Find out exactly how many agents you need to hit your service level target. Enter your contact volume and handle time, and the industry-standard Erlang C formula does the rest.
We've pre-filled a typical example. Adjust to match your center.
Calls, chats, or tickets arriving.
Use your busiest interval for safe staffing.
Talk time + hold + after-call work. 240s = 4 min.
The "20" in an 80/20 service level.
The "80" in an 80/20 service level.
Breaks, training, absence. Typical 25-35%.
Cap to avoid agent burnout. Above 85-90% is unsustainable.
Agents to schedule (with 30% shrinkage)
10 agents live on contacts, before shrinkage
Erlang C is the math behind nearly every contact center staffing decision. Created by Danish engineer A.K. Erlang in 1917 to model telephone traffic, it answers a deceptively tricky question: given a certain number of contacts arriving and how long each takes to handle, how many agents do you need so that most people are answered quickly rather than stuck in a queue?
The reason you can't just divide volume by capacity is randomness. Contacts don't arrive in a neat, even stream. They cluster and they go quiet, and those bursts are what create queues. Erlang C models that randomness to tell you the probability a contact has to wait, and from there it derives your service level, average speed of answer, and how busy your agents will be.
It runs in a few steps:
Worked example: 100 calls in an hour at 240 seconds each gives (100 × 240) ÷ 3600 = 6.67 Erlangs. You need at least 7 agents just to keep up, but to answer 80% of calls within 20 seconds you'll need around 10 on the phones. Apply 30% shrinkage and you need about 14 people scheduled (10 ÷ 0.70). The calculator above runs this Erlang C math live as you type.
Erlang C is the trusted industry standard, but it assumes contacts arrive randomly (a Poisson process), handle times follow an exponential distribution, and nobody abandons the queue. Because it ignores abandonment, it tends to slightly overestimate the agents you need, which errs on the safe side for planning. It is built for voice and works well for any synchronous channel like live chat (where one agent often handles a few concurrent sessions). It is less suited to asynchronous email or ticket queues, which you may prefer to plan with our cost per ticket calculator.
Erlang C makes one thing obvious: agents are your biggest cost lever, and the more contacts that reach the queue, the more people you have to schedule. So the most powerful way to ease staffing pressure is to reduce how many contacts hit the queue in the first place.
That's where Chatling helps. It's a no-code AI support agent that resolves repetitive questions automatically across your website, WhatsApp, and Instagram, around the clock, so a large share of contacts never need a human. Lower the volume reaching your team and the agent count this calculator returns drops with it. When a conversation does need a person, Chatling hands off with full context. Curious what that deflection is worth? Our AI support ROI calculator puts a number on it.
Let an AI agent resolve repetitive questions automatically so fewer contacts reach your queue. Hand off to humans when it matters. No coding required.
Find out what a single support ticket really costs your team, and how much an AI agent could save you.
See the return on automating customer support with AI: savings, payback, and your new cost per ticket.
Measure what share of support tickets you resolve without an agent, and how much that deflection saves you.
Calculate your Net Promoter Score from promoters, passives, and detractors, with instant benchmarks.
Work out your Customer Satisfaction Score from survey responses in seconds.
Create a free click-to-chat WhatsApp link with a pre-filled message, QR code, and ready-to-use button.