The number nobody tracks
Most companies track the cost of software subscriptions down to the dollar. Almost none track what it costs to move the same order, invoice, or lead record by hand between an email, a PDF, and a spreadsheet three or four times before it's usable.
According to a 2025 survey by Parseur and QuestionPro, manual data entry costs US companies roughly $28,500 per employee per year, and workers spend more than 9 hours a week moving data between emails, PDFs, and spreadsheets. That's not a rounding error. It's over a full workday a week, per person, spent re-typing information that already exists somewhere else.
Separately, research from Tier2 Systems found that operations professionals spend about 3.6 hours a week fixing spreadsheet errors — roughly 22 workdays a year, or close to a full month of paid time, cleaning up mistakes that manual re-entry introduces in the first place.
"The cost isn't the typing. It's the typing, plus the errors it causes, plus the time spent finding and fixing those errors later."
Neither number tells you what your company is losing. For that, you need to measure your own workflows. This article gives you the formula, a worked example, and a simple way to run the measurement yourself in one week.
The measurement formula
The math is simple on purpose — you want something a manager can calculate on a whiteboard, not a data science project.
Occurrences per week
x People involved
x Minutes each
÷ 60
x 50 weeks
x Fully loaded hourly cost
= Annual cost of the workflow
A few notes on each variable:
- Occurrences per week — how many times this task happens: orders processed, invoices entered, leads copied from a form into a CRM, etc.
- People involved — count everyone who touches the data, not just the person who "owns" the task. A sales coordinator who re-keys an order and a bookkeeper who re-checks it are both in scope.
- Minutes each — the real time per person per occurrence, including reopening the file, finding the source data, and saving/confirming — not just the typing itself.
- 50 weeks — a standard working-year assumption that nets out holidays and PTO. Adjust if your team's schedule is different.
- Fully loaded hourly cost — salary plus payroll taxes and benefits, divided by roughly 2,000 working hours a year. If you don't have this figure, a rough placeholder (salary ÷ 2,000 x 1.25) is a defensible illustrative starting point — label it as an estimate internally.
Worked example (illustrative only)
This example is illustrative, not a client result. It shows how the formula behaves with round numbers.
Say a company processes 40 orders a week. Each order gets touched by 4 people — sales, order entry, warehouse, and billing — and each person spends about 12 minutes on their piece of it: reading the order email, retyping it into the order system, checking it against inventory, and re-entering it again for invoicing.
40 orders/week x 4 people x 12 minutes ÷ 60 x 50 weeks x $35/hr fully loaded
= 40 x 4 x 12 = 1,920 minutes/week = 32 hours/week
= 32 hours x 50 weeks = 1,600 hours/year
= 1,600 x $35/hr ≈ $56,000/year
Thirty-two hours a week is nearly a full-time employee spent solely on re-typing the same order into different systems. That is the kind of number that turns "we should probably fix that someday" into a budgeted project.
Breaking down a typical re-entry workflow
Most manual data entry problems aren't one step — they're a chain of small handoffs, each adding a few minutes and a chance for error.
| Step | Who does it | Typical minutes | Common error introduced |
|---|---|---|---|
| Read incoming email/PDF | Sales or admin | 2–3 | Misread quantity or SKU |
| Re-key into order system | Order entry | 4–5 | Transposed digits, wrong customer |
| Cross-check against inventory | Warehouse | 2–3 | Missed stock discrepancy |
| Re-enter into invoicing/accounting | Billing | 3–4 | Duplicate or missing invoice |
| Fix an error found later | Whoever notices | 5–15 | New error introduced during the fix |
That last row is the one most companies forget to count. Rework isn't a separate problem from data entry — it's a direct cost of it, and it's exactly what the Tier2 Systems research on spreadsheet error fixing time is measuring.
How to run a one-week measurement
You don't need software to get a usable number. You need a shared log and one week of honest entries.
What to log:
- Task name (e.g., "enter wholesale order into system")
- Time started and time finished
- Number of people who touched this instance of the task
- Any rework: did someone have to redo or correct this later?
Who logs it: Ask the people actually doing the work to log their own time, not their manager's estimate of it. Self-reported logs are more accurate for this kind of task because managers usually only see the visible part of the work, not the reopening files, searching for the right version, and Slack messages asking "which number is right."
Common mistakes to avoid:
- Forgetting rework. If a mistake gets caught and fixed two days later, that fix belongs in this week's log too, tagged to the original task.
- Ignoring context-switching cost. The 12 minutes spent re-keying an order also costs the 3–5 minutes it takes to refocus on whatever the person was doing before. Most logs miss this entirely, which means most estimates are conservative, not inflated.
- Only counting one person. If four people touch an order, log all four. Undercounting people is the single biggest reason internal estimates come in low.
- Measuring during an unusually slow or busy week. Try to pick a representative week, and note if it wasn't.
At the end of the week, plug your logged numbers into the formula above. You'll have a defensible, company-specific annual cost — not an industry average.
What to do with the number
Once you have a dollar figure per workflow, rank your candidates by two things: how often the task happens, and how many people it touches. A task that happens 40 times a week with four people in the loop is a better first target than a task that happens twice a week with one person, even if the second one feels more painful.
Good first candidates for automation usually share these traits:
- The data already exists digitally somewhere (email, PDF, form, spreadsheet) — nobody has to start collecting new data.
- The rules for moving the data from A to B are consistent, not case-by-case judgment calls.
- More than one person currently has to touch it.
- Errors in it are expensive or embarrassing to fix (a wrong invoice, a missed order, a duplicate entry).
Not every manual process is worth automating. Rare, one-off tasks or ones that genuinely require human judgment on every instance usually aren't. The formula above is useful precisely because it stops "this feels tedious" from being the only test — it turns the decision into an annual dollar comparison against the cost of fixing it.
Where to go from here
The home page at / includes a lost-time calculator you can use to sanity-check a rough estimate before you commit to a full week-long log — it uses the same formula described here. The same calculator also appears on the /pricing page next to the offer ladder, so you can see roughly what a fix might cost against what the problem is costing you.
If you want a structured, outside look at your workflows instead of doing the log yourself, an Ops Teardown is a fixed-price $490 engagement: a 60-minute recorded workflow review, a written map of the three highest-ROI automations we find, and a fixed-price quote within 5 business days. The fee is credited toward a build if you move forward.
You can also start with a free AI automation audit if you want a lighter-weight first look before committing to a paid teardown.