Cost of Manual Processes: An Honest Worksheet

Key Takeaway

The cost of a manual process starts with one formula: minutes per run, times runs per period, times the people involved, times their loaded hourly cost. That is the floor, not the answer. The honest number adds what calculators leave out: rework, waiting, the owner's time priced at what it is really worth, the risk of one person holding the process, and what the people doing it could be doing instead. Then it subtracts what calculators over-count, because no fix removes 100% of the time and every fix costs something to build and run. Fill in the worksheet below with your own numbers, fix the process first, and price automation only against what is left. The fix often removes much of the time for free, which is why business process improvement runs in the order it does.

Why Most Process Cost Calculators Get the Number Wrong

Most process cost calculators are built by companies that sell automation, and they land on the same answer: the manual way costs a fortune and the software pays for itself almost at once.

They get there in two moves. They count only visible labor, which undercounts the real cost. Then they assume the tool removes all of it, which overcounts the savings. The errors do not cancel out. They make the number look precise while pointing it at the wrong decision.

Underneath both: most owners do not know how long the process takes. When we gap-analyzed 16 small businesses across 68 roles and 461 process areas, the average business had just 27% of its work documented. Just over half of role areas, 50.3%, had no documentation at all, and 82% of teams were below 50% coverage. If nobody has written a process down, nobody has timed it, and the number typed into a calculator is a guess. A guess multiplied by 48 weeks is just a bigger guess.

This worksheet is the opposite of a sales tool. You fill it in with numbers you measured, and it is just as happy to tell you not to buy anything.

The Core Formula

Minutes per run x runs per period x people involved, divided by 60, gives hours per period. Hours per period x periods per year x loaded hourly cost gives the annual base labor cost.

Each input has a trap.

Minutes per run. Time one real run with a stopwatch. Estimates tend to run low, because people remember the doing and forget the finding, checking, and fixing. Our guide on how to streamline a process walks through timing a run step by step.
Runs per period. Count from a calendar, a system log, or a sent folder. "A few times a week" might turn out to be eleven.
People involved. Everyone who touches a run, including whoever checks it. If each person spends a different amount of time, give each one a line.
Loaded hourly cost. Not the wage. Add up a year of wages, payroll taxes, benefits, and per-person costs like software seats, then divide by the hours actually worked. Use your own payroll, not a rule of thumb.
Periods per year. For a weekly process, use the weeks people actually work, not 52.

The Costs Calculators Leave Out

Base labor is the floor. Five more costs sit on top of it.

Rework and error correction

When the output is wrong, someone finds the mistake, fixes it, and sends it again. Rework rarely shows up in minutes per run, because whoever times the task times a clean run. Look back over the last quarter, list every correction, and attach the hours each took, plus any credit or write-off it caused.

Waiting and delay

A manual process also costs the time everything else spends waiting for it. An invoice that goes out three days late is three days of cash you do not have. Write down what waits and for how long, and price it where you can. If you are not sure where the waiting collects, start with how to find the bottleneck in your business.

Owner touches, at the owner's real value

Owner minutes do not belong at the staff rate. Price them at what an hour of the owner's time produces when it goes to the work only they can do: selling, leading, deciding where the business goes. One way to set it is to estimate the gross profit the owner's best work drives in a year and divide by the hours spent on that work. If you do not know how often a process pulls you in, keep an interruption log for a day and count.

Single-point-of-failure risk

If only one person can run the process, its cost includes what happens when that person is out. Price what you can see: the weeks of vacation or sick time when someone else covers slowly or the work just waits. The case you cannot fully price, the day they resign, is the one that should worry you most. We cover it in key-person risk.

Opportunity cost of who is doing it

The loaded rate tells you what the hours cost, not what they could have produced. A sales manager copying numbers is not selling. You do not always need a dollar figure here, but write down whose time it is and what they would do with it back.

The Costs Calculators Over-Count

An honest worksheet also subtracts.

Assuming 100% of the time disappears. No fix removes all of it. Someone still reviews the output, handles the exceptions, and repairs the integration when a vendor changes something. Walk the documented process, mark which steps the fix removes, and count only those. If you cannot say which steps go away, you are not ready to price the fix.

Ignoring the cost of building and maintaining the fix. Changing a process takes staff time. Automation has setup, a subscription, and someone who owns it when it breaks. Custom software has a build cost and upkeep. Put all of it on the same basis as the savings. Our article on how much custom software costs covers what moves the price of a build.

The Worksheet

Copy this into a spreadsheet and fill in the last column. Run it on one process at a time.

