Product Photos Per SKU: How Many Do You Really Need?
How Many Product Images Does an Article Need?
Six to twelve per article, depending on the category. Plain basics are covered by six, dresses and outerwear need eight to ten, high-priced pieces ten to twelve. What matters is not the number, though, but the variety: every image should answer a different question. Five different views do more than ten variations of the same front shot.
Two things make the question harder in practice than rules of thumb admit. Fashion systematically needs more images than hard goods, and marketplaces prescribe a minimum number of views that often dictates the requirement on its own. Both are broken down below, together with the point at which additional images stop having an effect.
Why Fashion Needs More Images Than Hard Goods
For a kettle, one good photo plus a spec sheet answers almost everything. Capacity, wattage, material and dimensions sit beside it as numbers, and the numbers are unambiguous. A second image from another angle is nice, but it rarely changes the purchase decision.
Fashion works differently, for three reasons:
- The decisive property cannot be described. How a garment falls on a body, how long it reads, how it behaves in motion — there is no number a customer can transfer to themselves. "Length 98 cm" is a correct specification that helps nobody who is not currently holding a tape measure.
- The customer has to project themselves into it. With hard goods you buy an object; with fashion you buy an idea of how you will look in it. That projection needs an image on a body, or it does not happen.
- Material and colour cannot be verified online. Structure, weight, sheen and shade decide satisfaction, and they can only be conveyed through close-ups and consistent lighting.
What follows is not a higher number for its own sake but a longer list of open questions — and each open question needs exactly one image. Shoes and bags sit between the two worlds: they need less fit information, but they do need a size reference.
Mandatory Views by Marketplace
Anyone selling through marketplaces often does not decide the image count at all. The overview below deliberately separates what is documented from what every merchant has to look up in their own partner account before listing:
| Channel | What is documented | What to check before listing |
|---|---|---|
| Zalando | Apparel: 3 mandatory images (a catalogue view plus two further views), aspect ratio 1:1.44 portrait, minimum 762 × 1100 px | Deviations by product group; Premium and Designer require five instead of three |
| Otto | Requirements live in the partner area and differ by assortment area; no publicly stable figure | Number of image slots, main image rule, aspect ratio, minimum edge length |
| Amazon (fashion) | Image slots and format rules are category-dependent and set out in the seller documentation | Slots for your own category, main image requirements, minimum size for the zoom function |
| About You | Requirements run through the partner onboarding documents | Number of required views, on-model requirement, format |
| Your own shop | No external requirement | Your own minimum standard per product group — otherwise chance decides |
Why only one hard number appears here: marketplace requirements change, sometimes several times a year and sometimes by assortment area. A figure that is correct today could be wrong in six months — and a rejected listing costs more than a look at the partner portal. The Zalando figure is listed because Zalando publishes it in its own partner documentation; even that should be cross-checked for your product group. One warning attached to it: the figure "7 views per article" circulates persistently for Zalando. It does not appear in Zalando's partner documentation, which requires three mandatory images for apparel. Where the seven comes from, and what is actually required, is in Zalando image requirements.
The practical consequence of the table is independent of any single marketplace: if you sell across several channels, do not produce for your own shop and then adapt for marketplaces. Produce for the strictest requirement in your channel mix from the outset. Seven portrait views cover considerably more ground than three square images that have to be reworked afterwards.
Diminishing Returns: When Additional Images Stop Paying
There is no reliable public figure for exactly where the curve flattens — anyone quoting one usually has it from a vendor deck. The shape of the curve is well understood, though, and it follows from the principle above:
- Images 1 to 3 answer the three basic questions: what does it look like, how does it sit, what does the back look like. Each has a large effect, because without it a question stays completely open.
- Images 4 to 7 answer the follow-up questions: material, construction, volume from the side, styling context. The effect per image gets smaller but stays clearly positive.
- From image 8 onward everything depends on whether a real question is still open. For a coat with an unusual cut and a lined interior: yes. For a plain T-shirt: no. Beyond this point another image only helps if it shows something new — not because it is the eighth.
