How many support agents do you actually need?
Most teams guess. Guessing high wastes payroll; guessing low means your agents drown and customers churn. Enter your real ticket volume and get a range you can defend in a budget meeting — plus an honest read on whether AI automation is worth it at your size.
No signup. No email required. Every assumption is editable, and the tool will tell you when automation is not worth buying.
Sets handle time and the realistic automation ceiling.
Order status, returns, shipping, payment issues
Time an agent actually spends per ticket, including wrap-up.
Wages + benefits + supervision + seats. Not the bare hourly rate.
If you automated support
Capped at 62% for this industry.
How Intercom Fin, Gorgias and Zendesk actually bill — typically $0.55-$1.20.
Nobody cuts 20 agents the week automation ships. Savings accrue via attrition.
Agents you need today
2–3
$8,286 – $12,429 / month fully loaded
After automation
1
1 head removable · 1,500 tickets / mo still reach a human
- Platform cost
- $975/mo
- Net monthly
- +$3,168
- Payback
- 0.8 mo
- Year-1 return
- 2.8x
Why we reached that conclusion
- 0.8-month payback and 2.8x first-year return on a 975/month platform fee.
Not sure whether the paid report is worth $49 to you?
Leave the number instead of the invoice. A person will look up teams like yours — 3,000 tickets in ecommerce — and tell you straight whether it is. The public benchmark tables are already free, so this is not a way of buying something you can read.
Share this with your team
Support sizing: 3,000 tickets/mo at ~8min avg handle time needs ~2-3 agents ($8,286-$12,429/mo fully loaded). Automation: 0.8mo payback, 2.8x year-1 return, $3,168/mo net.
The 90-day plan this belongs in
The free tool gives you a number. This gives you the plan around it: volume sensitivity, your break-even point, what to do in the next 90 days, and the four things that would change the answer. Every figure comes from the same model — it is not a different calculator with nicer prose.
Section 3 of your report — 3. Sensitivity — the part that decides the deal
Every table in every vendor deck assumes volume is flat. Yours will not be. The same plan at three volumes:
| Scenario | Headcount | Net / month | Verdict |
|---|---|---|---|
| Volume -30% | 2-2 | +$3,460 | strong |
| Volume flat | 2-3 | +$3,168 | strong |
| Volume +50% | 3-4 | +$2,680 | strong |
Generated live from your inputs above. If you change the numbers, this changes with them.
- • Current state, with the full derivation
- • What automation changes, line by line
- • Sensitivity at -30% / flat / +50% volume
- • Your break-even ticket volume
- • 90-day sequenced plan
- • Risks and reversal triggers
- • PDF + shareable permalink
- • Everything in the full report
- • 30 minutes with someone who has sized support orgs
- • We pressure-test your assumptions
- • Written follow-up you can forward to your CFO
The review is the product. More PDF would just be padding.
Current volumes sell out of relevance fast, so the report is generated for the numbers you enter rather than being a generic template. If it is wrong, it is wrong because your inputs are wrong — every constant is published on the methodology page so you can check it.
How the numbers are calculated
A calculator that hides its assumptions is a sales brochure. These are every constant we use.
Productive time
5.5 productive hours per 8-hour shift, after breaks, meetings, training and idle time. 4.6 shifts per agent to cover 7 days a week. Using 8 productive hours is the most common way these estimates come out 40% too low.
Handle-time range
We report headcount across your industry's 25th-to-75th percentile handle time — 7 to 12 minutes for e-commerce, 9 to 18 for SaaS. Real teams span this range, which is why a single-number estimate is never honest.
Automation ceiling
Capped at 62% for e-commerce, 55% for SaaS, 50% for marketplaces. If you enter a higher number we use the ceiling instead and tell you why. Pilot-period coverage that excludes escalations is the single most common way automation projects disappoint.
When not to buy support automation
- Under ~400 tickets a month. A decent help centre and one accountable human will beat a platform on both cost and quality.
- Before you have written anything down. If your FAQs live in three people's heads, automation has nothing to automate.
- While support is a strategic differentiator. Automated deflection saves money and quietly costs you the relationship.
Work in support or CX consulting?
If you put headcount sizing into client proposals, embed it directly. Same numbers as this page, no cookies, no email capture, safe inside a client-facing document. The attribution link stays — it's what keeps this free.
<iframe src="https://app.highkingflower.com/embed" width="100%" height="620" frameborder="0" title="Support headcount and automation ROI calculator" loading="lazy"></iframe>
Every constant behind it is public on the methodology page, so you can defend the number if a client challenges it. Happy to white-label the colours.
Frequently asked questions
- How do you calculate support headcount?
- Divide daily tickets by the number one agent can handle per shift, then multiply by the shifts needed to cover your service hours. We use 5.5 productive hours per 8-hour shift after breaks, meetings, training and idle time, and 4.6 shifts per full-time agent to cover 7 days.
- What is a realistic automation coverage number?
- For e-commerce support, 50-62% of tickets are typically repetitive enough to resolve without a human. Higher claims usually come from a pilot period that excluded escalations. We cap the model at the realistic ceiling for your industry rather than letting you enter 90% and get a fantasy number.
- What is a good tickets-per-agent benchmark?
- It depends almost entirely on handle time. At 7 minutes average handle time an agent clears about 47 tickets a shift. At 18 minutes, about 18. This is why the calculator reports a range across your industry's handle-time distribution instead of a single number.
- Is AI customer support worth it for a small team?
- Below roughly 400 tickets a month, usually not. A good help centre plus a founder answering tickets is cheaper. The calculator says this directly, and we would rather show you the honest answer than sell you a platform that does not pay back.
- What counts as a fully loaded agent cost?
- Wages, benefits, employer taxes, supervision overhead, seat licences, and recruiting and training amortised per agent. Using the bare hourly wage is the most common way teams understate support cost and then cancel their automation project in month three.
- How much does one support ticket actually cost?
- At a loaded cost of $26 per agent hour, a 7-minute e-commerce ticket costs about $4.43 and an 18-minute SaaS ticket about $11.56. The number that matters is cost per ticket multiplied by tickets per resolved problem, which is where automation pays off by cutting repeat contacts.
- Does this calculator need me to sign up?
- No. There is no account, no login, and your inputs are computed in your own browser and never sent to us. Only numbers are logged, so that we can tell which volume band and industry is most useful to build for.