The cost of manual product data
What product data onboarding actually costs
The Smart Station team6 min read
Every distributor knows that loading a new supplier range takes time. Very few can say how much. The work sits between departments, never appears as a line on a budget, and is absorbed by whoever has capacity. That invisibility is the problem — you cannot manage, resource or price something nobody has measured.
This is the arithmetic, and the three costs that never reach a budget.
Timing the work honestly
The instinct is to time the typing. That undercounts by roughly half, because the typing is the last step. A complete product record for a technical item requires:
- Finding the correct supplier page for the exact part number, including variants and superseded codes
- Reading the specification and deciding which attributes matter to your buyers
- Writing a short description and a long description that are not copies of each other
- Downloading the image at usable resolution, cropping it, and checking the background
- Finding the datasheet, which is often two clicks away from the product page
- Deciding the category, brand and any custom attributes
- Loading all of it and checking it rendered correctly
Timed properly, across a mixed technical catalogue, this lands between 20 and 40 minutes a SKU. Simple consumables sit at the bottom of that range. Anything with a specification table, a compliance certificate or multiple variants sits at the top.
Use 25 minutes as a working average unless you have timed your own. If your catalogue is heavily technical, use 30.
The worked example
Take a distributor onboarding 2 000 SKUs in a year — one substantial supplier range, or three smaller ones.
| Calculation | Result | |
|---|---|---|
| SKUs a year | — | 2 000 |
| Minutes each | measured average | 25 |
| Total minutes | 2 000 × 25 | 50 000 |
| Total hours | 50 000 ÷ 60 | 833 |
| Loaded rate | see below | R150/hr |
| Direct cost | 833 × 150 | R125 000 |
Eight hundred and thirty-three hours is roughly five months of one person's full working time, spent entirely on transcription. Most businesses discover this number and immediately dispute the hourly rate, which is the right instinct — so it is worth being precise about it.
R150 an hour is not a salary. It is a loaded cost: gross salary, plus the statutory contributions, plus the proportion of overhead that person consumes, divided by genuinely productive hours rather than contracted ones. For a junior administrator on R15 000 a month in South Africa, R150 an hour is a reasonable working figure and often conservative.
The three costs that never reach a budget
The R125 000 is the visible part. Three further costs are real, and none of them appears anywhere.
Rework. A proportion of records are wrong on the first pass — the wrong variant, a specification misread, an image that turns out to be the family shot rather than the part. Finding and fixing those costs more per record than creating them did, because somebody must first notice. Ten percent rework on 2 000 SKUs is another 80 hours.
Knowledge loss. The person doing this work accumulates knowledge that exists nowhere else: which supplier hides datasheets behind a login, which manufacturer uses a different part-number format in their PDF than on their website, which brand's images always need cropping. When they leave, that goes with them, and the next person rediscovers it at full price.
Opportunity cost. This is the largest and the least visible. It is the supplier range you did not take on, because the person who would have loaded it was already committed for five months. Nobody records a decision not to expand.
What the number is actually for
The point of calculating this is not to justify buying software. It is to make a genuinely invisible cost visible, so that it can be managed like any other.
Once the number exists, three conversations become possible that were not possible before: whether to hire, whether to outsource, and whether to automate. All three are reasonable answers. What is not reasonable is continuing to absorb a R125 000 annual cost that nobody has ever seen.
If you take one thing from this: time ten SKUs properly, end to end, with a stopwatch. The number will be higher than your estimate, and it will change the conversation.
A word of caution on the other side. Below roughly 800 SKUs a year, the arithmetic does not support buying anything — the manual process is cheaper than the software plus the review time. Measuring first protects you from that mistake as much as from the other one.
Common questions
- How long does it take to create one product record manually?
- Between 20 and 40 minutes for a technical product, timed end to end rather than timing only the typing. Simple consumables sit at the bottom of that range; anything with a specification table, a compliance certificate or multiple variants sits at the top. Use 25 minutes as a working average unless you have timed your own catalogue.
- What does it cost to onboard 2 000 products by hand?
- About R125 000 a year in direct labour: 2 000 SKUs at 25 minutes each is 833 hours, at a loaded cost of R150 an hour. That figure excludes rework, knowledge loss and opportunity cost, all of which are real and none of which appear on a budget.
- Why is the hourly rate R150 and not the person's salary?
- R150 an hour is a loaded cost, not a salary: gross pay plus statutory contributions plus the share of overhead that person consumes, divided by genuinely productive hours rather than contracted ones. For a junior administrator on R15 000 a month in South Africa it is a reasonable and often conservative figure.
- At what catalogue size does automating product data stop being worth it?
- Below roughly 800 products onboarded a year. Under that volume the software plus the review time costs more than doing the work by hand, and the honest answer is to carry on as you are.
Where to next
Who wrote this
The Smart Station team
Written by the engineering and delivery team at Smart Station (Pty) Ltd, the South African software company that builds GetShopSync. The material here comes from building catalogue systems for distributors — it is what we have measured and what we have got wrong, not a survey of the literature.