The most useful stopping test is a question put to the next planned image: would it change anybody's purchase decision? If the answer is no, it does not belong on the product page, at best in social media. Conversely, two cases justify more images than the rule of thumb: high prices, because the purchase hurdle is higher and more questions want answering, and unusual cuts, because imagination carries less far there.
Back View and Detail as Returns Levers
So far this has been about conversion. There is a second effect of image count that gets overlooked, because it shows up in a different part of the report: images that barely move conversion can still prevent returns.
The back view is the clearest example. It rarely persuades anyone to buy — but without it a proportion of customers order simply to see what the piece looks like from behind. Those orders come back with high probability. The material close-up behaves similarly: it influences the purchase decision less than the main image, but it prevents the unboxing disappointment that reliably leads to a return.
Optimise image count on conversion alone and you cut exactly the views that act on the cost side. Which views act on returns, in what order, and how to measure the effect cleanly, is in reducing fashion return rates.
Checklist per SKU
As a target by product group, each intended as minimum coverage:
- Basics and T-shirts: 6 images
- Trousers, skirts, knitwear: 7 to 8 images
- Dresses and outerwear: 8 to 10 images
- Shoes: 8 images, one of them with a size reference
- Bags and accessories: 6 to 8 images, one of them with a size reference
- High-priced pieces and unusual cuts: 10 to 12 images
The concrete shot list that fills those numbers is, for fashion, almost always the same:
- Main image, cut out or against a neutral background
- On-model front view, fully in frame
- Back view
- Side or three-quarter view
- Material close-up
- Construction detail (seam, closure, lining, print)
- Styling or context shot
- Size or proportion reference, where the size is not self-explanatory
Start with the first three and add the rest by product group and the order is right. The most common mistake is the reverse: six beautiful context shots and no back view.
What That Volume of Images Costs
The numbers get large quickly. An assortment of 300 articles at eight images each is 2,400 finished images per season. With a service provider, as a market observation rather than a survey, e-commerce images sit at roughly EUR 15 to 25 apiece — so EUR 36,000 to 60,000. The full breakdown by service type is in product photography pricing.
That calculation is where, in practice, it is decided whether the back view gets produced at all. The generated route does not change the image list, it changes the cost logic: with GridShot each published image costs USD 1 plus the compute actually consumed, a few cents, and one run returns 16 to 25 variations in 5 to 15 minutes to choose from. For the 2,400 images in the example that is roughly USD 2,400 to 2,700. New accounts start with USD 10 of credit, enough for a test on a handful of articles.
The honest limitation: a usable source photo per article has to exist. The generated route does not replace the first shot of a product, it replaces views two through eight — and those are the bulk of this calculation.
Frequently Asked Questions
How many images do the marketplaces actually require?
Zalando requires three compliant images per fashion article — a catalogue view plus two further views — and five for Premium and Designer, at a 1:1.44 portrait ratio. Amazon works with five to seven images per fashion listing, one main image plus up to six alternates, and requires the main image to be a product-only shot on pure white. Slots and format rules are category-dependent and change, so check your own partner account before planning the shot list.
Does every SKU need the same number of images?
No. A plain T-shirt is fully explained by six frames; a coat with an unusual cut and a lined interior is not. Set a minimum standard per product group and add images for high-priced articles and complex construction, rather than applying one number across the entire catalogue.
When does an additional image stop paying for itself?
When it no longer answers an open question. The first three frames cover the basic questions and carry most of the effect; images four to seven answer follow-up questions with a smaller one. From the eighth onward, the only test that matters is whether the image would change the decision of any customer.
Do channel variants count as separate images?
For planning purposes yes, but they are produced differently. Marketplace and social versions are a downstream crop and export step from your master images, not a second shoot. Budget them as production time rather than photography, and expect vertical social formats to need a genuine reframe rather than a tighter crop.
How do you close the on-model gap without booking another shoot?
Generate that layer from the product photos you already have. It needs at least one usable source image per article; with GridShot a run takes 5 to 15 minutes and returns 16 to 25 pose variations, and you pay USD 1 per image you actually publish plus a few cents of compute. The run itself costs USD 1 too, credited toward the first image you deliver from it, so only runs you never deliver from keep their dollar.