Consumers and Their Brands in Selfies: A Question of Balance

A most popular hobby among users of smartphones is taking ‘selfie’ photo images (e.g., on trips, in restaurants, at parties), and it is furthermore popular to upload those self-images to their accounts on social media networks for sharing with others; some images are ‘joint-selfies’ (e.g., with friends and relatives). Consumers additionally share images they have shot of branded-products they relate to, and they may appear themselves in those images. A recent research article on “The Power of Brand Selfies” (2020) reveals how the form of presenting branded-products in user-generated photo images can matter by effecting differently the response behaviour of other users observing the images.

Hartmann, Heitmann, Schamp, and Netzer [*] show in their article that there is significance to whether consumers appear explicitly in selfies, or just implicitly, together with the branded-product presented in their images. The researchers identified three types of user-generated content (UGC) of photo images shared on social media which present branded-products: (1) The first type is the ‘consumer selfie‘ because it is more like the customary selfie image: the consumer-user is seen holding the branded-product with his or her face appearing in the image; (2) In the second type of image the consumer appears in a subtle, symbolic way: the branded-product is seen in a hand holding it, presumably of the consumer-user, but without seeing the consumer’s face, hence this image type is called by the researchers a ‘brand selfie‘; (3) The third image type, named ‘packshot‘, is a stand-alone display of the branded-product (i.e., with no apparent intervention of the consumer-user within the photo image).

Hartmann and his colleagues carried out their research on a little more than 250K images posted on social media platforms, primarily in Twitter (92% of post-images, ~214K), and secondarily in Instagram (8% of post-images, ~43K). Instagram was added to the research design to represent a different, image-specialised kind of social media network. Therefore, the research likens a field study of post-images published in their digital-virtual environment of the social media platform. The main part of the research applies to packaged consumer food and beverage products in a number of categories, which seems to explain also the name ‘packshot’ given to the third image type. (Note: consumers may present other products that are not packed, e.g., fashion garments and accessories, kitchenware, electronic devices; a product pack has the advantage of making the brand name or logo more visible to observers).

The large majority of images (64%) were actually classified as packshots, that is, they included a branded product alone (see table below). Second came the brand selfie images. Only around 10% of images posted were recognised as consumer selfie images wherein the identifiable face of the consumer posting the image can be seen while holding and even consuming the branded product (e.g., a soft drink). This outcome might reflect, as the researchers suggest, consumers’ reluctance to photo themselves with products. It might occur out of concern of consumers that they could be perceived as visibly promoting the branded-product. This point is of importance because it relates to a difference in how consumers view user-generated images with branded products and how the companies or brand owners view them.

Brand-Image TypeTwitterInstagramOverall
Brand Selfie29%20%27%
Consumer Selfie8%12%9%
Packshot63%69%64%
Number of Post-Images214.5K43.5K258K
Classification of images (see explanation of the methodology below)

The style of a brand selfie image, with a hand putting forward the product to viewers, usually receives little attention, almost neglected, in discussions on product images. Although this type of image is not the most common, it seems to hold a key to making greater impact on other social media users observing an image of this type compared with the two other types. It is furthermore interesting that consumers-users choose to take this kind of photo image with a branded-product — holding the camera phone in such a position in relation to one’s hand holding the product, taking the photo from a first-person perspective (they might be more serious hobbyist photographers) — findings from the research suggest that this kind of perspective is better accepted by others observing the image.

When a social media user observes a consumer selfie image of a peer user holding a product of a certain brand, we might expect that the observer would visualise oneself in place, holding and having that product. But instead of performing this kind of mental simulation, a form of self-reference thoughts connected with the product, the consumer selfie image can elicit engagement of the observer in thoughts about the other person in the image (‘sender’), that is other-thoughts. A key thesis of the researchers is that the kind of thoughts a portrait or selfie image is more likely to trigger is contingent on familiarity of the observer with the person whose portrait is displayed in the image.

On seeing the face of the sender, the observer is likely to have thoughts like how he or she relates or compares to the other person in the image, and this would inhibit self-thoughts. If the face of the person is less familiar to the image viewer, as probable in a display ad, having other-thoughts would be more difficult, because of lacking a basis for developing them, and hence would still leave room for developing self-thoughts like mental simulations. However, if the person in the image is more visibly and closely familiar to the image viewer, as is more likely for selfies among users in a social media network, the potency for developing other-thoughts about the user in the consumer selfie image would be stronger, and self-thoughts get blocked; moreover, in this case there is also a stronger drive for the user-observer to communicate with the user-sender.

