What we build

What one repeated task is costing you a year

Most businesses know something is slow. Very few can say what it costs. Eight boxes and you will have a number you can take to a decision.

Which page you want. This is the calculator. If you want the argument rather than the arithmetic, the hidden cost of manual work covers where the money actually goes.

Put your own numbers in

This manual work cost calculator turns one repeated task into an annual figure. You enter how often the task happens, how long it takes, how many people touch it and what an hour of their time costs. The arithmetic is printed underneath, and the answer comes back as a range rather than a single number nobody could defend.

Nothing here is stored or sent anywhere. The calculation runs in your browser, and closing the page loses it.

Use one task, not all of them. The number is more useful when it is about something specific enough to fix.

Fill the boxes above

    The arithmetic, so you can check it: people multiplied by times a week, multiplied by minutes, divided by 60, gives hours a week. Hours a week multiplied by 52 gives hours a year. Multiply by the redo share to add the corrections. Multiply by cost per hour for the money. The automatable share is applied last, and the range is that figure at 70 and 100 percent, because no automation removes a task entirely on day one.

    What this does not count: the cost of the errors that reach a customer, the revenue lost while somebody was doing this instead of selling, and the delay itself. Those are real and usually larger. Estimating them here would mean inventing a multiplier, so the number you see is the floor rather than the answer.

    Why the answer is a range

    A single figure invites a precision nobody can defend. Two businesses running the same task will get different results from automating it, because the share that is genuinely rules-based differs and because no automation removes a task completely on the first day.

    The low end assumes seven tenths of the automatable work actually goes away. The high end assumes all of it does. Real projects usually land between them and then improve, which is why the range is stated rather than the midpoint.

    The number is a floor, not the answer

    Three costs are deliberately excluded because estimating them would mean inventing a multiplier.

    The first is what an error costs once it reaches a customer. The second is the revenue lost while someone was doing this instead of selling. The third is the delay itself, which in a business that quotes or responds to enquiries is often the largest of the three.

    What to do with the number

    A yearly figure is a decision tool, not a headline. Compare it against what fixing the task would cost once, and against the same figure for the other two or three tasks competing for the same budget.

    The task to automate first is rarely the one that annoys people most. It is the one with the highest product of frequency and cost, which is usually something quiet that happens forty times a day and that nobody complains about because it has always been that way.

    A sanity check before you commit

    If the yearly cost is smaller than a month of the fix, do not automate it. Write the task down properly, hand it to one person, and spend the money on something with a bigger number.

    Where this calculator is the wrong tool

    It measures a repeated task with a stable shape. If the work changes every week, the maintenance cost of automating it outruns the saving and the honest answer is to stabilise the process first.

    It also assumes the task should exist. Some of the most valuable automation work is discovering that a report nobody reads has been produced weekly for four years. Deleting a task beats automating it every time, and no calculator will tell you which ones those are.

    What the number is worth against, and what it cannot tell you

    This figure cannot tell you whether automating the task will work. It tells you what the task costs, which is a different question. Two things decide the first one and neither is in the eight boxes: whether the process is written down anywhere, and whether somebody inside the business will own the result. Where a process lives only in one person's head, an automation has nothing to be accurate against and will be confidently wrong for months before anybody notices. The readiness assessment asks both of those directly, and one poor answer there outweighs a large number here.

    The most useful thing to do with the total is not to send it to a supplier. It is to put it beside the same figure for the other two or three tasks competing for the same budget, and then to ask whether your own team could take the first one. A rules-based task at high frequency is often within reach of somebody in-house who enjoys this sort of work, and where that person exists you should keep it there. Whoever builds it, ownership belongs with the business: the workflow, the credentials and the data in accounts under your own logins, so you own the system whether or not any arrangement continues.

    Nothing built on the back of a number like this runs entirely unattended. Anything touching money, pricing, a public statement or a serious complaint waits for human approval, and a workflow meeting a case it was not built for hands it to a person rather than improvising around it. The published version of this arithmetic is Quillon, where the delivery line became a single audited automation of 34 AI nodes, the gross margin moved from under 35 per cent to 94, and the AI cost of a full unit pack came out around four dollars. Moiz Khan owns automation architecture at Wobble, which is based in Karachi, bills month to month and carries 25 engagements across six countries.

    Common questions

    How do I work out a fully loaded cost per hour?

    Take the annual salary, add employer contributions, equipment, software licences and a share of overheads, then divide by the hours actually worked in a year rather than the hours on the contract. For most roles the fully loaded figure lands somewhere between 1.2 and 1.5 times the bare salary rate, which is why using the salary alone understates every calculation of this kind.

    Should I include time spent checking the work?

    Yes, and it is the number people leave out most often. Checking is part of the task, and in work that gets copied between systems it is frequently longer than the copying. Leaving it out is the single largest reason these estimates come out too low.

    What counts as rules-based work?

    Anything where two competent people following the same instructions would produce the same result. Copying, sorting, routing, formatting, reminding and reconciling are rules-based. Deciding whether to make an exception for a good customer is not, and that part stays with a person.

    Why does the calculator not tell me what automation would cost?

    Because that figure depends on what the task touches, how many systems are involved and whether any of them has an interface worth connecting to. Any tool quoting a price before looking at that is quoting a sales number. The point of this calculator is to tell you whether the question is worth asking at all.

    Is my data sent anywhere?

    No. The calculation runs entirely in your browser. Nothing is transmitted, nothing is stored, and closing the tab loses the numbers you entered.

    What if the task takes a different amount of time each occasion?

    Use the typical case rather than the worst one, and use the median rather than the average if a few occasions are very long. Overstating the input to make a stronger case produces a number that falls apart the first time someone senior checks it.

    See where this applies to your business

    The AI Readiness Call is a short, free conversation about where automation would actually pay back in your business. The call is free. The diagnosis is not.

    Book AI Readiness Call