Ad creative is everything a shopper sees and reads in an ad: image or video, headline, words and type. Peer-reviewed studies show that it works differently depending on what you sell (pleasure or function) and on the kind of ad (promotional or not), so the best ad creative is the one that fits the product. This article collects 11 studies, each with authors, journal, year, what was tested and a link to the DOI.
None of the studies was run on Meta or on Shopify. They come from laboratories, social posts, emails and TV spots. That is the reason to read them with the sample and the metric in front of you, which is how each one is reported below.
How should you read studies on ad creative?
Read each study as a hypothesis for your own ads, not as a rule. Look at three things: what was measured (attitude, purchase intent, likes, clicks, rarely sales), who was measured (students, social media users, a single company’s TV spots) and what the authors say about when the result flips.
Every figure reported here comes from the study’s own abstract or text.
Should the ad match the product: pleasure or function?
Yes: three studies point the same way, creative that matches the product’s role (pleasure or function) is judged better.
Affonso and Janiszewski (Journal of Marketing, 2023) ran laboratory experiments, a field test on ads and industry data. Structured design favored utilitarian brands, unstructured design favored hedonic ones. The study is about brand type and visual design, not about a platform.
Casado-Aranda, Sánchez-Fernández and Viedma-del-Jesús (Journal of Interactive Marketing, 2022) put banner ads in an fMRI scanner. Hedonic banners activated areas linked to reward and emotion, utilitarian banners areas linked to reasoning and control. The neural data predicted an additional 9% to 18% of online behavior beyond what people declared about the ads’ effectiveness.
Zarantonello, Jedidi and Schmitt (International Journal of Research in Marketing, 2013) analyzed 256 tests of TV spots from one multinational consumer goods company in 23 countries. In developed markets the experiential route weighed the most, in emerging markets the functional route weighed relatively more. It is a single company, so it shows a pattern, not a law.
Our take. Before you brief an ad, write one line: is this product bought for pleasure or for function? Then check that image, layout and words say the same thing. A catalog that mixes both, say a pleasure-led line and a technical one, deserves different creative for each, not one template.
What should the image or video show?
The studies on imagery say show how the product is used and felt, and give pleasure-led products motion. Four studies, four different settings.
Hartmann, Heitmann, Schamp and Netzer (Journal of Marketing Research, 2021) analyzed more than 250,000 brand posts from 185 brands. Compared with selfies showing a face, posts with a hand holding the product showed more purchase intent expressed in the comments (an estimate based on language), while face selfies got more likes. The data are observational, the authors test the mechanism separately.
Sharma and Estes (International Journal of Research in Marketing, 2024) ran seven studies plus four supplementary ones. Images that evoke a smell improved evaluations and choices, more for people with a strong need for smell input, and only when smell matters for the product. Images evoking a bad smell hurt.
Roggeveen, Grewal, Townsend and Krishnan (Journal of Marketing, 2015) compared video with a still image. Their abstract reports that, across the experiments, video increased the choice of the hedonic option by more than 79%. The result concerns hedonic products and services.
Yoon, Bang, Choi and Kim (International Journal of Advertising, 2020) ran four studies on TV spots. Slow objects, or a slowed spot, led to abstract thinking and a preference for benefit and quality appeals; fast ones led to a preference for features and price. The studies measure which appeal people prefer, not sales.
Our take. Two of these studies give you a creative brief you can write in a sentence: a hand using the product for the post, a slower video for a benefit message and a faster one for a price message. Test them as two ads that differ in that one thing.
How should the words in the ad sound?
Fewer claims, said with certainty, and framed in a way that makes the brand look powerful. The three studies below measure impressions and engagement, not purchases.
Shu and Carlson (Journal of Marketing, 2014) used scenarios with 521 students and sources with a persuasive intent. Impressions were 4.37 with three positive claims and 3.96 with four. The measure is an impression rating, not a purchase.
