Help desk ROI: the math for a small business

Help desk ROI is a simple idea buried under bad examples. Nearly every guide you will find runs the same one. Ten agents. Twelve thousand tickets a month. A fully loaded labour rate of $35 an hour. A satisfying 320 percent at the bottom of the page. If you are one person answering forty emails a week, not one of those numbers is yours, and the answer they produce tells you nothing.

So here is the small business version. The numbers are smaller, the math is shorter, and it is much harder to fudge.

What help desk ROI actually measures

Help desk ROI is what you get back from the software, minus what it cost you, divided by what it cost you. Multiply by 100 and you have a percentage. Nothing about support software makes that arithmetic special.

What makes it slippery is the first half. For a fifty person support team, "what you get back" is labour, and labour is real cash on a payroll run. Cut four minutes off twelve thousand tickets and someone in finance can point at a line item that got smaller.

You do not have that line item. If a tool saves you five hours a week, nobody sends you a cheque. The saving is real, but it only turns into money when those five hours go somewhere that earns, or when they go back into your evening and you decide that is worth paying for. Both are legitimate. They are just different, and the vendor calculators quietly assume the first one.

That is the whole reason a small business needs its own version of this. The formula does not change. What changes is that you are the one who has to say honestly what a returned hour is worth, and nobody audits your answer.

Which is why we would rather you work out a payback date than a help desk ROI percentage. A percentage compares two numbers you invented. A payback date says "this covers itself by the second week of November," and you either see it happen or you do not.

The standard help desk ROI formula, and where it breaks

Every help desk ROI calculator on the internet runs the same three lines:

• Hours saved per month = tickets per month × minutes saved per ticket ÷ 60
• Value of that time = hours saved × your loaded hourly cost
• ROI = (annual value − annual cost) ÷ annual cost × 100

That is sound. Three things go wrong when you run it at small scale.

You do not lay yourself off

A big team converts saved minutes into fewer hires. You convert them into a shorter Tuesday. If you never name what the freed time is for, the calculation produces a number you do not believe, and you talk yourself out of a $15 subscription for no good reason.

Write it down before you start. "These four hours go into quoting jobs" or "these four hours go into not working Sundays." Then you can price them.

The hourly rate is borrowed from someone else

Calculators default to $30 or $35 an hour because that is roughly what a support agent costs a mid-sized company once you add benefits and overhead. For a sense of the underlying wage, the federal wage figures put the median for customer service representatives at $21.53 an hour as of May 2025, with the bottom tenth under $15.27.

If you employ someone, use their real cost. If it is you, that number is wrong in both directions. Your time is cheaper than $35 in the sense that nobody bills for it. And it is far more expensive, because the hour you spend retyping a returns policy is an hour you did not spend selling. Owners we talk to who price honestly usually land somewhere between $40 and $80.

The denominator is tiny, which changes the question

Enterprise help desk ROI is a real spreadsheet exercise because the software costs $40,000 a year and getting it wrong matters. At small business prices, the whole question collapses into something you can do in your head.

Take a $22 hour, which is close to that median wage. A $15 a month plan breaks even at 41 minutes saved a month. A $39 plan breaks even at one hour and 46 minutes. Not per week. Per month. If a tool cannot save you two hours in a month, the problem is not the price, and you should look hard at whether you needed software at all.

A help desk ROI calculation you can actually do

Four steps, about twenty minutes, and you need a timer. Two inputs decide the whole thing, so spend your effort on those and guess at the rest.

Step 1: count messages and time three of them

Open your inbox and count the customer messages from the last full week. Then handle three of them with a stopwatch running, and include everything. Hunting for the order. Scrolling back to find what you told them in March. Checking whether your colleague already replied. That hunting is where the time actually goes, and it is invisible until you time it.

Three messages is enough. You are not running a study, you are trying to find out whether the answer is closer to two minutes or closer to eight, because that single number swings a help desk ROI calculation further than anything else you will enter.

Step 2: price your hour

Use the real cost if you pay someone. Use your next best hour if it is you. Pick one number and keep it for the whole calculation.

Step 3: add the money you stop losing

This is the half the enterprise guides skip, and for a small business it is often bigger than the time saving. Count only things you can point at:

• The quote request that sat in a shared inbox for two days and went cold.
• The refund you gave because nobody answered for a week and the customer got angry enough to want one.
• The repeat order from someone who got a straight answer in ten minutes.

One saved $400 job a quarter beats every minute-shaving argument on this page. If you cannot name a real example from the last six months, put zero here and move on. Guessing at this line is how vendor calculators get to 758 percent.

A note on the third bullet, because it is the one people push back on. You cannot prove the ten minute reply caused the repeat order. Nobody can. So either leave it out and accept your help desk ROI is understated, or put a small number in and label it as a judgement call. What you should not do is bury a big guess in the middle of a total and present the total as a finding.

Step 4: do the division

Here is a worked example from a two person business, and the numbers are deliberately unimpressive.

Before. 55 customer messages a week, all in a shared Gmail account. Timed at about 6 minutes each once you count the hunting. That is 330 minutes, or 5 hours 30 minutes a week.

