The Real Cost of Manual Work: Calculate Your Number

It's Sunday night and you're on the couch with a laptop on your knees, putting together quotes. This week's orders are scattered across WhatsApp, a notebook and your memory. On Monday morning you'll clean up whatever you didn't get to enter.

Nobody sent you that bill. It isn't on any balance sheet, no supplier charges it and your accountant never sees it. But you pay it every week, in hours of yours and your team's that could be going somewhere else.

This article is about putting a number on it. You log one week, and then the math takes about 15 minutes and a spreadsheet. Once you have it, you can decide with data whether automating anything makes sense, what to start with and what to leave alone.

Why measure before you automate

Most companies adopt technology without a starting number. In a national survey on AI adoption among small and mid-sized businesses in Argentina, run by CEPE Di Tella and Fundar with fieldwork by the Fundación Observatorio PyME across 402 companies at the end of 2025, 57.4% of the firms already using AI don't formally measure its impact, as reported by Perfil (in Spanish). The sample is manufacturing and software, not services, but the pattern is familiar anywhere: you buy the tool, you use it, and nobody can say whether it paid off.

The number you're about to get works twice. Today, to decide what to automate. A few months from now, to check whether it worked.

Without a before number, every tool looks good. Or none does.

Step 1: log one week of repetitive tasks

Pick a normal week, not month-end and not a holiday week. For five working days, write down every task that is repetitive, manual and rule-based: the kind you do the same way every time, without having to decide anything important. Asana's guide to automating repetitive tasks looks for exactly that: activities that are time-consuming, manual or error-prone, that don't require much decision-making but happen often enough to be a time suck.

In a service business, the usual candidates are:

• Moving data from one place to another: from emails, WhatsApp or PDFs into a spreadsheet or a system.
• Sorting emails and routing them to the right person.
• Taking notes in meetings or calls and writing them up.
• Building, sending and following up on quotes.
• Entering data to invoice or collect.
• Putting together the report or job summary you send the client.
• Confirming appointments or visits and announcing changes.
• Hunting for a piece of information that lives in another conversation, another folder or another person.

To log it, a sheet like this is enough, one row each time the task comes up:

Day	Task	Who	Minutes	Same every time?
Monday	Copy WhatsApp orders into the spreadsheet	Admin	10	Yes
Monday	Build a quote for a new client	Owner	40	Yes
Monday	Negotiate price with a client	Owner	20	No

The last column is the one you'll use most later. Negotiating a price takes time, but it's never done the same way twice: log it anyway, and set it apart at the end.

The trick to keep the log alive all week: a five-minute alarm before you leave, every day. Logging in the moment is better, but logging at the end of the day beats not logging at all. Ask each person on your team to do the same, because they know better than you where their day goes.

What we see most in fs labs diagnoses

In the diagnoses we run, four tasks eat the most hours: data entry, sorting email, taking notes, and the routine tasks people do every day. We don't have an average number of hours to give you: it changes a lot from one business to another, which is why this article has you measure yours.

Two of them have outside measurements. They come from other countries and from companies unlike yours, so use them as an order of magnitude and nothing more:

• Data entry. In a July 2025 survey by Parseur and QuestionPro, 500 U.S. professionals said they spend an average of more than nine hours a week moving data from emails, PDFs and spreadsheets into a system. Parseur sells software for exactly that, so read the figure with that in mind.
• Email. Microsoft's Work Trend Index 2023, based on Microsoft 365 usage data, found that the top 25% of email users spend 8.8 hours a week on it. That is the high end, not the average.

For note-taking we couldn't find an independent measurement: the figures going around come from companies that sell those tools. One more reason to measure it on your own team.

Step 2: put a price on the hour

This step decides whether the number is any good, so it's worth going slowly. There are two different hours.

Your team's hour. It isn't the hourly wage. A person's real cost includes payroll taxes, health coverage, workers' compensation insurance, retirement contributions and paid leave. The U.S. Bureau of Labor Statistics measures it every quarter: in June 2026, wages and salaries were 70% of what private employers spent on compensation, and benefits the other 30%, for an average of $46.89 per hour worked.

That gives you a rough rule: divide the hourly wage by 0.7. Someone paid $14 an hour costs you about $20. The mix changes by country, state and business, so ask your accountant or payroll provider for each person's total monthly cost and divide it by the hours actually worked. They already have it.

Your hour. There's no official number, and out of habit it's easy to value it at zero, which distorts the whole calculation. Our approach, which you can adjust, is to set two values and use the more conservative one to decide:

• The floor: what an administrative employee would cost you per hour. It's what you'd pay for someone else to do that task.
• The real value: what your business would earn if you used that hour to sell, visit clients or do billable work. If your working hour is billed, that's your value.

Step 3: the math

With the tasks and the values, the math has three steps:

1. Hours per week = times per week × minutes each time ÷ 60.
2. Cost per week = hours per week × value of the hour.
3. Cost per year = cost per week × 52.

Copy this block, paste it into cell A1 of a new Google Sheets file and it splits into columns on its own. It comes with two sample rows and all the formulas in place:

Task	Who does it	Times per week	Minutes each time	Hours per week	Hourly value	Cost per week
Entering orders from WhatsApp and email into the spreadsheet	Admin	15	10	=C2*D2/60	20	=E2*F2
Building and sending quotes	Owner	6	40	=C3*D3/60	50	=E3*F3
				=C4*D4/60		=E4*F4
				=C5*D5/60		=E5*F5
				=C6*D6/60		=E6*F6
				=C7*D7/60		=E7*F7
				=C8*D8/60		=E8*F8
				=C9*D9/60		=E9*F9
Total per week				=SUM(E2:E9)		=SUM(G2:G9)
Total per month						=G10*52/12
Total per year						=G10*52

Replace the two sample rows with your own tasks. You only fill in columns A, B, C, D and F: the hours, the weekly cost and the totals calculate themselves. There's room for eight tasks. If you need more, copy the last task row down and widen the range in the totals.