Pezzuti, Leonhardt and Warren (Journal of Interactive Marketing, 2021) analyzed Facebook and Twitter texts and ran two experiments. Words of certainty were associated with more engagement, through the perceived power of the brand. The moderator was an individual’s belief in power distance, not the national culture.
Pezzuti and Leonhardt (Journal of the Academy of Marketing Science, 2023) ran four studies, including field data on 53 million interactions. Messages with negations increased engagement because the brand looked more powerful, especially among people with a high need for status.
Our take. The first lever is the cheapest to test: write the same ad with three claims and with five, and keep the visual identical. The other two give you a wording variant to put against your current one when engagement is the goal.
Does the typeface of a promotion matter?
In promotional ads, one study says yes: right-slanted fonts raised intentions and clicks. Mead, Richerson and Li (Journal of Retailing, 2020) ran five experiments and an email field study. Right-slanted fonts raised intention to visit and to buy, and clicks, only in promotional ads, especially among people who look for deals.
Our take. Treat this as a rule for sale and offer creative only. For a brand story or a product-led ad, the study gives you no reason to expect the same effect.
How do you turn these studies into tests for Meta and Google?
Pick one study, change one variable and measure what the study measured. A short way to run it:
- One study per test and one change per ad: the rest of the creative stays identical.
- Write the hypothesis in the study’s own terms, with its limit: promotional only, hedonic products only, one company only.
- Match the metric: clicks for a clicks study, purchase intent or add to cart for an intent study, purchases only when the study measured purchases.
- Keep the product and audience constant, because several results flip with the type of product.
Our take. The value of the research is the reason behind each test. A test with a stated reason teaches you something even when it loses.
What happens after the click?
The ad brings the visit, the product page decides the sale. If the store behind your ads is on a platform that makes every change slow, our Shopify migration service moves it to Shopify; the whole process is in the Shopify migration guide. For the page itself, read Shopify CRO: what 21 peer-reviewed studies prove.
If you are already on Shopify and want advertising run month after month, the Data-Driven Team, our dedicated advertising and growth team for Shopify stores, manages advertising for its clients, as in the Cumini case study.
Frequently Asked Questions
What is ad creative?
Ad creative is everything a shopper sees and reads in an ad: the image or video, the headline, the body text and the typography. In paid social and search it is the part you control after targeting and budget, so it is where research on persuasion applies most directly.
What do peer-reviewed studies say about ad creative best practices?
They agree on one thing: there is no single best creative, only creative that fits the product and the goal. The studies in this article tie design, images, motion, wording and typography to the type of product (pleasure or function) and to the type of ad (promotional or not).
Do these studies apply to Meta ads?
Not directly. The studies ran in laboratories, on social posts, emails and TV spots, and none was designed for Meta or for Shopify. Use each result as a hypothesis for a test in your own account, matching your metric to what the study measured.
How many claims should an ad make?
One study of persuasive messages (Shu and Carlson, Journal of Marketing, 2014) rated impressions at 4.37 with three claims and 3.96 with four, in scenarios with 521 students. It measured impressions, not sales, so treat three as a starting point for a test.
Is video better than a still image?
For pleasure-led products, one set of experiments (Roggeveen and colleagues, Journal of Marketing, 2015) reports that video, compared with a still image, raised the choice of the hedonic option by more than 79%. The result is about hedonic products and services, not every category.
How do I turn a study into an ad test?
Take one result, change one variable, keep the rest of the ad identical, and read the metric the study used: clicks for a click study, purchase intent for an intent study. Write down the study’s limit before the test starts.
The Shopify Launch Sprint moves your store to Shopify in 14 days with our Custom AI System: a fixed calendar, a public price and 4 written guarantees.
Antonio Manitta
Founder & eCommerce Manager at NoProb Agency
For over 7 years I have helped fashion, supplements, and DTC brands scale their eCommerce. I work in long-term partnership with three brands at a time, building stable systems that compound.
About NoProb →
