — while the queue is plainly not working.
How it drives headcount
Service level is the input the Erlang C model solves against. Fix the target, the volume and the handle time, and the agent count falls out.
The relationship is not linear, and the last few points are expensive. Moving 80/20 to 85/20 costs a few agents. Moving 90/20 to 95/20 costs far more, because you are now staffing for progressively rarer arrival clusters. Before buying the last five points, check they are worth more than the occupancy you spend to get them — the agents that deliver them sit idle most of the day by design.
Measuring it per interval
A daily service level of 82% can hide an 11am at 55%. Callers experience the interval they called in, not the daily average, and so does your team.
Report per 15- or 30-minute interval, and compare against the same interval last week rather than yesterday — contact centres are weekly-cyclical, and Monday looks nothing like Thursday.
The service level calculator works the achieved figure from answered and offered calls per interval, and the Erlang C calculator goes the other way, from a target to a headcount.
Frequently asked questions
What is a good call center service level? 80% of calls answered within 20 seconds is the common default (Call Centre Helper), but the right target depends on what the queue is for. Emergency lines should be higher; low-urgency queues can be lower.
Should abandoned calls count against service level? Include them. Excluding abandons lets a queue report a healthy figure while callers are giving up, which is the situation the metric exists to catch.
What is the difference between service level and average speed of answer? Service level is the share answered within a threshold. Average speed of answer is the mean wait. A queue can have a good average and still fail a large minority badly.
Why is 95/20 so much more expensive than 80/20? Because you are staffing for increasingly rare arrival spikes. The extra agents are idle most of the time by design, which is what makes the last few points cost more than the first eighty.
