Erlang C Calculator: How Many Customer Service Agents Do You Actually Need?

If you have ever searched for an erlang calculator, you probably landed on a page built for someone managing a 200-seat call center. The interface asks for "traffic intensity in Erlangs," "service level targets," and "average handle time" — terms that mean nothing when you are a Shopify merchant trying to figure out whether you need one hire or two. The Erlang C formula is genuinely useful, but almost every resource about it assumes you already know what it is and that you operate at enterprise scale.

You do not. You run a small ecommerce store, and you need a straight answer: given your order volume and the number of customer questions that come in, how many people should be handling support? That is exactly what the Erlang C model tells you — once someone strips away the jargon. Try our free staffing calculator to skip the math entirely and get your number in seconds.

What Is the Erlang C Formula (and Why Should You Care)?

A Danish mathematician named Agner Krarup Erlang published his formula in 1917 while working for the Copenhagen Telephone Company. He was solving a specific problem: how many telephone operators does a switchboard need so that callers do not wait too long? The math he developed — now called the Erlang C formula — calculates the probability that an incoming request has to wait in a queue before someone handles it.

Over a century later, that same formula is the backbone of workforce management in every major contact center platform. NICE, Calabrio, and Assembled all use Erlang C under the hood when they tell managers how many agents to schedule per shift.

The reason you should care: the formula works just as well for a five-person ecommerce support team as it does for a 500-seat call center. It takes your ticket volume, the average time each ticket takes to resolve, and your target response time, then outputs the minimum number of people you need on duty.

The problem is that nobody explains it that way. They explain it in Erlangs, Poisson distributions, and service-level percentages. So here is the plain-English version.

How the Erlang C Calculator Works — Plain English Version

Forget the math notation. The Erlang C formula answers one question: if X tickets arrive per hour and each one takes Y minutes to handle, how many agents do I need so that Z% of tickets get a response within W minutes?

Four inputs. That is it.

1. Tickets per hour. How many customer inquiries arrive during a typical hour. Not per day — per hour. If you get 40 tickets a day across an 8-hour window, that is 5 per hour.

2. Average handle time. How long it takes to fully resolve a ticket, from first response to close. For email-based ecommerce support, this is usually 8-15 minutes of active work per ticket. For live chat, 5-10 minutes.

3. Target response time. How fast you want to respond to most tickets. A common benchmark is 80% of tickets answered within 60 seconds for chat, or within 30 minutes for email.

4. Service level percentage. What fraction of tickets should hit that target. The industry standard is 80%, sometimes written as "80/20" (80% of calls answered within 20 seconds) in call center shorthand.

The formula takes those four numbers, runs them through a probability model based on random arrival patterns, and outputs the number of agents needed. It accounts for the fact that tickets do not arrive in a neat, even stream — they cluster. Three come in at once, then nothing for 20 minutes, then five in a row. That clustering is precisely what makes staffing hard, and what the Erlang C model was built to handle.

You can try our free staffing calculator to run these numbers for your own store without touching the underlying math.

How to Use the Erlang C Calculator for Ecommerce Support

Most Erlang C content uses phone call examples, but ecommerce support is increasingly chat, email, and social media. The formula still applies — you just need to adjust your inputs.

Step 1: Find your real ticket volume. Pull the numbers from your helpdesk or inbox. If you do not have a helpdesk, count the emails, chat messages, and DMs you handled last week and divide by the number of hours you were actively working support. Smaller ecommerce businesses generate roughly 88 support tickets for every 100 orders, while larger operations see closer to 56. If you ship 500 orders a month, expect around 440 tickets.

Step 2: Estimate your handle time. Time yourself over a few days. Most small ecommerce stores spend 8-12 minutes of active work per ticket, but this varies. "Where is my order?" takes two minutes. A complicated return with a damaged item and a frustrated customer takes 20. Average them out.

Step 3: Decide your target response time. This depends on your channels. If you offer live chat, your target should be under two minutes. For email, aiming for a response within one hour puts you ahead of most competitors — the industry average for ecommerce email responses is 12 hours. Pick a target that is realistic for your current setup but fast enough to keep customers from leaving.

Step 4: Plug the numbers in. Use our staffing calculator to run the Erlang C math. It will tell you how many agents need to be actively working support during each hour to hit your service target.

Step 5: Add shrinkage. This is the workforce management term for all the time your staff is not actually handling tickets — breaks, lunch, meetings, training, bathroom visits. The standard shrinkage factor is 30-35%. If the calculator says you need 2 agents, divide by 0.65 to get the real headcount: about 3 people scheduled to keep 2 active.

What the Numbers Usually Tell Small Business Owners

Here is what happens when most small ecommerce stores run their numbers through an Erlang C calculator for the first time: the result is smaller than they expected.

