Hasil untuk "Advertising"

Naomi Wolf

Chenfei Li, Hantao Zhao, Weixi Yao et al.

Optimizing reranking in advertising feeds is a constrained combinatorial problem, requiring simultaneous maximization of platform revenue and preservation of user experience. Recent generative ranking methods enable listwise optimization via autoregressive decoding, but their deployment is hindered by high inference latency and limited constraint handling. We propose a constraint-aware generative reranking framework that transforms constrained optimization into bounded neural decoding. Unlike prior approaches that separate generator and evaluator models, our framework unifies sequence generation and reward estimation into a single network. We further introduce constraint-aware reward pruning, integrating constraint satisfaction directly into decoding to efficiently generate optimal sequences. Experiments on large-scale industrial feeds and online A/B tests show that our method improves revenue and user engagement while meeting strict latency requirements, providing an efficient neural solution for constrained listwise optimization.

Kolagani Paramahamsa

We investigate a seller's revenue-maximizing mechanism in a setting where a desirable good is sold together with an undesirable bad (e.g., advertisements) that generates third-party revenue. The buyer's private information is two-dimensional: valuation for the good and willingness to pay to avoid the bad. Following the duality framework of Daskalakis, Deckelbaum, and Tzamos (2017), whose results extend to our setting, we formulate the seller's problem using a transformed measure $μ$ that depends on the third-party payment $k$. We provide a near-characterization for optimality of three pricing mechanisms commonly used in practice -- the Good-Only, Ad-Tiered, and Single-Bundle Posted Price -- and introduce a new class of tractable, interpretable two-dimensional orthant conditions on $μ$ for sufficiency. Economically, $k$ yields a clean comparative static: low $k$ excludes the bad, intermediate $k$ separates ad-tolerant and ad-averse buyers, and high $k$ bundles ads for all types.

Andreas Stoller

This study assesses the impact of tobacco billboard bans on smoking in Switzerland, exploiting their staggered adoption across regions, i.e., the cantons. Based on retrospective smoking histories from the Swiss Health Survey, a panel of individuals' annual smoking status is reconstructed, containing more than one million observations from 1993 to 2017. Estimation relies on staggered difference-in-differences as well as a complementary latent factor model, which relaxes the common trends assumption. The findings indicate that tobacco billboard bans lead to a reduction in smoking rates. Reductions of up to 0.9 percentage points correspond to an approximate 3% decline in the smoking rate. The effect is driven by women and individuals aged 25-44 and 65+. Overall, this evidence suggests that even partial tobacco advertising bans, such as billboard bans, can effectively reduce smoking rates and serve as a valuable policy tool within comprehensive tobacco prevention strategies.

Nosacheva, Marina I.

The object of the study is pragmatonyms-names of dental products. As the subject of study the word-formation aspect is considered. The aim is to identify the ways of how the pragmatonyms are formed, taking into account their specific nature – the intermediate position between consumer goods and medical goods. In the course of research it was established that there is a correlation between the special position of dental pragmatonyms among other products, on the one hand, and medical products, on the other hand, and the specific nature of the ways of their formation. Thus, lexical-semantic (onimization and transonimization) and lexical-syntactic methods (formation of pragmatonyms – word groups and sentences), typical for the formation of pragmatonyms-names of other groups of goods, as well as other advertising names are widely used. On the other hand, the proximity to the medicinal products determines the recourse to productive models of medical terminological formation, which predetermines the dominance of compounding and clipping, including the use of Greek and Latin components. Among the identified groups of methods, morphological (26.4%), lexical-syntactic (22.5%) and morphological-syntactic (17.1%) methods for the formation of pragmatonyms predominate; 12,5% of the studied units are formed by the complex ways (lexical-semantic or lexical-syntactic combined with one of morphological, morphological-syntactic ways). Pragmatonyms formed by special means – numbering, using graphic or mathematical symbols – are few, their share is 1,0%. Pragmatonyms with opaque internal word-form (5.0%) comprise a separate group because of the inability to determine the way of their formation.

Cristina Perales-García, Reinald Besalú, Laura Pérez-Altable et al.