After 30 days. They moved to a help desk, wrote eight help articles over a weekend, and saved seven canned replies. About 12 of the 55 weekly messages now get answered by the help centre before anyone sees them. The remaining 43 take about 4 minutes each, because the history and the saved replies are right there. That is 172 minutes, or 2 hours 52 minutes a week.

Saved. 2 hours 38 minutes a week, call it 11.4 hours a month.

Value. At $22 an hour, about $251 a month, so roughly $3,012 a year. Knock off a month of ramp and call the first year $2,761.

Cost. A $39 plan is $468 a year. Setup took about four hours of writing and fiddling, which at the same rate is $88. First year cost: $556.

Help desk ROI = ($2,761 − $556) ÷ $556 × 100 = 396 percent.

Now the honest part. That 396 percent rests entirely on the six minutes and the four minutes. We guessed at both. Every help desk ROI calculator you will meet does the same thing, it just does not tell you. Which is why the payback number from earlier is the more useful one: this business needed to save 1 hour 46 minutes a month and saved 11 hours. You do not need three decimal places to see that it cleared.

Notice too how much of the return came from the eight articles and the seven canned replies, not from the software itself. The subscription bought the place to put them. The weekend of writing is what moved the number, and a business that skips that weekend gets a much worse result on exactly the same plan.

What actually moves help desk ROI at this size

Four things do nearly all the work, roughly in order:

1. Answers customers find without you. A published help centre with ten good articles takes a weekend and keeps paying every week after. Nothing else on this list comes close. If you only do one thing, do this one, and our guide to cutting your ticket volume covers which ten to write first.
2. Saved replies for the eight questions you retype. Everyone has eight. Writing them once turns a four minute answer into a twenty second one.
3. One place instead of three. Most of the six minutes in that example was not typing. It was looking. Email, chat and a phone note in three systems is the single most expensive habit in small business support.
4. A billing model that does not punish you for growing. Per agent pricing taxes hiring a weekend helper. Per resolution pricing taxes a busy month, and taxes it hardest on exactly the cheap repetitive questions automation handles best. We wrote about the costs that show up after you sign, because they land squarely in the denominator of this calculation.

On the support bot, since it belongs in the deflection line: rules search what you have already written, and then a model writes the reply from those articles in plain sentences rather than handing back a list of links. If your articles do not cover the question, it says so and hands the conversation over instead of inventing a policy you never wrote. Anything it drafts for your knowledge base is held for you to approve before it goes live. It is included on every SupportifyGPT plan with no per-answer fee, which keeps it out of the cost side of the sum entirely.

Why published help desk ROI figures are mostly fiction

You will see 320 percent, 758 percent, "150 to 400 percent in year one." Every one of those is an output, not a finding. Someone chose the inputs, and they chose them while trying to sell software.

Three moves show up again and again:

• Counting the same saving twice. One popular example lists "reduced manual routing," "automated response handling" and "improved first contact resolution" as separate benefits. They overlap heavily. Adding them is like counting the same twenty minutes three times.
• Enterprise labour rates on a small business page. The rate does most of the work in the formula, and it is always the least examined number on the page.
• No admin time in the costs. Someone has to write the articles, tidy the tags and fix the routing rule that misfires. That is real hours, and leaving them out inflates the result every time.

Here is a test that takes ten seconds. Change one input by twenty percent and see how far the answer moves. If shaving the minutes-saved figure from five to four drops a 320 percent return to 240, you are not looking at a measurement. You are looking at a guess in a suit. Publish your assumptions next to your answer or the answer means nothing.

None of this makes help desk ROI a useless idea. The arithmetic is honest work when the inputs are yours and you can see them. What is worthless is a stranger's percentage, calculated from a stranger's ticket volume, on a page whose job is to sell you something.

When the honest answer is that help desk ROI does not clear

We sell this software and we would still tell you to skip it in these cases.

Under about ten customer messages a week, and only you answering. A tidy Gmail inbox with three labels is fine. Come back when a second person starts replying, because that is when duplicate answers and lost threads begin, and that is a real cost you can feel.

Every question is genuinely one of a kind. Custom fabrication, consulting, bespoke work. Self-service saves you nothing if no answer is ever reused, and self-service is most of the return. You might still want the shared inbox for the history. Just do not expect deflection to pay for it.

You need to act on an order from inside the reply. Issuing refunds or editing a shipment without leaving the conversation is a real time saving we do not offer, because we have no native Shopify or WooCommerce order sync. If that is most of your day, a purpose-built store desk will earn more than we will, and our honest comparison with Gorgias lays out where the line falls.

Do the twenty minute version

Count last week's messages. Time three of them. Pick an honest number for your hour. Multiply, then compare against what the software costs. That is the entire help desk ROI exercise, and it beats any calculator with a vendor logo in the corner because every input in it is yours. If you want a cleaner baseline first, our walkthrough on working out your cost per ticket gives you the same inputs in a form you can reuse.

Our own numbers are on the plans and pricing page, and they are the same three lines for everybody: $15, $39 or $79 a month, every feature on every plan, seats being the only difference. Trials run 14 days, which is long enough to time your own before and after rather than trusting ours. That is the only help desk ROI calculation worth much.