Where our numbers come from

Marketing statistics in this industry rarely survive a second click. This page lists the numbers we cite on cobiro.com, what each one actually measures, and where it comes from, so you can check every figure yourself.

More images, more clicks

Our homepage says product offers with more than one image get 76% more impressions and 32% more clicks. The figure is Google’s own, published in its announcement of Product Studio: a comparison across Shopping offers with one image against offers with several. It measures clicks, not conversions, and we quote it as clicks for exactly that reason. Most catalogues still ship one packshot per product, so the gap is open to anyone who adds real secondary images. The figure has its own page in this guide, with the caveats spelled out.

For calibration, the honest conversion numbers in this field are single digits. eBay’s research across 6.8 million listings puts better photos at a 4.5% higher likelihood of selling, and Meta’s own estimate for AI image generation on its ads platform is a 7% lift in conversions. A vendor promising a double-digit conversion lift from imagery alone is quoting an uncontrolled case study, not a measurement.

What a product image costs

The 90% cost figure is a price-list comparison you can re-run in ten minutes. A professional studio packshot costs $39 to $75 at current published rates, and on-model or lifestyle photography runs $200 to $500 per shot once models, sets and retouching are in; soona publishes its rate card openly. On the other side, the image models behind Cobiro price generation at a few cents per image on Google’s published price lists. Even allowing several generations per usable image and the time to review them, a finished AI product image lands under $5, on the same day. The full breakdown is on its own page.

Complete attributes, more searches

Google’s guidance for Merchant Center is plain: providing all relevant attributes increases a product’s chances of showing for the most relevant queries. Colour, material, size, category: every field the feed leaves blank is a search the product cannot match. That is the mechanism behind attribute enrichment, in Google’s own words. The coverage measure has its own page as well.

Attribute coverage is also the rare number you can measure on your own catalogue in an afternoon: the share of products with colour, material, size and category filled in, against every product in the feed. Run the count before enrichment and after it, and you have a before-and-after nobody can argue with.

Why volume beats polish

Motion’s 2026 creative benchmarks, built on 1.3 billion dollars of tracked Meta spend across 550,000 ads, found that only 5 to 8% of ads become winners, and that plain product-image-and-text creative wins slightly more often than high-production creative. The advertisers who find winners simply test more creative, and what limits testing is the cost of each image.

Pointing the same way, NCSolutions measured roughly 450 CPG campaigns against purchase panels and attributed 49% of sales lift to the creative itself, more than reach, targeting and recency combined.

Numbers we choose not to use

• “75% of shoppers rely on product photography.” The most-quoted image statistic online traces to a trade article that names no survey, no sample and no year. Every citation leads back to the same dead end.
• “3D content converts 94% better.” A single 2020 social post comparing shoppers who chose to interact with 3D content against everyone else. That measures the shopper’s intent, not the content.
• Unnamed case studies. A “+157% conversion” with no merchant, no method and no control group is an anecdote. We would rather cite a smaller number with a source you can click.