Line item	How to measure it	Your number
Minutes per run	Stopwatch one real run, start to finish, for each person
Runs per period	Count from a calendar, log, or sent folder
People involved	Everyone who touches a run, including checkers
Loaded hourly cost	A year of wages, taxes, benefits, and per-person costs, divided by hours worked
Periods per year	Weeks or months the process actually runs
Base labor cost per year	Minutes x runs x people, divided by 60, x periods per year x loaded hourly cost
Rework and error correction	Last quarter's corrections x hours each, scaled to a year, plus credits
Waiting and delay	What waits, for how long, and the cost where you can price it
Owner touches	Owner hours per year x the owner's real hourly value
Single-point-of-failure risk	Absences per year x extra cover hours x loaded hourly cost
Opportunity cost	Whose time it is and what they would do with it back
Total annual cost	Sum of every priced line
Realistic savings per year	Only the steps the fix removes, never 100%
One-time cost of the fix	Staff time to change the process, plus setup or build
Running cost per year	Subscriptions, plus upkeep hours x loaded hourly cost
Payback in months	One-time cost divided by net monthly savings

A Worked Example: A Hypothetical Weekly Report

Here is the worksheet filled in for a made-up process. The inputs are round example numbers chosen so you can check every line. They are not benchmarks and do not come from any client. Swap in your own.

The process. Every week, an office coordinator pulls sales numbers out of two systems and pastes them into a spreadsheet. A manager then rebuilds the same totals by hand to check them, because nobody trusts the inputs. The owner reads the report and sends questions back.

Input	Example value
Minutes per run, per person	240
Runs per week	1
People involved	2 (coordinator and manager)
Loaded hourly cost, staff	$40
Working weeks per year	48
Owner's real hourly value	$150
Owner review time	30 minutes a week
Corrections	1 week in 4, 2 hours each
Coverage	Coordinator out twice a year, 6 extra hours each time

Step 1: Base labor

240 minutes x 1 run x 2 people = 480 minutes a week
480 minutes divided by 60 = 8 hours a week
8 hours x 48 weeks = 384 hours a year
384 hours x $40 = $15,360 a year

Step 2: The lines calculators leave out

Rework: 48 weeks divided by 4 = 12 corrections. 12 x 2 hours = 24 hours. 24 hours x $40 = $960 a year
Owner touches: 30 minutes x 48 weeks = 1,440 minutes. 1,440 divided by 60 = 24 hours. 24 hours x $150 = $3,600 a year
Single point of failure: 2 absences x 6 extra hours = 12 hours. 12 hours x $40 = $480 a year
Waiting and opportunity cost: not priced here, but written down. The owner decides on last week's numbers until the report lands, and the manager's four hours come out of time for the team.

Total annual cost: $15,360 + $960 + $3,600 + $480 = $20,400 a year.

The owner's 24 hours cost $3,600. The 24 hours of rework cost $960. Same hours, almost four times the cost, which is why owner time gets its own line.

A typical calculator would stop at Step 1, assume all 384 hours disappear, and promise $15,360 a year in savings. Here is the honest version.

Step 3: Fix the process first

Before pricing any tool, the business maps the report and makes two decisions. Every number gets one agreed source, pulled into one place. With trusted inputs, the manager's hand rebuild has nothing left to catch, so it goes. Assume the coordinator alone now builds the report in 180 minutes.

180 minutes x 1 run x 1 person = 180 minutes a week
180 minutes divided by 60 = 3 hours a week
3 hours x 48 weeks = 144 hours a year
144 hours x $40 = $5,760 a year
Savings: $15,360 minus $5,760 = $9,600 a year, and $9,600 divided by 48 weeks = $200 a week

Assume the fix takes 20 hours of staff time: 20 hours x $40 = $800, with no software. Payback: $800 divided by $200 a week = 4 weeks.

To stay conservative, the rework, owner, and coverage lines stay unchanged, even though a cleaner process may shrink them.

Step 4: Price automation against what is left

The automation now competes for 3 hours a week, not 8. Assume an integration removes 2 of them, and the coordinator keeps 1 hour to review the output and handle exceptions.

Gross savings: 2 hours x 48 weeks = 96 hours. 96 hours x $40 = $3,840 a year
Subscription: $40 a month x 12 months = $480 a year
Upkeep: 1 hour a month x 12 months = 12 hours. 12 hours x $40 = $480 a year
Net savings: $3,840 minus $480 minus $480 = $2,880 a year, and $2,880 divided by 12 = $240 a month
One-time setup: assume $2,400
Payback: $2,400 divided by $240 a month = 10 months

Ten months is a reasonable return, but it is a different decision from the calculator's. Priced against the original $15,360, the same setup looks like it pays back almost at once: $15,360 divided by 12 = $1,280 a month, and $2,400 divided by $1,280 is under two months. Worse, that automation would likely have been scoped to reproduce the manager's duplicate check, a step one decision removed for free.