Take a store doing 1,000 orders per month. Using the 88-tickets-per-100-orders benchmark, that is roughly 880 tickets a month, or about 40 per business day. Spread across 8 hours, that is 5 tickets per hour with an average handle time of 10 minutes.

Run that through Erlang C with a target of 80% of tickets responded to within 30 minutes, and the formula says you need 1 agent. Maybe 2 during peak hours.

A store doing 3,000 orders per month — roughly 2,640 tickets, or about 120 per day — needs 2-3 agents during business hours to maintain the same service level.

The math consistently shows that small ecommerce stores need fewer agents than their gut tells them. The real issue is not headcount — it is coverage. You need someone available during all your support hours, and you need a plan for spikes (holiday weekends, a product going viral, a shipping carrier having a bad day). That is where most small stores break down. Not because they lack people, but because their one or two people cannot be available 12 hours a day, seven days a week.

When coverage gaps appear, response times collapse and customer churn follows. Fixing that gap matters more than raw headcount for stores your size.

When Erlang C Breaks Down: The AI Variable

The Erlang C formula was built in 1917 for telephone switchboards. It has no concept of a support agent that never sleeps, never takes a break, and handles 10 conversations simultaneously. AI changes the staffing equation in a way Erlang never anticipated.

Here is why the formula's assumptions start to crack for modern ecommerce support:

Erlang C assumes every ticket needs a human. It does not. Where is my order? Return policy questions. Shipping time inquiries. Store hours. Product availability. These repetitive, high-volume tickets make up the majority of most ecommerce support queues. Companies using AI for customer service now report 50-70% of tickets resolved without human intervention, and Gartner predicts that number will reach 80% by 2029.

Erlang C assumes a fixed handle time. AI resolves routine tickets in seconds, not minutes. That collapses the average handle time input and changes the output dramatically.

Erlang C assumes customers wait in a queue. An AI agent responds instantly. There is no queue for the tickets it handles, which means the queuing model the formula is built on simply does not apply to that portion of your volume.

What this means in practice: if AI handles 60% of your tickets, you only need to staff for the remaining 40%. Go back to that 1,000-order-per-month store. Instead of 880 tickets needing human attention, the number drops to roughly 350. That is 16 per day, or 2 per hour. The Erlang C math now says you need one person working part-time — and only for the complex stuff that actually requires human judgment, like handling a frustrated customer complaint or negotiating a tricky refund request.

Ernest is built for exactly this scenario. It handles the routine 60-70% — order tracking, policy questions, product inquiries, basic troubleshooting — so you only staff humans for the 30-40% that requires real judgment. Instead of hiring a full-time agent at $38,000-$50,000 per year, you pay a fraction of that and get 24/7 coverage on top of it.

The staffing equation flips from "how many agents do I need?" to "how few hours of human support do I actually need to cover what AI cannot?"

Try the Free Staffing Calculator

We built a free staffing calculator at heyernest.ai/tools/staffing-calculator that runs the Erlang C math for you. Plug in your monthly order volume, average handle time, and target response time. It shows you two numbers side by side: how many agents you need with humans handling everything, and how many you need when AI covers the routine tickets.

For most small stores, the difference between those two numbers is one full-time hire. That is $38K-$50K per year in salary alone, before you factor in training, benefits, scheduling headaches, and the fact that human agents do not work nights and weekends unless you pay extra.

The calculator does not ask you to know what an "Erlang" is. It does not require a background in queuing theory. You enter the numbers you already know about your business, and it tells you what you need.

Stop Guessing, Start Calculating

Most small store owners either overstaff support (hiring a full-time agent when they need 15 hours a week of human coverage) or understaff it (trying to do everything themselves until response times crater and reviews suffer). Both are expensive mistakes — one costs you money directly, the other costs you customers.

The Erlang C formula exists to replace guessing with math. It is not perfect — it was not designed for AI-assisted support, it assumes customers never abandon the queue, and it works best when your volume is high enough for statistical patterns to emerge. But even with its limitations, running your numbers through the model gives you a starting point grounded in data instead of anxiety.

Run your numbers through our free staffing calculator. See what the math says. Then look at what AI can take off the human pile — the order tracking questions, the policy lookups, the "do you ship to Canada" messages at 2am.

For most stores, the answer is clear: you do not need more agents. You need an AI agent handling the predictable volume and a human covering the rest. Ernest starts free for up to 50 conversations a month — enough to see the impact before you commit. If you are building your support operation from scratch, our customer service guide for small businesses covers the full framework, and our software comparison helps you pick the right tools for each stage of growth.

Your customers are already sending you tickets. The only question is whether you are staffing based on math or on a hunch.