The Spanish general election held on July 23, 2023, was significantly influenced by the discourse on feminism and gender-based violence, issues that all the political parties were determined to address either directly or indirectly. This research paper explores the electorate’s reception of the feminist agenda, evaluating the extent to which the gender perspective is integrated into political discourse and resonates with the general public and investigating the gender-specific perceptions and biases that influence their assessment of political candidates. The responses to a survey administered between July 19 and 21, 2023 to 1,590 Spanish residents aged 18 and above, demographically weighted by gender, age, and geographical location, reflect the profound influence of gender on electoral perceptions and reveal nuanced differences in the evaluation of conservative candidates according to respondents’ gender. Notably, the analysis highlights the critical role of women in moderating the advance of conservative and far-right political narratives. This study contributes to the broader discourse on gender politics by underscoring the complex interplay between feminist policies and political perceptions within the context of the 2023 Spanish general election.

Lingwei Kong, Lu Wang, Changping Peng et al.

Click-Through Rate (CTR) prediction models are integral to a myriad of industrial settings, such as personalized search advertising. Current methods typically involve feature extraction from users' historical behavior sequences combined with product information, feeding into a discriminative model that is trained on user feedback to estimate CTR. With the success of models such as GPT, the potential for generative models to enrich expressive power beyond discriminative models has become apparent. In light of this, we introduce a novel model that leverages generative models to enhance the precision of CTR predictions in discriminative models. To reconcile the disparate data aggregation needs of both model types, we design a two-stage training process: 1) Generative pre-training for next-item prediction with the given item category in user behavior sequences; 2) Fine-tuning the well-trained generative model within a discriminative CTR prediction framework. Our method's efficacy is substantiated through extensive experiments on a new dataset, and its significant utility is further corroborated by online A/B testing results. Currently, the model is deployed on one of the world's largest e-commerce platforms, and we intend to release the associated code and dataset in the future.

Joel Nitu, Heidrun Mühle, Andreas Stöckl

Autonomous multimodal language models are rapidly evolving into web agents that can browse, click, and purchase items on behalf of users, posing a threat to display advertising designed for human eyes. Yet little is known about how these agents interact with ads or which design principles ensure reliable engagement. To address this, we ran a controlled experiment using a faithful clone of the news site TT.com, seeded with diverse ads: static banners, GIFs, carousels, videos, cookie dialogues, and paywalls. We ran 300 initial trials plus follow-ups using the Document Object Model (DOM)-centric Browser Use framework with GPT-4o, Claude 3.7 Sonnet, Gemini 2.0 Flash, and the pixel-based OpenAI Operator, across 10 realistic user tasks. Our results show these agents display severe satisficing: they never scroll beyond two viewports and ignore purely visual calls to action, clicking banners only when semantic button overlays or off-screen text labels are present. Critically, when sweepstake participation required a purchase, GPT-4o and Claude 3.7 Sonnet subscribed in 100% of trials, and Gemini 2.0 Flash in 70%, revealing gaps in cost-benefit analysis. We identified five actionable design principles-semantic overlays, hidden labels, top-left placement, static frames, and dialogue replacement, that make human-centric creatives machine-detectable without harming user experience. We also evaluated agent trustworthiness through "behavior patterns" such as cookie consent handling and subscription choices, highlighting model-specific risk boundaries and the urgent need for robust trust evaluation frameworks in real-world advertising.

Henri Claude Moungui, Hugues Clotaire Nana-Djeunga, Che Frankline Anyiang et al.

BackgroundAmong the millions of mobile apps in existence, thousands fall under the category of mobile health (mHealth). Although the utility of mHealth apps has been demonstrated for disease diagnosis, treatment data management, and health promotion strategies, to be effective they must reach and be used by their target audience. An appropriate marketing strategy can ensure that apps reach potential users and potentially convert them to actual users. Such a strategy requires definitions of target end users, communication channels, and advertising content, as well as a timeline for effectively reaching and motivating end users to adopt and maintain engagement with the mHealth app. ObjectiveThe aim of this study was to identify strategies and elements that ensure that end users adopt and remain engaged with mHealth apps. MethodsA systematic search of the PubMed, PsycINFO, Scopus, and CINAHL databases was conducted for suitable studies published between January 1, 2018, and September 30, 2022. Two researchers independently screened studies for inclusion, extracted data, and assessed the risk of bias. The main outcome was dissemination strategies for mHealth apps. ResultsOf the 648 papers retrieved from the selected databases, only 10 (1.5%) met the inclusion criteria. The marketing strategies used in these studies to inform potential users of the existence of mHealth apps and motivate download included both paid and unpaid strategies and used various channels, including social media, emails, printed posters, and face-to-face communication. Most of the studies reported a combination of marketing concepts used to advertise their mHealth apps. Advertising messages included instructions on where and how to download and install the apps. In most of the studies (6/10, 60%), instructions were oriented toward how to use the apps and maintain engagement with a health intervention. The most frequently used paid marketing platform was Facebook Ads Manager (2/10, 20%). Advertising performance was influenced by many factors, including but not limited to advertising content. In 1 (10%) of the 10 studies, animated graphics generated the greatest number of clicks compared with other image types. The metrics used to assess marketing strategy effectiveness were number of downloads; nonuse rate; dropout rate; adherence rate; duration of app use; and app usability over days, weeks, or months. Additional indicators such as cost per click, cost per install, and clickthrough rate were mainly used to assess the cost-effectiveness of paid marketing campaigns. ConclusionsmHealth apps can be disseminated via paid and unpaid marketing strategies using various communication channels. The effects of these strategies are reflected in download numbers and user engagement with mHealth apps. Further research could provide guidance on a framework for disseminating mHealth apps and encouraging their routine use.

