Most teams can tell you how many products they onboarded last year. Far fewer can tell you what that work cost, because it never shows up as one line in a budget. It sits in the hours of people in purchasing, product data and e-commerce who open supplier files, find the values, type them in and check them.
The openProd onboarding cost calculator puts a number on that work. This article explains what it asks for, how it calculates, which figures are measured and which are assumptions, and what it leaves out on purpose. If you plan to show the result to a finance owner, these are the questions they will ask.
What the calculator asks for
Four inputs on the front, two more in the Assumptions panel.
| Input | What it means | Default |
|---|---|---|
| Products per year | New products your team onboards from supplier data each year | 5,000 |
| Minutes per product today | Finding, typing and checking one product by hand | 25 min |
| Loaded cost per hour | Salary plus overheads of the people doing the work | EUR 30 |
| How supplier data arrives | Spreadsheets, PDFs or mixed | Mixed |
| Human review per product with AI | A person checking and correcting what the AI prepared | 5.5 min |
| Months to full use | How long until all of the volume runs through the new process | 3 |
The defaults are a starting point, not a benchmark. The most useful thing you can do with the calculator is replace “minutes per product today” with a number you have timed on your own supplier files.
The formulas in plain words
There is no hidden model. The whole calculation is five steps:
- Today. Products a year x minutes per product today / 60 = hours a year. Hours x loaded cost per hour = cost a year.
- With AI-assisted onboarding. Products a year x human review minutes per product / 60 = hours a year, at the same cost per hour.
- Saved. Today minus with AI, in hours and in euros. The calculator never shows a saving below zero.
- Capacity. Hours saved / 1,720 = FTE. The 1,720 hours are 215 working days of 8 hours.
- First 12 months. The annual saving x the average share of products on the new process while use ramps up evenly.
Step 2 is the one to understand. With AI-assisted onboarding the machine extracts and maps the supplier data, but a person still checks every product. The calculator counts that person’s review time as the staff time with AI, and nothing else.
Review time is a planning assumption
The suggested review time depends on how supplier data arrives:
| How supplier data arrives | Suggested human review per product |
|---|---|
| Spreadsheets | 5 min |
| Mixed | 5.5 min |
| PDFs | 6 min |
These figures are planning assumptions, not measurements. A PDF layout needs a closer check than spreadsheet columns, so it gets more time. You can change the value in the Assumptions panel, and you should if your products carry long attribute lists or need technical checks after extraction.
If you set review time as high as today’s manual time or higher, the calculator says plainly that no time is saved. That is intended. A calculator that shows a saving whatever you enter is not measuring anything.
The measured run behind the time figures
The time figures are checked against one measured run, an openProd import test on real customer data, presented at Infoshare 2026:
Measured on a real supplier catalogue: 48 pages, 80 products. Manual entry: 30 to 50 hours. AI extraction in openProd: 16 minutes, plus human review.
Per product, manual entry on that catalogue took 22.5 to 37.5 minutes. The calculator’s default of 25 minutes sits near the low end of that range, so the default scenario does not lean on the most expensive case. The AI extraction came to about 12 seconds of machine time per product. That is machine time, and the calculator does not count it as staff time.
One limit should be stated directly: review time was not part of that measurement. That is why the calculator adds review time from its Assumptions instead of presenting it as measured. One catalogue is one data point, and it is shown as one.
A worked example with the defaults
Take the calculator as it opens: 5,000 products a year, 25 minutes per product today, EUR 30 per hour, mixed sources, 5.5 minutes of review per product and 3 months to full use.
| Hours a year | Cost a year | |
|---|---|---|
| Today, manual entry | 2,083 | EUR 62,500 |
| With AI-assisted onboarding | 458 | EUR 13,750 |
| Saved | 1,625 | EUR 48,750 |
The arithmetic behind it:
- Today: 5,000 x 25 / 60 = 2,083 hours. At EUR 30 per hour that is EUR 62,500.
- With AI: 5,000 x 5.5 / 60 = 458 hours, or EUR 13,750.
- Saved: EUR 48,750 and 1,625 hours. 1,625 / 1,720 is about 0.9 FTE.
The annual figures assume full use of the new process. The first-year figure takes the ramp-up into account. With 3 months to full use, a third of the volume goes through the new process in month one, two thirds in month two and all of it from month three on. Averaged over 12 months that is 11/12 of the volume, so the first 12 months come to about EUR 44,700 (EUR 48,750 x 11/12 = EUR 44,687.50, which the calculator rounds to the nearest hundred).
Next to the result the calculator also shows a volume range. If you onboard 20% fewer or more products, 4,000 to 6,000, the saving with the same assumptions is EUR 39,000 to EUR 58,500 a year.
How much the review assumption moves the result
Because review time is an assumption, it is worth testing. Same 5,000 products, 25 minutes today and EUR 30 per hour; only the review time changes:
| Review per product | Hours with AI | Saved a year | Capacity |
|---|---|---|---|
| 5 min (spreadsheets) | 417 | EUR 50,000 | 1.0 FTE |
| 5.5 min (mixed) | 458 | EUR 48,750 | 0.9 FTE |
| 6 min (PDFs) | 500 | EUR 47,500 | 0.9 FTE |
| 10 min | 833 | EUR 37,500 | 0.7 FTE |
Even with nearly double the suggested review time, the saving shrinks but does not disappear, because today’s manual time is the larger number in the calculation. That is also the point: minutes per product today drive the result more than any other input, which is one more reason to time them on your own files.
What the calculator does not include
The calculator shows the work and only the work. It leaves out:
- Software cost, including AI processing. Pricing depends on volume and on the sources you work with, so it is a separate conversation. Set it against the saving once you have it.
- Setting up the process. The time to configure it, train the team and connect your systems, in other words implementation and change management.
- Errors and rework downstream, and the value of products going live sooner. Both are real, and both are hard to put a number on honestly, so they stay out of the figure.
- Tax, salary changes and currency effects.
Implementation and change management belong in a full business case. They depend on your systems and your team, which a public calculator cannot know. Leaving them out means the figure says exactly one thing: what the onboarding work itself costs, today and with AI-assisted onboarding.
Hours saved are capacity, not a payroll cut
The euro figure values the hours saved at your loaded cost per hour. That does not mean the money leaves the payroll. The calculator states it in one line: “Hours saved are capacity your team gets back, not a promised payroll cut.”
In practice those hours go back to work that was waiting: more suppliers, more attributes per product, earlier launches, closer checks. 0.9 FTE of capacity is a statement about time. Present it that way. A finance owner who is promised a payroll cut will look for it a year later.
How to run it on your own numbers
- Time the manual work on a recent supplier file: from opening the file to a checked product in your system, divided by the number of products in it.
- Count products from supplier data over the last 12 months, not the size of your whole catalogue.
- Ask finance for a loaded hourly cost, not a salary.
- Pick the source mix honestly and keep the suggested review time unless you know your products need longer checks.
- Keep the result. The share link contains only your inputs, and the PDF report shows the same numbers as the page.
If you want to see where the review minutes go, the import workflow shows how openProd extracts and maps a supplier file and where a person checks the result before it is saved. You can use openProd in front of your PIM, or keep your catalogue in openProd.
Sources
- openProd onboarding cost calculator: the formula, the assumptions and what is not included
- openProd import test on real customer data (48-page supplier catalogue, 80 products), presented at Infoshare 2026. Event write-up: openProd at the Infoshare 2026 Pomeranian Startup Contest
Zofia Kamińska is an AI character created by openProd, not a real person. This article was reviewed by Olgierd Mrozik.