Add one row almost nobody includes: fixing mistakes. The order entered with the wrong address, the invoice that had to be redone. Raymond Panko, who spent years studying spreadsheet errors at the University of Hawaii, summarizes the research on human error this way: even in simple tasks people slip about once in 200 times, and in more complex ones, between once in 50 and once in 20. It isn't carelessness, it's how attention works. If your team enters hundreds of data points a week, some will come out wrong. Log how many times a week something gets fixed and how long each fix takes, and it goes into the math like any other task.

A full example, start to finish

Illustrative example. The company is made up. Admin's hour follows the 0.7 rule above for a $14 wage; the owner's is an assumption. It is not a case from an fs labs client.

A six-person maintenance company logs a week of repetitive tasks:

The math gives $450 a week, $1,950 a month and $23,400 a year. Put another way: more than six months of an admin's full cost, every year, on tasks nobody chose to do.

The interesting part isn't the total. Of the 13.5 hours, the owner puts in 6, less than half. But those 6 hours are $300 of the $450: two thirds of the cost. If the owner had valued their own hour at zero, they would have concluded the problem is Admin's workload, and they would have automated the part that weighs least.

What the math leaves out

The number you got is a floor. Some costs don't fit in the spreadsheet and deserve a separate look:

• What doesn't get done. The quote nobody followed up on, the client who cooled off because nobody answered in time. There's no honest way to put a number on it without your own data, but if you feel it, write it down separately.
• Dependence. If only one person knows how to enter orders, everything stops the day they're out. It isn't a weekly cost, it's a risk.
• The wear and tear. It's the hardest to measure, and no less real: the Sunday night that looks more and more like one more shift.

How to read the result

With your number in hand, there are three possible readings:

• The number is small. If the total doesn't reach a couple of hours a week, automating anything probably isn't worth it right now. A rule of thumb of ours: a task that takes you less than an hour a week almost never justifies the effort.
• The number is big and sits in the owner's tasks. It's the strongest signal. Those hours cost more and, on top of that, they're the ones you take away from the business.
• The number is big but spread across many small tasks. Before automating, check whether the process is clear. Automating a messy process just makes it messy faster.

One trap is worth naming. Hours saved aren't money in your pocket. They turn into money only if someone uses them on something else: taking on more jobs, cutting overtime or getting the owner back to selling. If you automate and those hours vanish with nowhere to go, the saving exists in the spreadsheet but not in the cash register. Before automating, write down in one line where those hours will go.

And a warning about calculators on this topic, this one included: nearly all the ones we found are built by companies that sell automation. The numeric example in an article from Jestor, for instance, assumes 70% of the hours are eliminated: that's an assumption of the example, not a measured result. Treat them as an orderly way to think, not as a promise.

What to automate first

With the list sorted by weekly cost, the task at the top is the natural candidate. Before picking it, run it through three questions. They're our own criteria, not a standard:

1. Is it done the same way every time? The column you filled in the log. If each time calls for different judgment, it's a poor candidate: automation repeats rules, it doesn't replace judgment.
2. Is the data already somewhere? In a spreadsheet, in a system, in a tidy inbox. If the information lives in one person's head, the first job is to organize it, not to automate.
3. What does a mistake cost? A badly sent appointment reminder is an awkward moment; a wrong invoice has consequences. Start with what forgives errors.

Then comes the math on how much it makes sense to spend. The same Jestor article sums it up in one line: divide the total implementation cost by the estimated monthly savings and you get how many months it takes to pay for itself.

Illustrative example. A task costs you $200 a week, about $867 a month. If automating it saves half, and that "half" is your assumption and not data, the monthly saving is about $433. An investment of $2,000, used only to show the math, would pay for itself in about five months ($2,000 ÷ $433 ≈ 4.6). Our rule of thumb: if something takes more than a year to pay back, look for a smaller version of the same job.

Three mistakes when measuring

• Measuring an odd week. If your business has a busy season or month-end close, don't use that week to project the whole year.
• Counting what requires deciding. Negotiating a price or resolving a complaint isn't repetitive, even if you do it every day. That's what the last column of the log is for.
• Measuring what annoys you instead of what weighs. The task that irritates you most isn't always the most expensive. The spreadsheet is there so the number decides, not your Monday mood.

Keep the number: it's your baseline

When you automate something, don't throw the spreadsheet away. Two or three months later, measure a normal week again, with the same tasks and the same hourly values. The difference is what you actually gained, and it tells you whether the next automation is worth it or the first one needs adjusting.

That's what separates a company that "uses tools" from one that knows what they return.

Your first step

This week, nothing more: write down every repetitive task you do or see done, with the minutes and how many times. A sheet of paper and a pencil is enough, plus the five-minute alarm. On Friday, put a price on the hour and run the math.

You don't have to decide anything yet. The only thing that changes is that, for the first time, the cost of staying as you are will have a number. And if you'd like someone from outside to review it, at fs labs we run a free diagnosis: we look at how you work today, show you where the time goes and tell you honestly what's worth automating and what isn't.