Lilyana Georgieva Karadjova

This paper focuses on installation art and its potential to employ and elaborate psychological concepts. As Claire Bishop argues, installation art has a psychologically absorptive character because it activates and immerses the viewing subject. To analyze this immersive experience, she refers to Maurice Merleau-Ponty and the premise that subject and object are intertwined and reciprocally interdependent. In this paper I refer to these registers of artistic perception in order to explore the concept of ‘excursive object’, which was introduced by the contemporary artist and theorist Peter Tzanev in a site-specific installation (Credo Bonum Gallery, Sofia, 2018). It refers to both Elizabeth Helsinger’s term ‘excursive sight’ and to Michel Foucault’s notion of ‘discursive object’. Whereas the latter is a discursive textual formation, which consists of apparent internal relations inside a statement, the excursive object of art emerges as a less perceivable configuration of elements. It does not result in a clear perceptual experience and reflects the unstable phenomenological interdependence of subject and object. Thus, Tzanev’s novel excursive objects depart from the usual modes of perception. His concept reveals the viewer’s particular experience of unstable configurations of elements in installation art. Furthermore, it could be explored as a resourceful term to describe perceptive situations in the psychology of contemporary art.

Anastasiia Konovalova, Tatiana Petrova

This study examines the process of reading polycode advertising posters, focusing in particular on the effect of a pun in the headline. The pun, or a sequence of lexical items that can be perceived as ambiguous, is contained in the headline and different meanings of this sequence are supported by the picture and text. The results of the preliminary experiment showed that advertisements with puns are rated as more attractive, original, effective and positive compared to advertisements without puns. We hypothesized that puns in the headlines increase cognitive effort in processing posters, leading to higher evaluations. The main experiment tested this and examined differences in eye movement when reading posters with and without puns. Fifty-five Russian participants viewed advertisements while their eye movements were recorded. Our results showed no fundamental differences in the general pattern of viewing advertisement posters with and without puns. We found that readers start to perceive polycode advertisements from the text and spend more time reading the text than looking at an image. These findings shed light on how attention is distributed between verbal and non-verbal components of polycode texts, and which type of poster is more effective for information retrieval at different processing levels.

Matthew A. Ladwig, Christopher N. Sciamanna, Gavin Luzier et al.

Abstract Introduction Adolescents who drop out of sport often report that it had become less ‘fun’ and ‘enjoyable’ over time. Although preadolescent sport typically emphasizes experiences of fun, emphases on competition and elite performance often dominate during adolescence. We theorized that adherence to adolescent sport might be improved if the overarching goal were to maximize repeated experiences of fun during sport and, subsequently, increase reflective evaluations of sport enjoyment. To that end, this manuscript reports on the rationale and design of the PlayFit Youth Sport Program (PYSP), as well as its preliminary feasibility and acceptability. The main objectives were to evaluate the feasibility of recruitment strategies and data collection procedures and the acceptability of the intervention. Setting An outdoor, multipurpose grass field at a south-central Pennsylvania middle school. Methods A mixed-methods, single-arm feasibility trial lasting for 8 weeks (August–October 2021) offered 3-times per week for 1-h per session. The equipment, ruleset, and psychosocial environment of the PYSP sport games were modified to reduce several of the constraints theorized to impair experiences of fun during sport and hamper reflective evaluations of enjoyment afterward. Results Eleven healthy, but sedentary adolescents in grades 5–7 completed the program. The median number of sessions attended (of 16 possible) was 12 (range = 6–13). Post-intervention, 9/10 respondents indicated that they ‘looked forward’ to the PYSP, 8/10 would recommend it to a friend, and 8/10 were interested in continuing the program. Ten of 11 participant guardians expressed interest in reenrolling their children if the PYSP were offered again. Some changes recommended were to improve recruitment via advertising the positive aspects of the program and “word of mouth” techniques, offering the program immediately following the school day, having contingencies for inclement weather, and minor changes to the sport equipment to improve the experience among the population the PYSP intends to attract. Conclusions The adjustments recommended in this preliminary work could be used to further refine the PYSP. A future efficacy trial could explore whether the PYSP may reduce attrition for adolescents who experience existing sport programs negatively by offering an alternative that better matches their unique needs and preferences.