Step 5: The ceiling for a custom build

After the process fix, everything left adds up to $5,760 + $960 + $3,600 + $480 = $10,800 a year. Suppose a build removed every dollar of that, which no build does. For payback within two years, the most it could cost is 2 x $10,800 = $21,600, before running costs. That is a ceiling, not a quote. For one weekly report, it points at the automation. A build earns its place when the same math runs across several processes that share the same data, so their savings stack against one cost.

Compare the Fix Options

Option	What it costs	What it removes	When it is the right call
Fix the process	Staff time to map and change it	Duplicate steps, unneeded checks, waiting at handoffs	Always first. In the example, $800 removed $9,600 a year
Configure or automate	Setup, a subscription, upkeep hours	Repetitive, rule-based steps that survive the fix	What is left is mostly data moving between tools you own
Custom build	Build cost plus ongoing upkeep	Work off-the-shelf tools fight, data spread across spreadsheets, several processes at once	Leftover cost across processes clears the ceiling you set before seeing a quote

If the real data lives in a workbook the business depends on, start with how to turn a spreadsheet into an app, and if you are weighing a build, read build vs buy for custom software first. When the numbers do justify a build, our sister company, The Software Effect, builds custom software you own, not rent, shaped around the process you already mapped.

A Simple Payback Calculation

Net monthly savings = (realistic annual savings minus annual running cost) divided by 12

Payback in months = one-time cost of the fix divided by net monthly savings

Use realistic savings, never 100%. Include every running cost, upkeep hours too. And decide what payback you will accept before you see a quote, so the quote does not decide it for you. If you need to take the number to a partner or a board, the business case for systems shows how to frame it.

Why You Fix the Process Before You Price the Automation

In the example, the process fix removed 5 hours a week and the automation removed 2: more than twice the hours, for a third of the setup cost, with no subscription. The numbers are made up, but the reasons behind the order are not.

The fix often removes time for free. Duplicate checks, retyped data, and habit steps are decisions, not purchases.

Automation multiplies whatever it touches. Automate a duplicate check and you pay to run it forever. That is why the four stages of business process improvement put document, diagnose, and improve before solve.

The documented process makes the savings line honest. You cannot mark which steps a tool removes until the steps are written down, and the same document becomes the specification for whatever comes next.

We hear the raw material for this on discovery calls. One owner told us a routine report took about eight hours, most of it moving numbers from one place to another. Another business spent about three days each cycle auditing commissions by hand. A third spent around three hours of data entry on every regulatory submission. The first useful question in each case is not which tool to buy, but why the work happens by hand at all. We walk through where each one breaks in business process improvement examples.

Where to Start

Pick the manual process that eats the most hours or pulls in the owner most often. Time one real run. Fill in the worksheet, hidden lines included. Fix the process and run the new version for a few weeks. Then price the automation or the build against what is left, not against what you started with.

Frequently Asked Questions

How do you calculate the cost of a manual process?

Multiply minutes per run by runs per period by the people involved, divide by 60, then multiply by periods per year and the loaded hourly cost. That is the base labor cost. Then add what calculators skip: rework, waiting, owner time at the owner's real value, the cost of covering for the one person who knows the process, and the opportunity cost of whose time it is.

What is a loaded hourly cost?

It is what an hour of an employee's time really costs the business. Add up a year of wages, payroll taxes, benefits, and per-person costs like software seats and equipment, then divide by the hours that person actually works. Use your own payroll figures, because the result drives every other line on the worksheet.

How do you calculate automation ROI for a small business?

Fix the process first, then measure what is left. Count only the steps the automation removes, multiply those hours by the loaded hourly cost, and subtract the subscription and upkeep to get net annual savings. Divide by 12 for net monthly savings, then divide the one-time setup cost by that figure for payback in months.

Are online process cost calculators accurate?

Most are built to sell a product, so they lean one way. They count only visible labor, then assume the tool removes all of it and ignore setup and upkeep. Borrow the formula if you like, but use your own measured numbers and subtract what the fix costs to run.

Should I fix a process before automating it?

Yes. The fix often removes a large share of the time for nothing more than a decision and some staff hours, and automating first means paying to run steps you never needed. Fixing first also leaves you with a documented process, which shows exactly which steps a tool would remove.

What Does a Manual Process Really Cost? (A Worksheet)