Photography in marketing is often sold with big numbers and little proof. Therefore, this article keeps only what we could trace to a named source: two usability studies on how people look at images, three rules on retouching, and the EU’s labelling duty for AI-generated pictures. For each, we say what the source shows and what it does not show.
However, the earlier version of this page was a 4,000-word guide full of general claims about neuroscience, conversion rates and ROI. We could not verify them, so we removed them. As a result, this version is shorter, but you can check every point.
Photography in marketing at a glance
| Topic | What the source reports | Source |
|---|---|---|
| Which images people look at | Informative images get attention; decorative stock images get ignored | Nielsen Norman Group, 2010 |
| Images on product pages | 56% of testers looked at product images first | Baymard Institute, 2018 |
| Retouched models in France | A “Retouched photographs” label, from 1 October 2017 | Osborne Clarke summary |
| Retouched advertising in Norway | A label on edited body or skin, from 1 July 2022 | Norwegian Consumer Authority |
| Beauty filters in the UK | Filters that exaggerate a product’s effect can mislead | ASA and CAP, 2021 |
| AI-generated images in the EU | Disclosure duties under Article 50, from 2 August 2026 | artificialintelligenceact.eu |
What eye-tracking says about images
First, Jakob Nielsen published “Not All Images Are the Same” in 2010, drawing on Nielsen Norman Group eye-tracking studies. He reports a dramatic gap: users scrutinise informative images and skip decorative ones. For example, on a FreshBooks team page, one test user spent about 10% more time on the staff portraits than on the bios, even though the bios took up far more space.
By contrast, users ignored a stock-style photo on a Yale School of Management application page as “pure filler.” On an Amazon television category page, photos drew only 18% of viewing time, while text drew 82%. In addition, Nielsen adds that product photos on product pages attract even more attention, and users often click for enlargements.
Two limits apply. First, the article gives no participant counts or detailed methods. Second, these are 2010 desktop pages. Therefore, treat the finding as a design principle, not as a conversion statistic.
Photography in marketing and e-commerce: images come first
Next, Baymard’s 2018 article by Edward Scott reports that 56% of its testers looked at product images first when they reached a product page, even when the description was the faster route. However, images cannot carry everything. Details such as scent, weight or weather resistance are hard to show in a photo, and Baymard found that captions or small graphics on images helped people spot features. It also reports that 52% of sites do not add descriptive text or graphics to their top-selling product images.
Still, Baymard does not state how many participants or sites it tested in that article. Consequently, the percentages describe its own sample, not all shoppers.
The punctum: a lens for reading ad photographs
Meanwhile, Roland Barthes separated the studium, a general cultural interest in a photograph, from the punctum, a detail that pricks the viewer, in Camera Lucida (1980). Marketers sometimes borrow the idea to talk about a telling detail in a campaign image. It is a way of reading pictures, not a metric, and Barthes did not write about advertising conversion. Our article on Camera Lucida explains the terms in more depth.
Photography in marketing and retouching law
Three jurisdictions regulate how photography in marketing may use editing, and they differ in scope.
- France: According to Osborne Clarke, commercial photographs in which software altered a model’s body must carry the label “Retouched photographs” from 1 October 2017. The rule excludes TV commercials, and fines reach €37,500, multiplied by five for organisations.
- Norway: The Norwegian Consumer Authority says advertising must carry a label from 1 July 2022 when retouching or filters change a person’s body shape, size or skin. Moreover, influencers who benefit from promoting a product are covered. Makeup, lighting and general brightness edits do not need a label, and the Authority advises advertisers to keep their original files.
- United Kingdom: On 11 February 2021, the ASA and CAP applied the existing cosmetics rules to beauty filters. Two rulings on tanning products found that filters which darkened skin likely exaggerated the product’s effect. The advertiser must show that a filter does not mislead.
Because these rules differ, a global campaign needs a local check. This is general information, not legal advice.
AI-generated images and the EU labelling duty
Article 50 of the EU AI Act applies from 2 August 2026. According to artificialintelligenceact.eu, organisations that use AI to create deepfakes must disclose that the image is artificially generated or manipulated. The rule covers AI images that resemble existing people, objects, places or events and would falsely appear authentic. Providers have a separate duty to mark outputs in a machine-readable way.
Furthermore, the disclosure must come no later than first exposure, and hidden footers or faint labels do not qualify. For artistic, creative, satirical or fictional works, the duty shrinks to an appropriate disclosure that does not spoil the work. The same source says clearly fantastical content falls outside the definition, citing draft guidelines. Therefore, check the final guidance before you label, or skip a label, on a campaign image. Our AI photography facts and stats article covers the wider debate.
A short history: from the Kodak ad to the billboard
In fact, photography and advertising grew up together. The Tekniska museet dates a Kodak advertisement from the Eastman Company of Rochester to 1891. Its slogan, “You press the button, we do the rest,” sells the ease of taking pictures rather than the quality of the lens.
Likewise, photographers also documented advertising itself. Around 1926, Stuart Thomson photographed a lit Stevenson’s Bread billboard in Vancouver for the billboard company Duker and Shaw Ltd.
What we could not verify
For example, we removed several kinds of claims from the old article: conversion-rate and ROI figures, statements about neuroscience, the “average person scrolls hundreds of images” claim, and discussions of Cindy Sherman and museum influence that did not concern marketing. Instead of replacing them with unsourced numbers, we left the gaps. If you need campaign data, use your own analytics.
Sources and limits
- Nielsen Norman Group: Not All Images Are the Same (2010)
- Baymard Institute: Include Descriptive Text or Graphics for Some Product Images (2018)
- Osborne Clarke: French label for retouched photographs (secondary source)
- Norwegian Consumer Authority: Guideline on labelling retouched advertising
- ASA: The (mis)use of social media beauty filters (11 February 2021)
- artificialintelligenceact.eu: A practical guide to Article 50
Overall, we rewrote this article on 10 November 2026. Our research tools return summaries, so we did not read the French decree, the Norwegian act or the AI Act text itself. Also, the AI Act guide is a secondary source and refers to draft guidelines. The three AI-generated illustrations are marked in their captions, and the two historical images come from Wikimedia Commons.