Melissa P S Toh, Chui Yuen Yang, Phei Cze Lim et al.

BackgroundPerinatal mood disorders such as depression and anxiety are common, with subclinical symptomology manifesting as perinatal mood disturbances being even more prevalent. These could potentially affect breastfeeding practices and infant development. Pregnant and lactating women usually limit their exposure to medications, including those for psychological symptoms. Interestingly, the naturally occurring probiotic Bifidobacterium longum (BL) NCC3001 has been shown to reduce anxious behavior in preclinical models and feelings of low mood in nonpregnant human adults. During the COVID-19 pandemic, mental health issues increased, and conventionally conducted clinical trials were restricted by social distancing regulations. ObjectiveThis study, Probiotics on Mothers’ Mood and Stress (PROMOTE), aimed to use a decentralized clinical trial design to test whether BL NCC3001 can reduce symptoms of depression, anxiety, and stress over the perinatal period. MethodsThis double-blind, placebo-controlled, randomized, and 3-parallel-arm study aimed to recruit 180 women to evaluate the efficacy of the probiotic taken either during pregnancy and post partum (from 28-32 weeks’ gestation until 12 weeks after delivery; n=60, 33.3%) or post partum only (from birth until 12 weeks after delivery; n=60, 33.3%) in comparison with a placebo control group (n=60, 33.3%). Participants consumed the probiotic or matched placebo in a drink once daily. Mood outcomes were measured using the State-Trait Anxiety Inventory and Edinburgh Postnatal Depression Scale questionnaires, captured electronically at baseline (28-32 weeks’ gestation) and during e-study sessions over 5 further time points (36 weeks’ gestation; 9 days post partum; and 4, 8, and 12 weeks post partum). Saliva and stool samples were collected longitudinally at home to provide mechanistic insights. ResultsIn total, 520 women registered their interest on our website, of whom 184 (35.4%) were eligible and randomized. Of these 184 participants, 5 (2.7%) withdrew after randomization, leaving 179 (97.3%) who completed the study. Recruitment occurred between November 7, 2020, and August 20, 2021. Advertising on social media brought in 46.9% (244/520) of the prospective participants, followed by parenting-specific websites (116/520, 22.3%). Nationwide recruitment was achieved. Data processing is ongoing, and there are no outcomes to report yet. ConclusionsMultiple converging factors contributed to speedy recruitment and retention of participants despite COVID-19–related restrictions. This decentralized trial design sets a precedent for similar studies, in addition to potentially providing novel evidence on the impact of BL NCC3001 on symptoms of perinatal mood disturbances. This study was ideal for remote conduct: because of the high digital literacy and public trust in digital security in Singapore, the intervention could be self-administered without regular clinical monitoring, and the eligibility criteria and outcomes were measured using electronic questionnaires and self-collected biological samples. This design was particularly suited for a group considered vulnerable—pregnant women—during the challenging times of COVID-19–related social restrictions. Trial RegistrationClinicalTrials.gov NCT04685252; https://clinicaltrials.gov/ct2/show/NCT04685252 International Registered Report Identifier (IRRID)DERR1-10.2196/41751

Digbalay Bose, Rajat Hebbar, Tiantian Feng et al.

Advertisement videos (ads) play an integral part in the domain of Internet e-commerce as they amplify the reach of particular products to a broad audience or can serve as a medium to raise awareness about specific issues through concise narrative structures. The narrative structures of advertisements involve several elements like reasoning about the broad content (topic and the underlying message) and examining fine-grained details involving the transition of perceived tone due to the specific sequence of events and interaction among characters. In this work, to facilitate the understanding of advertisements along the three important dimensions of topic categorization, perceived tone transition, and social message detection, we introduce a multimodal multilingual benchmark called MM-AU composed of over 8.4K videos (147 hours) curated from multiple web sources. We explore multiple zero-shot reasoning baselines through the application of large language models on the ads transcripts. Further, we demonstrate that leveraging signals from multiple modalities, including audio, video, and text, in multimodal transformer-based supervised models leads to improved performance compared to unimodal approaches.

