Images that win the row

A Shopping result is a row of near-identical products at near-identical prices. The image is doing most of the work of the ad, against direct competitors, at thumbnail size.

The rules first

Before anything about quality, the main image has to pass. It must show the product being sold and nothing else: no promotional text, badges, borders, watermarks or overlays burned into it, no placeholder or generic graphic, and no image of a different variant.

Minimum sizes are 100 by 100 pixels for non-apparel and 250 by 250 for apparel. Treat those as floors, not targets: Shopping surfaces render large on desktop and larger on Lens.

What wins the click

• Fill the frame. The most common weakness is a product occupying a third of the image. At thumbnail size that reads as a smaller product and a less serious seller.
• Clean, even light. Hard shadows and colour casts read as amateur next to a competitor’s studio shot of the same item.
• A consistent background. White is the safe default; what matters more is that your own products look like a set rather than a scrapbook.
• The recognisable angle. Whatever angle buyers picture the product from, because that is the one they will scan for.
• Additional images. Angles, scale, detail. They do not appear in the row but they carry the click once it lands.

Variants need their own image

A blue shirt with a photograph of the red one is a mismatch: it converts worse, it generates returns, and it risks disapproval on grounds of accuracy. If your catalogue has variant-level colour, it needs variant-level images.

Where generation helps

Relighting, background removal and sharpening are corrective work on your own photograph, and they are the highest-return use of generation on a catalogue that was shot on a table. Placing the product in a scene belongs in additional images and on the product page rather than in the main image, where accuracy rules apply hardest. The product content chapter covers both.