Customer support budget: the four lines that matter

Most small businesses never sit down and write a customer support budget. The money leaks out in pieces instead. A $39 software bill on the first of the month. Four hours of your Saturday answering the same shipping question. A $30 refund handed over to make an angry email go away. Nobody adds it up, so nobody can say whether the total is too high, too low, or about right. Here's how to add it up in an afternoon, using four line items and a worked example for a shop handling about 120 questions a month.

What actually goes in a customer support budget

Four things. Software, time, the cost of whatever people are writing in about, and a buffer for the bad months. Most of the budgets we see only count the first one, which is why they tend to be wrong by a wide margin.

1. Software

This is the line everyone starts with, and it is usually the smallest. A shared inbox, ticketing, live chat and a knowledge base, either bundled into one tool or bought as separate subscriptions.

Two things move this line. The first is seats, since most help desks charge per person per month, so hiring one part-timer can double the bill overnight. The second is metering. A lot of tools now bill automated answering by the resolved conversation, which means your bill goes up every time you write a good help article. We covered that trap in more detail in our post on the hidden costs of help desk software.

Verified August 2026, entry paid plans across the popular tools run from roughly $7 to $25 per seat per month, with automated answering often priced on top. Check each vendor's own site before you budget, because these change often. SupportifyGPT charges a flat $15, $39 or $79 a month for 1, 3 or 10 seats, and every feature including the support bot is on every plan at no extra charge. Our pricing page lays out what sits in each tier.

One more thing worth writing into the software line: the billing term. Annual plans usually save 15% to 20%, which is real money on a small budget, but they lock you in before you know whether the tool suits you. Run the trial monthly, give it a full busy week, and switch to annual once you are sure. Paying monthly for two extra months is cheaper than being stuck for ten.

2. Time

Here's the big one. It gets left out because nobody sends you an invoice for it.

Count it anyway. Take a normal month, count the conversations, and multiply by how long an average reply takes including the reading and the looking things up. Six minutes is a fair opening guess for email. Live chat runs longer, because someone is waiting on you while you type.

If you have never tracked conversations, you do not need a tool to start. Sort your support mailbox by date, count the threads for one ordinary week, and multiply by four. Skip a holiday week and skip the week you launched something. Then time yourself honestly on ten replies, including the part where you go and check the order, and take the average. Two hours of counting gives you a number that is close enough to budget against, and you can refine it later once a help desk is doing the counting for you.

Then put an hourly rate on those minutes. Federal wage data puts the median hourly wage for customer service representatives at $20.59 as of May 2024, which is a sensible floor if you would be hiring. If you are doing the answering yourself, use what an hour of your own time is worth doing the next best thing, which for most owners is a good deal more than $20.

3. The cost of what people write in about

Refunds, replacement units, reshipping, goodwill credits, discount codes handed out to calm someone down. This money usually sits in a different account, so it never shows up in support planning at all.

It belongs here, because support volume drives it. It is also the line with the most give in it. Fix the packaging that keeps arriving dented and both the refunds and the emails about them disappear together. That is the real argument for spending an afternoon on cutting ticket volume at the source rather than trying to type faster.

4. A buffer

Support is lumpy. A December rush, one bad batch from a supplier, a price change, an outage on a Tuesday. Add 10% to 20% on top of the first three lines so a busy month is a busy month and not an emergency.

Size the buffer from your own history rather than a rule of thumb. Find your busiest month last year, work out how much more volume it carried than an average month, and use that. A shop whose December runs 40% above normal needs a bigger cushion than a service business that ticks along evenly all year.

Why percentage benchmarks break for a small business support budget

Go looking for a support spending benchmark and you get percentages of revenue. Somewhere between 2% and 8%, depending on whose table you read.

Those tables are built from software companies with recurring revenue in the millions and a support team on payroll. Run one against a $400,000 business and a 5% benchmark says you should spend $20,000 a year answering customer questions, which for most small businesses is a fantasy. The same business at half the ticket volume would want a completely different figure, and the benchmark cannot tell the two apart.

Benchmarks are fine as a sanity check after you have done the work. They are a bad place to start. Build a customer support budget from the bottom up, then look at what percentage you landed on and decide if it feels right.