Allison Koenecke, Eric Giannella, Robb Willer et al.

Algorithmically optimizing the provision of limited resources is commonplace across domains from healthcare to lending. Optimization can lead to efficient resource allocation, but, if deployed without additional scrutiny, can also exacerbate inequality. Little is known about popular preferences regarding acceptable efficiency-equity trade-offs, making it difficult to design algorithms that are responsive to community needs and desires. Here we examine this trade-off and concomitant preferences in the context of GetCalFresh, an online service that streamlines the application process for California's Supplementary Nutrition Assistance Program (SNAP, formerly known as food stamps). GetCalFresh runs online advertisements to raise awareness of their multilingual SNAP application service. We first demonstrate that when ads are optimized to garner the most enrollments per dollar, a disproportionately small number of Spanish speakers enroll due to relatively higher costs of non-English language advertising. Embedding these results in a survey (N = 1,532) of a diverse set of Americans, we find broad popular support for valuing equity in addition to efficiency: respondents generally preferred reducing total enrollments to facilitate increased enrollment of Spanish speakers. These results buttress recent calls to reevaluate the efficiency-centric paradigm popular in algorithmic resource allocation.

Laurenz Aisenpreis, Gustav Gyrst, Vedran Sekara

Ensuring transparency and integrity in political communication on climate change has arguably never been more important than today. Yet we know little about how politicians focus on, talk about, and portray climate change on social media. Here we study it from the perspective of political advertisement. We use Meta's Ad Library to collect 602,546 ads that have been issued by US Congress members since mid-2018. Out of those only 19,176 (3.2%) are climate-related. Analyzing this data, we find that Democrats focus substantially more on climate change than Republicans, with 99.7% of all climate-related ads stemming from Democratic politicians. In particular, we find this is driven by a small core of Democratic politicians, where 72% of all impressions can be attributed to 10 politicians. Interestingly, we find a significant difference in the average amount of impressions generated per dollar spent between the two parties. Republicans generate on average 188% more impressions with their climate ads for the same money spent as Democrats. We build models to explain the differences and find that demographic factors only partially explain the variance. Our results demonstrate differences of climate-related advertisements of US congress members and reveal differences in advertising characteristics between the two political parties. We anticipate our work to be a starting point for further studies about climate-related ads on Meta's platforms.

Jessica M Rath, Siobhan N Perks, Kenneshia N Williams et al.

BackgroundMass media campaigns are effective for influencing a broad range of health behaviors. Prior to launching a campaign, developers often conduct ad testing to help identify the strengths and weaknesses of the message executions among the campaign’s target audience. This process allows for changes to be made to ads, making them more relevant to or better received by the target audience before they are finalized. To assess the effectiveness of an ad’s message and execution, campaign ads are often rated using a single item or multiple items on a scale, and scores are calculated. Endorsement of a 6-item perceived message effectiveness (PME) scale, defined as the practice of using a target audience’s evaluative ratings to inform message selection, is one approach commonly used to select messages for antitobacco campaigns; however, the 6-item PME scale often does not produce enough specificity to make important decisions on ad optimization. In addition, the PME scale is typically used with adult populations for smoking cessation messages. ObjectiveThis study includes the development of the Message Assessment Scale, a new tobacco prevention message testing scale for youth and young adults. MethodsData were derived from numerous cross-sectional surveys designed to test the relevance and potential efficacy of antitobacco truth campaign ads. Participants aged 15-24 years (N=6108) responded to a set of 12 core attitudinal items, including relevance (both personal and cultural) as well as comprehension of the ad’s main message. ResultsAnalyses were completed in two phases. In phase I, mean scores were calculated for each of the 12 attitudinal items by ad type, with higher scores indicating more endorsement of the item. Next, all items were submitted to exploratory factor analysis. A four-factor model fit was revealed and verified with confirmatory factor analysis, resulting in the following constructs: personally relevant, culturally relevant, the strength of messaging, and negative attributes. In phase II, ads were categorized by performance (high/medium/low), and constructs identified in phase I were correlated with key campaign outcomes (ie, main fact agreement and likelihood to vape). Phase II confirmed that the four constructs identified in phase I were all significantly correlated with main fact agreement and vape intentions. ConclusionsFindings from this study advance the field by establishing an expanded set of validated items to comprehensively assess the potential effectiveness of advertising executions. This set of items expands the portfolio of ad testing measures for ads focused on tobacco use prevention. Findings can inform how best to optimize ad executions and message delivery for health behavior campaigns, particularly those focused on tobacco use prevention among youth and young adult populations.

