An inbound call center is a team, or a part of a contact center, that answers calls customers make to a business. Callers usually pass through a phone menu and a queue before a person or an AI agent picks up.
A 2003 review of call center research by Noah Gans, Ger Koole and Avishai Mandelbaum lists typical inbound work. It includes customer support, help desks, reservations and order taking. At banks, telcos and insurers it also covers card blocks, billing disputes, claims and payment arrangements.
How an inbound call reaches an agent
The call arrives from the carrier, often over SIP trunking. The phone system knows the caller's number and the number dialed, and can use both for routing.
An IVR or conversational IVR then asks why the person is calling. Some callers finish there, for example by checking a balance.
The rest pass to the automatic call distributor (ACD). It picks an agent by the rules described under call routing. When no suitable agent is free, the caller waits on hold, and some hang up before being served.
Service level, ASA and abandonment
Service level. The share of calls answered within a target time. The same review calls it the telephone service factor and says the target is typically 20 or 30 seconds.
Average speed of answer (ASA). The average time callers spend on hold before an agent answers. ASA leaves out the wait of callers who hung up, so the review says a full picture needs the abandonment rate too.
Abandonment rate. The share of callers who hang up before being served. It shows congestion and caller patience at the same time.
These figures come from ACD reports, usually in half-hour intervals. Call center metrics covers the wider set, including handle time and first contact resolution.
Staffing an inbound queue with Erlang C
Common practice, the review says, is the Erlang C queueing model. It estimates the chance a caller waits, given calls per interval, average handle time and the number of agents.
The review lists its assumptions. Calls arrive at random, handle times vary at random, callers are served in order, and busy signals, abandonment and retrials are ignored.
As a worked example, take 100 calls in a half hour with a three-minute handle time. The load is 10 Erlangs (100 × 180 seconds ÷ 1,800 seconds). Answering 80% of calls within 20 seconds needs 14 agents; 13 agents reach about 79.6%.
Double the calls to 200 and the target needs 24 agents, ten more. Larger queues need proportionally fewer extra agents, which the review explains with square-root safety staffing.
That number is agents on the phones. Planners then gross it up for breaks, training, meetings and absence, which the review notes as a standard step.
Frequently asked questions
What is the difference between an inbound call center and a contact center?
A contact center handles other channels next to the phone, such as email and chat. The 2003 review describes the contact center as an extension of the call center into those media. Most inbound phone teams today sit inside a wider contact center.
What is a good service level for an inbound call center?
There is no single standard. Targets of 20 or 30 seconds are common, and the share of calls to meet them is a business choice. It helps to set the target per queue, since a card block and a billing question can tolerate different waits.
How many agents does an inbound call center need?
Planners usually start from forecast calls and handle time per half hour, then use Erlang C to find the agents needed for the target. They add a gross-up for time agents are scheduled but off the phones. Erlang C ignores callers who hang up, and the 2003 review recommends Erlang A, which includes them, as the better standard.
What is the difference between ASA and service level?
ASA is an average wait across answered calls. Service level is the share of calls answered within a set time. Two queues with the same ASA can have very different service levels, so many teams report both, plus abandonment.
How we approach this at Zowie
Zowie's AI agents work on voice, chat, and email, with over 100 million conversations a year across our clients. On a call, the language model speaks. A separate, deterministic Decision Engine returns every amount, rate, offer, disclosure, and escalation from your rules, with an audit log. To see it with your own call types, you are welcome to book a demo.
Explore: Outbound call center, Automatic call distributor (ACD), Workforce optimization (WFO), Call routing, Call center metrics, Conversational IVR