How Many Support Agents Do You Need? A Simple Staffing Model

“How many agents do we need?” is one of the most common questions support leads face, and one of the easiest to answer badly. Guessing from gut feel leads either to an overworked team and slipping response times, or to a budget request nobody can justify.

You don’t need specialist workforce-management software to get a defensible estimate. This guide walks through a simple workload-based model using numbers most help desks already report, then explains how to adjust it for real-time channels and coverage hours.

The Core Idea: Workload Divided by Capacity

Staffing comes down to two quantities:

Workload — the total agent time your incoming requests require.
Capacity — the productive time one agent actually has available to handle those requests.

Divide the first by the second and you get the number of agents needed to keep up. Everything else is about measuring those two quantities honestly.

Step 1: Measure Your Ticket Volume

Pull the number of tickets created per month for at least the last three to six months from your help desk’s reporting. Look for:

The average month, as a baseline.
The busiest month, since you need to survive peaks, not just averages.
Growth trend, if volume is rising along with your customer base.

If you’ve recently launched a self-service help center or changed your product significantly, weight recent months more heavily — older data may no longer represent your volume.

Step 2: Measure Average Handle Time

Handle time is how long an agent spends actively working on a ticket across all replies — not the time the ticket sits open waiting on the customer. Many help desks track this automatically; if yours doesn’t, have agents log time on a sample of tickets for a couple of weeks.

Handle time usually varies a lot by ticket type. A password reset might take a few minutes, a technical investigation an hour. If your mix is uneven, calculate handle time per category and weight it by volume rather than relying on one average. A consistent tagging system makes this much easier.

Step 3: Calculate Productive Hours per Agent

This is where most estimates go wrong. An agent scheduled for 40 hours a week doesn’t spend 40 hours answering tickets. Subtract time for:

Team meetings, training, and one-on-ones.
Breaks, holidays, and sick leave.
Non-ticket work: writing help articles, QA reviews, product feedback, internal projects.

Then apply a utilization target — the share of remaining time you expect agents to spend actively working tickets. Planning for 100% utilization leaves no slack for complex tickets, interruptions, or a sudden spike, and tends to burn people out. Many teams plan well below that; choose a figure that reflects how your team actually works and revisit it once you have real data.

Step 4: Do the Math

The basic formula:

Agents needed = (Monthly tickets × Average handle time) ÷ Productive hours per agent per month

A worked example

The numbers below are illustrative, not benchmarks — substitute your own.

A team receives 3,000 tickets per month.
Average handle time is 12 minutes (0.2 hours) per ticket.
Workload = 3,000 × 0.2 = 600 agent-hours per month.

For capacity, assume each agent is scheduled for about 160 hours a month. After meetings, training, leave, and non-ticket work, roughly 120 hours remain. At a planned utilization of 75%, each agent contributes 90 productive ticket-handling hours per month.

Agents needed = 600 ÷ 90 ≈ 6.7, so plan for 7 agents.

If the busiest month in the last year reached 3,600 tickets, the same math gives 720 ÷ 90 = 8 agents to absorb peaks without sacrificing response times — or 7 agents plus a plan for temporary help during peak periods.

Step 5: Adjust for Coverage Hours

The workload model tells you how many agent-hours you need, not when you need them. If you offer support across multiple shifts, time zones, or weekends, you may need more people than the raw number suggests, because each covered hour needs at least one agent present even when volume is low.

Check coverage separately:

List the hours you promise support (for example, 8 a.m.–8 p.m. on weekdays).
Determine the minimum number of agents who must be on shift during each block — at least one, often two so nobody is ever alone with an urgent issue.
Total the shift hours required and compare them with your workload estimate. Staff to whichever is higher.

Small teams often find coverage, not volume, is the binding constraint. A team with low volume but a 24/7 promise needs enough people to fill every shift regardless of how few tickets arrive overnight.

Step 6: Adjust for Real-Time Channels

The model above works well for email and other asynchronous channels, where tickets can wait in a queue for a while. Live chat and phone are different: customers expect an answer within seconds or minutes, so you need agents available at the moment demand arrives, not just enough total hours across the month.

For real-time channels:

Look at volume by hour of day, not just per month. Most help desks can show arrivals by hour.
Plan staffing around peak hours, accepting that agents will be less busy during quiet periods.
Consider concurrency for chat. Agents can often handle more than one chat at once, but the practical limit depends on conversation complexity.

Contact centers with heavy phone volume typically use queueing models such as Erlang C to calculate staffing for a target wait time. For most small and mid-sized teams, the hourly-volume approach is a reasonable approximation until real-time volume becomes large.

Step 7: Factor In Your Service Targets

Your staffing level and your SLA targets are directly linked. The model above estimates the staff needed to keep up with volume over a month — but keeping up isn’t the same as responding within an hour. Tight first-response targets require more agents available at any given moment than loose ones.

If you’re consistently missing first-response targets while resolution quality is fine, that usually points to a staffing or scheduling gap rather than a performance problem.

Ways to Reduce the Number You Need

Before hiring to match rising volume, check whether the volume itself can come down:

Deflect repetitive questions with a better knowledge base — see what a knowledge base is and how it reduces tickets.
Lower handle time with well-written canned responses and macros and automation for routine actions.
Fix root causes. If a single confusing feature generates a large share of tickets, a product fix may be cheaper than another hire.
Route tickets better, so specialists aren’t spending time on tickets anyone could answer — covered in our guide to automating ticket routing.

FAQ

How many tickets should one support agent handle per day?

There’s no universal number — it depends on handle time, which varies enormously between a quick billing question and a complex technical issue. Rather than setting a ticket-count target, calculate capacity from your own handle time and productive hours. Ticket-count quotas tend to encourage rushing through easy tickets and avoiding hard ones.

When should a support team hire its next agent?

Watch the trend rather than waiting for a crisis. If first-response times are rising, backlog is growing week over week, or your forecast shows workload exceeding capacity within the time it takes to hire and train someone, start hiring then. Our guide to reducing ticket backlog covers early warning signs.

Should I staff for average volume or peak volume?

Usually somewhere between the two. Staffing for the absolute peak leaves people idle most of the year; staffing for the average means every busy period becomes a backlog. Many teams staff near their typical busy level and cover true peaks with overtime, temporary help, or deliberately looser targets during known spikes.

Does AI reduce how many support agents you need?

It can reduce workload for well-documented, repetitive requests, but the effect depends heavily on your ticket mix. Model it honestly: estimate the share of tickets AI could realistically resolve end-to-end, subtract only that portion from your workload, and remember that the remaining tickets are often the harder ones. Our guide to AI customer support covers what AI can and can’t reliably handle.