Ask most vendors how they got to the saving figure on their homepage. The honest answer is usually a spreadsheet nobody kept.
Our own free AI ROI calculator makes the same kind of estimate. So it deserves the same scrutiny we would apply to anyone else's.
Here is exactly what it multiplies. The constants sitting quietly behind the sliders. And what the arithmetic cannot see, however carefully you fill in the boxes.
The four numbers it multiplies
You give the calculator four things. How many people are on the team. How many hours each person loses to repetitive work in an average week. A fully loaded hourly cost for that time. And the share of that work you believe AI could take on, picked from three preset tiers rather than typed in freehand.
From there the arithmetic is plain multiplication. Nothing hidden underneath it.
Hours saved in a week is team size times hours lost per person times the automation share. Hours saved in a year is that weekly figure times a fixed working-weeks constant. The pound saving is annual hours times the hourly cost you entered. The full-time-equivalent headcount is annual hours divided by a standard full-time-hours constant.
That is the whole model. No compounding multiplier. No productivity ramp. No discount for the months it takes a team to trust a new system before leaning on it.
The constants under the bonnet
Two figures never appear in the input boxes. Both are worth naming, because both do a lot of quiet work.
The first is a working-weeks-per-year constant. UK bank holidays and standard annual leave are already stripped out of it. It turns your weekly saving into an annual one.
The second is a standard working week length. It only converts an annual hours figure into a full-time-equivalent count.
Both are fixed. They apply the same way to a small agency and a large warehouse, because neither is measured against your business. They are only assumed for it.
A calculator that asked for your real holiday allowance and your real contracted hours would need several more fields. A form nobody finishes is worse than a form that makes one clean assumption and says so.
The automation share works on the same principle. Rather than asking you to guess a percentage out of thin air, the tool offers three labelled tiers.
Conservative is what off-the-shelf tools reach with no change to how the team currently works. Typical is what we see from a focused first automation project. Ambitious needs the workflow redesigned around AI, not AI bolted onto the process you already have.
Choosing a tier picks the number behind the label for you. You are not estimating a percentage. You are picking a level of ambition and letting the calculator hold the number.
What the tool already says it leaves out
To its credit, the page is upfront about two exclusions. They sit right there in the copy under the sliders.
It measures time saved only. So it leaves out revenue upside that is genuinely harder to price. Leads answered while they are still warm. Invoices chased before they go stale. Fewer mistakes needing rework later.
It also leaves out implementation cost entirely. The stated reasoning is that for most small businesses that cost is smaller than the annual saving it is weighed against.
Both are honest, declared omissions. They sit in the FAQ for anyone who reads past the number.
Three things the arithmetic assumes away
None of that is where the model actually gets fragile. The real gaps sit where the formula never looks.
Whether anyone keeps using it. The maths assumes the automation share you selected holds steady, every week, for the rest of the year. There is no field for a system that gets built, demonstrated once, and then quietly stops being opened. That happens when it does not fit how the team works day to day. That gap, between what a tool can do and what a team keeps reaching for, is where most automation projects stall. No multiplication catches it.
What it costs to keep working. The calculator treats the automation share as free to hold onto once you have it. Set it and forget it. In practice, something has to keep it working. A workflow breaks when a connected app changes its interface without warning. A prompt needs retuning when the underlying task shifts. Someone has to notice when the output has quietly got worse, before a customer does. None of that shows up in the four inputs, because there is nowhere in the formula for it to go.
Where the freed hours actually land. Hours saved only turn into pounds saved if someone deliberately redirects that time to something worth the hourly cost you typed in. Left alone, freed time gets absorbed back into a slightly slower version of the same day. The inbox still takes as long to clear. Meetings still run to their old length. The calculator cannot know what your team would do with time back each week. That is a decision your business makes, not a step in the arithmetic.
What the number is actually for
None of that makes the estimate worthless. It just marks it as a sizing tool rather than a business case.
The figure tells you whether a topic is worth an hour of proper investigation. It does not tell you what to put in front of a board or a lender.
Treat it the way you would treat a sharp guess from a colleague who knows your business reasonably well. A reason to look closer. Never a reason to stop looking.
That is also the honest limit of a page like this one. We could add fields for adoption, upkeep, and redirected time. But a calculator that asked you to forecast your own team's habits would stop being quick. It would become a guess dressed up as precision. That is exactly what we are trying not to build.
Want to see the formula in front of you? Open the calculator and put your own team size, hours, and hourly cost through it. The working shown above is what runs behind the sliders.
If the result makes automation worth investigating, Talk to us. We will work through which process would actually survive adoption, upkeep, and a place to send the freed hours. On the call, not in a spreadsheet.