What is an Erlang?

An Erlang is the standard unit of telecommunications traffic, equal to one full hour of continuous channel occupancy in one hour of time. Named after Danish engineer A. K. Erlang, it is the foundational measurement that contact centers, trunk planners, and network engineers use to translate offered traffic into the number of lines or agents required to carry it.

How an Erlang is calculated

The simplest definition:

Traffic in Erlangs = (Number of calls × Average call duration in seconds) ÷ 3600

So 30 calls of 6 minutes each in one hour: (30 × 360) ÷ 3600 = 3 Erlangs

3 Erlangs of traffic means an average of 3 channels are in use at any moment during that hour. The number does not tell you how many channels you need — only the average load. Sizing depends on which Erlang formula you apply.

Erlang B vs. Erlang C

The two formulas the industry uses every day, with very different assumptions:

• Erlang B assumes blocked calls are lost — the caller gets a busy tone and goes away. Used for trunk sizing where overflow is not an option (PRI lines, SIP trunking capacity).
• Erlang C assumes blocked calls are queued — the caller waits until an agent is free. Used for contact-center staffing where calls join a queue rather than dropping.

The same traffic in Erlangs returns very different channel or agent counts depending on which formula matches the operating reality. Misapplying Erlang B to a contact center systematically understaffs it.

How contact centers use Erlang C

Erlang C powers nearly every workforce-management tool on the market. Inputs:

• Traffic in Erlangs for each interval (typically 15- or 30-minute slots).
• Target service level — the share of calls to answer within a fixed wait (commonly 80% in 20 seconds).
• Average handle time — talk time plus after-call work.
• Patience or maximum wait that bounds the queue.

Output: the minimum number of agents required to hit the target. Schedule the result with a shrinkage allowance for breaks, training, and absence, and the operation is correctly staffed for the forecast.

Limits of Erlang C

• Assumes infinite patience — every caller waits as long as it takes. Real callers abandon, which makes Erlang C slightly conservative for staffing.
• Assumes steady-state traffic within each interval — bursty arrivals violate this and push real waits above the formula’s prediction.
• Does not model multi-skill routing natively — skills-based routing operations need simulation-based tools, not pure Erlang C.
• Ignores agent productivity variance — schedule adherence, occupancy ceilings, and break patterns must be layered on top.

Erlang frequently asked questions

What is one Erlang?

One Erlang equals one full hour of one telecommunications channel being continuously occupied during one hour of time. It is the unit of average traffic load and is the input to Erlang B and Erlang C, the two formulas used to size trunks and contact-center agents.

What is the difference between Erlang B and Erlang C?

Erlang B assumes blocked calls are lost (the caller gets a busy signal) and is used to size carrier trunks. Erlang C assumes blocked calls are queued and is used to size contact-center agents. The same offered traffic produces different headcount or trunk counts depending on which model fits.

Why do call centers use Erlang C?

Erlang C answers the practical question every operation faces: given this volume and average handle time, how many agents are needed to hit the service-level target? Every modern workforce-management tool uses Erlang C (or a simulation extension of it) as the core staffing engine.

What is the Erlang formula used for?

Erlang formulas convert offered traffic — calls per hour multiplied by average duration — into the number of channels or agents required at a given probability of blockage (Erlang B) or wait time (Erlang C). They are foundational to telecom-network sizing and contact-center workforce management.

See how DialPhone fits

DialPhone’s contact center reports interval-level traffic in Erlangs alongside service level, ASA, and occupancy — feeding directly into Erlang-C-based scheduling so staffing decisions are grounded in measured load, not estimates.