Eight copies of one shelf
Counting is the cheapest competitive research there is, which is why it is the easiest to get wrong. Two catalogues can report very different totals and stock nearly the same things, because one is counting products and the other is counting rows. The unit is the whole result, and almost nobody writes it down before comparing.
Published — written and checked by hand.
At some point most sellers do the arithmetic. You look at a competitor, you count what they have published — products, listings, photos, pages — you count your own, and you come away with a ratio. Forty times. Five times. Whatever it is, it feels like a measurement, and it tends to set the plan for the next quarter.
It is a measurement. The problem is that it is usually a measurement of two different things, and nothing about the result says so. A total arrives as a single number with no unit attached, and a number with no unit will compare happily against anything.
A row is not a product
Consider what sits behind a catalogue total. The same jacket in four sizes and three colours is twelve rows and one product. The same jacket listed in two currencies, or on two marketplaces, is more rows again. A storefront that publishes each item in several languages multiplies its own total by the number of languages, without adding a single thing a buyer could not already purchase.
None of that is dishonest. Every one of those rows is real, has a URL, and needs maintaining. But if you are counting in order to answer "how much more do they sell than we do", rows are the wrong unit, and the gap you compute from them is inflated by a factor you have not measured.
This site got it wrong, in exactly this way
It is worth being concrete, and the clearest case to hand is our own. This site runs a standing comparison against the other tools in its category, and for six consecutive weeks that comparison reported one of them as publishing dozens of times more pages than we do. The number was correct. It was also the same few hundred pages, published once per language, counted once per language.
What makes this worth telling is not the error. It is that no amount of re-reading the notes could have caught it. Every figure was internally consistent, the ratio moved sensibly week to week, and the conclusion it supported — that we are behind on published volume — happens to be true. The only thing that exposed it was going back to the source and counting again with the split visible. A wrong number that behaves like a right one will survive any amount of review.
Your own count has multipliers too
This cuts both ways, and the same discipline is worth applying before you quote your own total to anybody — an investor, a marketplace, yourself.
- Variants. Size, colour, pack quantity and material each multiply rows against a fixed number of products. A catalogue of 200 products can be 2,000 rows without anything being added.
- Formats. The same photograph exported square, portrait and vertical is three files and one image. Nitaris renders photo sets at 1:1, 3:4 and 9:16 because those are different placements, not different pictures — useful to have, and not three times the catalogue.
- Locales and channels. One product listed on your own store, a marketplace and a social shop is one product. Counting it three times tells you about your distribution, not your range.
- Reissues. Last season's item relisted under a new code is a new row and an old product.
Any of these can be the right thing to count. Rows are exactly what you want when the question is about maintenance load, or about how many pages need photography, or about what a migration will cost. They are the wrong thing when the question is about range.
Write the unit down first
The fix is smaller than the problem and takes about a minute. Before you count anything, finish this sentence: one unit equals one ______. Distinct purchasable product. Distinct photograph. Published page in one language. Then count that, on both sides, and only then take the ratio.
If you cannot count the comparable unit on the other side — and often you cannot, because what a competitor publishes is not broken out the way yours is — the honest move is to record it as unmeasured rather than to use the number you can get. An unmeasured gap you know is unmeasured costs you nothing. A measured gap in the wrong unit sets a budget.
What is usually left when you do this
A smaller gap, and a more useful one. Collapsing somebody else's total to the comparable unit often cuts it by several times over, and what remains is a number you can actually plan against — because it is now denominated in the thing you would have to make.
In our own case the corrected gap is still a real gap, and still large. The difference is that it is now the size of a content programme rather than the size of a translation budget, and those are not the same project. We would have spent the next quarter solving the wrong one.
Read enough? Watch your own product.
One photo in — a cinematic product video and a full set of marketing photos out. Your first one is free.