Wenjia Ba, J. Michael Harrison, Harikesh S. Nair

We present a data-driven algorithm that advertisers can use to automate their digital ad-campaigns at online publishers. The algorithm enables the advertiser to search across available target audiences and ad-media to find the best possible combination for its campaign via online experimentation. The problem of finding the best audience-ad combination is complicated by a number of distinctive challenges, including (a) a need for active exploration to resolve prior uncertainty and to speed the search for profitable combinations, (b) many combinations to choose from, giving rise to high-dimensional search formulations, and (c) very low success probabilities, typically just a fraction of one percent. Our algorithm (designated LRDL, an acronym for Logistic Regression with Debiased Lasso) addresses these challenges by combining four elements: a multiarmed bandit framework for active exploration; a Lasso penalty function to handle high dimensionality; an inbuilt debiasing kernel that handles the regularization bias induced by the Lasso; and a semi-parametric regression model for outcomes that promotes cross-learning across arms. The algorithm is implemented as a Thompson Sampler, and to the best of our knowledge, it is the first that can practically address all of the challenges above. Simulations with real and synthetic data show the method is effective and document its superior performance against several benchmarks from the recent high-dimensional bandit literature.

Zhipeng Zhang, Xinglin Hou, Kai Niu et al.

Recently, online shopping has gradually become a common way of shopping for people all over the world. Wonderful merchandise advertisements often attract more people to buy. These advertisements properly integrate multimodal multi-structured information of commodities, such as visual spatial information and fine-grained structure information. However, traditional multimodal text generation focuses on the conventional description of what existed and happened, which does not match the requirement of advertisement copywriting in the real world. Because advertisement copywriting has a vivid language style and higher requirements of faithfulness. Unfortunately, there is a lack of reusable evaluation frameworks and a scarcity of datasets. Therefore, we present a dataset, E-MMAD (e-commercial multimodal multi-structured advertisement copywriting), which requires, and supports much more detailed information in text generation. Noticeably, it is one of the largest video captioning datasets in this field. Accordingly, we propose a baseline method and faithfulness evaluation metric on the strength of structured information reasoning to solve the demand in reality on this dataset. It surpasses the previous methods by a large margin on all metrics. The dataset and method are coming soon on \url{https://e-mmad.github.io/e-mmad.net/index.html}.

Marie Bragg, Samina Lutfeali, Tenay Greene et al.

BackgroundWorldwide obesity rates have prompted 16 countries to enact policies to reduce children’s exposure to unhealthy food marketing, but few policies address online advertising practices or protect adolescents from being targeted. Given adolescents spend so much time online, it is critical to understand how persuasive Instagram food advertisements (ads) are compared with traditional food ads. To strengthen online food marketing policies, more evidence is needed on whether social media ads are more persuasive than other types of ads in shaping adolescents’ preferences. ObjectiveThis study examined whether adolescents could identify food companies’ Instagram posts as ads, and the extent to which Instagram versus traditional food ads shape adolescents’ preferences. MethodsIn Part 1, participants aged 13-17 years (N=832) viewed 8 pairs of ads and were asked to identify which ads originated from Instagram. One ad in each pair was selected from traditional sources (eg, print; online banner ad), and the other ad was selected from Instagram, but we removed the Instagram frame—which includes the logo, comments, and “likes.” In Part 2, participants were randomized to rate food ads that ostensibly originated from (1) Instagram (ie, we photoshopped the Instagram frame onto ads); or (2) traditional sources. Unbeknownst to participants, half of the ads in their condition originated from Instagram and half originated from traditional sources. ResultsIn Part 1, adolescents performed worse than chance when asked to identify Instagram ads (P<.001). In Part 2, there were no differences on 4 of 5 outcomes in the “labeled ad condition.” In the “unlabeled ad condition,” however, they preferred Instagram ads to traditional ads on 3 of 5 outcomes (ie, trendiness, P=.001; artistic appeal, P=.001; likeability, P=.001). ConclusionsAdolescents incorrectly identified traditional ads as Instagram posts, suggesting the artistic appearance of social media ads may not be perceived as marketing. Further, the mere presence of Instagram features caused adolescents to rate food ads more positively than ads without Instagram features.