テレフォニー
Erlang C
別名: Erlang C formula
Erlang C is a queuing formula that calculates the probability an arriving call has to wait, and from that the agents required to answer a given volume within a target time.
It models a queue where calls arrive at random and wait until an agent is free. Feed it offered load and a service-level target and it returns the minimum agents needed, along with the service level, average speed of answer and occupancy any given headcount would produce.
Erlang C rests on four assumptions, and all four break in real operations: calls arrive at random (Poisson), no caller ever abandons, every agent handles every call identically, and conditions hold steady across the interval. The no-abandonment assumption is the consequential one — where callers do hang up, Erlang C over-staffs, and Erlang A exists to correct for it.
Erlang C is distinct from Erlang B, which models systems with no queue at all — a blocked call is simply lost. Erlang B is the right model for trunk sizing; Erlang C is the right model for agent staffing. In practice Erlang C is computed via the Erlang B recurrence, which stays numerically stable at headcounts where the textbook factorial form overflows.
よくある質問
What is the difference between Erlang B and Erlang C?
Erlang B models systems with no queue — a call that finds every line busy is lost. Erlang C models systems where calls queue and wait. Use Erlang B for trunk and line sizing, Erlang C for agent staffing.
When should you not use Erlang C?
When abandonment is high, because Erlang C assumes nobody hangs up and will over-staff as a result. Also when routing is skills-based, since a single pooled calculation understates what each skill group needs, and when arrivals spike sharply inside the interval you are modelling.
関連用語
Erlang Cを実際の通話で見る。
30分のデモを予約して、Finnがインバウンドとアウトバウンドの通話を最初から最後まで処理する様子をご覧ください——組み立てるべきスタックは不要です。