A worked example you can copy

Say you run a small online shop. $400,000 in revenue last year, you plus one part-timer, and roughly 120 customer questions a month across email and chat. Here is what a customer support budget looks like for that business.

• Software: one tool covering inbox, chat and a knowledge base at $39 a month.
• Time: 120 conversations at 6 minutes each is 12 hours a month. At $20.59 an hour, that is about $247.
• What people write in about: around 10 of those 120 end in a refund or credit averaging $25, so $240.
• Buffer: 15% on the $526 subtotal, or about $79.

Total: roughly $605 a month, near enough $7,260 a year, which is 1.8% of revenue. It also works out to about $5 per conversation, and that is a figure you can watch month to month. Our post on cost per ticket covers what to do with it once you have it.

Now change one assumption. If the owner handles support personally and an hour of their time is worth $60, those same 12 hours cost $720 a month, the total lands closer to $1,270, and the software bill turns into a rounding error. That is the whole reason to write the budget down. It shows you which line is worth attacking, and it is rarely the one you were staring at.

What the same budget looks like at three sizes

Run the same four lines at different volumes and the picture gets clearer. These use the same six-minute handling time, the same $20.59 hourly rate, the same $25 average refund on roughly 8% of conversations, and the same 15% buffer.

• Solo, about 30 questions a month. Software $15, time 3 hours at about $62, refunds around $60. With the buffer, close to $158 a month.
• Two people, about 120 questions a month. The example above. Software $39, time $247, refunds $240. With the buffer, about $605 a month.
• Four people, about 400 questions a month. Software $79, time 40 hours at about $824, refunds around $825. With the buffer, roughly $1,990 a month.

Notice what stays still. Per conversation, all three land within a few cents of $5. That is not a happy accident, and we are not claiming it as a law. We held the handling time and the refund rate steady on purpose, and that is exactly the finding: your total support cost tracks your conversation count almost perfectly. Everything else is noise around it.

Which means the only lever that changes the shape of the budget rather than just its size is volume. Halve the questions and you halve the bill, no matter which tool you bought.

What to cut first when the number is too big

If your customer support budget comes out bigger than you can stand, cut in this order. Each step is worth more than the one after it.

1. Cut the reasons. Read your last 100 conversations and count the top five questions. In most small businesses the top three cover about half the volume, and they are usually about delivery, returns or something the website does not say clearly. Fix the page. Change the confirmation email. Write the help article. A question you prevent costs nothing forever.
2. Cut the minutes. Saved replies for anything you answer weekly. A published help center the support bot can quote from, so a routine question gets a real answer at 11pm without waking anybody. Neither is glamorous and both take a third or more off the average handling time.
3. Cut the tools last. Software is normally the smallest line, and a cheaper tool that makes every reply slower costs you more in time than it saves in fees. The exception is duplicate spend. If you pay for a chat tool, a separate ticketing tool and a separate knowledge base, folding all three into one product is a real saving.

How to tell your budget is set wrong

There are two ways to get this wrong, and both are easy to spot once you are tracking the number.

Set too low looks like replies going out after midnight because that is the only hour left, first-reply times creeping past a day, refunds climbing because problems sat too long, and you personally dreading the inbox. Underfunded support does not announce itself as a budget problem. It shows up months later as customers who quietly stopped buying, which our post on reducing customer churn gets into.

Set too high looks like per-seat fees for people who answer two tickets a week, three overlapping subscriptions nobody has audited, or a plan tier bought in case you grew into it. Compare your seat count against who actually logged in last month. That one check pays for itself more often than it should.

There is a third case that looks like neither. The budget is fine on paper and the work still is not getting done, because the hours exist but they belong to someone whose job is something else. When the time line passes roughly 20 hours a month, you are paying a person's part-time wage in owner hours and getting no product work in exchange. That is the moment to price a hire properly, which our post on when to hire customer support walks through.

Rerun the four lines of your customer support budget once a quarter. It takes about twenty minutes after the first time, and it is the difference between guessing at what support costs you and knowing. If you want a real number for the software line without doing per-seat arithmetic, you can start a 14-day trial and fill in the first row today.