What a manual process actually costs per year

How the calculation works

There is no hidden model inside this tool — just four lines of arithmetic you could run on paper. The tool only makes it faster and stops you from skipping a line:

1. Annual hours = people × hours per person per week × working weeks per year. The default of 46 weeks represents a working year with annual leave, public holidays and sick days already removed. Change it if your calendar differs.
2. Direct labour cost = annual hours × fully loaded hourly cost.
3. Rework burden = monthly error count × hours to fix one error × 12 months × hourly cost. This line is shown separately on purpose: most companies can estimate the labour but have never measured the correction time.
4. Annual invisible cost = direct labour + rework.
5. Automatable share = annual invisible cost × the percentage you entered.
6. Rough payback = pilot budget ÷ (automatable share ÷ 12), expressed in months and based on direct labour only.

The assumptions we refuse to make

Most calculators of this kind bury invisible multipliers inside the maths to inflate the result: “automation removes 70% of process work”, “the average error rate is 12%”, “the investment pays back in six months”. Putting a number like that on screen requires having measured it. We have not measured your process, so:

• No savings rate is assumed. You enter the automatable share yourself, and the field label states plainly that it is an estimate.
• No error rate is assumed. Error count and correction time are two separate fields. Leave them both at zero and the rework line drops out of the calculation entirely.
• Opportunity cost is excluded. The sales calls, design work and customer attention the same team could not deliver while re-keying data is a real loss, but it is not measurable here — so it stays out of the total. The figure on screen is a floor, not a ceiling.
• No inflation or wage growth is applied. The calculation covers a single year at today’s rates. For a multi-year view, switch to the 5-year TCO comparison.

How to fill in each field

People

Think about one process, not the whole company. Pick a single flow — order intake, stock counting, approval chasing, production reporting — and count everyone who touches it at least once a week. If the warehouse fills in the form, finance checks it and a manager approves it, that is three people. The number is only meaningful per process; to measure several, run the tool separately for each and add the results.

Weekly hours

This is where the biggest mistake happens: people only count “the time I spend in the spreadsheet”. But the process starts before the file is opened. Chasing the information, the “can you resend that line” exchange, comparing two files side by side, re-checking in the evening — all of it is the same job. Keeping rough notes for a single week produces a far better input than a guess.

Fully loaded hourly cost

The correct definition is total cost to the business: gross pay plus employer contributions and benefits, divided by annual working hours. Using take-home pay makes the result meaningfully smaller. For a mixed team — warehouse staff, engineer, manager — a weighted average is close enough.

Errors and correction time

An error is not only a wrong number: a short shipment, a duplicate record, stock that does not reconcile at month end, a delivery sent to the wrong address. When estimating correction time, include the time it took to notice the error and the communication around it — that is usually where most of the time goes, not in the fix itself.

Automatable share

Software takes over work whose rules can be written down. A dealer order form that creates the record directly, a barcode count that adjusts stock instantly, an approval that advances with one tap on a phone — all automatable. Price negotiation, exception decisions and customer conversations are not. Split your process into those two buckets and derive a rough percentage. A generous number makes the output look better and makes the decision worse.

How to read the result

The figure is not a “loss” — it is a measure of displacement. That money is already being spent; it is simply hidden inside payroll, which is why nobody puts it next to a software budget. A healthy reading looks like this:

• Well below your pilot budget: this process is not urgent. Measure a different, more repetitive one instead.
• Close to it: cost alone cannot decide. Speed, traceability, audit trail and handover ease all matter and none of them are priced here.
• Clearly above it: the question stops being “should we” and becomes “which module first”. In that case a single-process pilot is far healthier than trying to build the whole system at once.

Read the payback period conservatively too. It counts direct labour only and ignores the transition period, when the old and new ways run in parallel, plus training and adjustment time. In practice the first few months add work rather than removing it. Treat the number of months on screen as the best case.

Three common mistakes

1. Rolling the whole company into one calculation. “Everyone loses three hours a week” in a 40-person firm produces a large, unusable number. A small figure measured per process is what actually drives a decision.

2. Picturing only a good week. Month end, stock-count periods and seasonal peaks pull the average up noticeably. Think of a typical week across the year, not the quietest one.

3. Skipping the rework line. The labour line is visible and easy to predict; the rework line is invisible and usually more expensive. A rough estimate beats leaving it at zero.

What to do next

Once you have the number there are two useful paths. To test different scenarios, change the fields and recalculate. To compare the multi-year cost of off-the-shelf software against a custom build, move to the 5-year TCO comparison; to see which published band your scope falls into, use the budget and timeline estimator. For the same subject in prose, our process automation page and guides cover it in more depth.

Frequently Asked Questions