Let us turn first to differences in reference of users-observers to the sender or to the branded-product, based on frequency of the common responses in social media, likes and comments. Hartmann and colleagues find that post-images with visible faces (consumer selfies) gain more likes than face-absent post-images of brand selfies and packshots. Consumer selfie images also receive more comments in response from users than packshots {Note $}. The faces in consumer selfies draw more attention and interest of peer users-observers, indicating their engagement with the user-sender rather than the brands of products. In particular, users seem more inclined to compliment the senders, not necessarily or as much complimenting the branded-products. This outcome is even more in accordance with the expectation that users in social media are more familiar with the peer (‘connected’) senders (i.e., by generating more other-related thoughts about the sender).

Yet, we should also notice that brand selfie post-images draw more likes and comments than the packshot post-images displaying stand-alone branded-products. It can be taken as a specific indication that including the user-sender’s hand presenting the product is meaningful to users, potentially generating brand engagement (e.g., by eliciting self-reference thoughts connected with the brand). Thereby, the hand in the brand selfie appears to play a contributing role.

  • {Note $} The effect of getting more comments to consumer selfies than brand selfies is less strong or ‘mixed’ — the effect is consistent with the pattern described above though not statistically significant on Twitter whereas the effect is greater and significant on Instagram.

The differences in response between the three image types indicate quite clearly that when images show the faces of the senders (consumer selfies) the observers are more likely to be engaged with the sender. Hartmann and his colleagues were still looking for more convincing evidence that when the images focused on the branded-products, users-observers were indeed more inclined to be engaged with the brand. Their measure of interest and engagement with the product and its brand was purchase intention, as could be verbally detected in comments posted by users observing the post-images (i.e., the researchers applied a language model to retrieve words {e.g., “got”, “where”, “it”} that could predict an interest in purchasing and having the branded-product displayed — see more explanation below).

The researchers found, in accordance with their expectations, that brand selfie images as well as packshot images receive more purchase intentions (through comments) relative to the consumer selfie images. Furthermore, brand selfie images have an advantage in generating purchase intentions compared with the packshot images. First, the brand selfie and packshot images, wherein branded-products take a more central place, show greater potency to drive brand engagement vis-à-vis consumer selfie images with faces seen that are more likely to drive sender engagement. Second, and importantly, even with just a subtle presence of the hand ‘serving’ the branded-product, the brand selfie image is better able to induce purchase intention than a stand-alone display of a branded-product in a packshot image. Based only on the number of likes and comments, one may miss when brand engagement truly occurs, and especially mistaking sender engagement for brand engagement. (Note: the incidence rate of image-posts that receive comments expressing purchase intention is just 5% — but where they occur, the researchers argue that these can be seen as “strongest and most favourable” reactions to be found in social media.)

Another aspect of content that can be retrieved from comments brings us back specifically to the distinction between self-reference thoughts and other-reference thoughts discussed above. In this case the text analysis of comments detects references in first-person (of the type of “I” words) and second-person (of the type of “you” or “he/she” words) by users-observers, respectively. A net-self metric measures the number of self-thought verbal expressions over and above other-thought expressions. The results confirm that post-images without faces — brand selfies and packshots — receive more self-oriented thoughts from observers than consumer selfies, suggesting that senders’ faces indeed are shifting self-thoughts towards other-thoughts.

However, brand-selfies are not found significantly different from packshots in triggering (more) self-related thoughts. That is, we cannot conclude from here that the presence of a sender’s hand encourages more net self-thoughts compared with a stand-alone display of a branded-product. Hence, the research indicates that brand selfie images are likely to induce more purchase intentions than packshots but without connecting them with self-oriented thoughts of the observer. The problem may reside, as the researchers clarify, in that their latter text analysis actually did not link the self- and other-related thoughts with brand-related thoughts (e.g., intentions). A separate lab experiment conducted by Hartmann et al. provides some remedy, and adds sense to the results. Briefly, findings from the controlled experiment (featuring a burger sandwich) supported the advantage of brand selfie images with respect to purchase intention: brand selfies yield higher stated purchase intention than packshot images as well as consumer selfies, and also support a mediating (‘underlying explanation’) role of self-thoughts versus other-thoughts aimed at the sender (inadvertently the experiment also affirms that when a sender’s face is not really familiar it fails to elicit other-thoughts).

The visible familiarity of faces is only the initial trigger, and we should follow deeper by considering the degree of familiarity with the sender as a person and closeness of sender-observer relationship in driving other-reference thoughts. It is also practically difficult to make generalizations about differences in familiarity between display ads and social media networks. On the one hand, the product presenters in ads can be public figures or celebrities (‘endorsers’) on whom consumers know more, even if not knowing them in person, versus unfamiliar, anonymous model presenters (results of the analysis with display ads are not cited here). On the other hand, users in social media networks connect with others whom they may know only by face and scantly from the social media environ or more closely as friends, relatives and colleagues from the physical world; we may distinguish at least between connections as ‘friends’ and ‘followers’ to represent more distinct levels of familiarity.

  • A deep learning technology and methodology (convolutional neural networks, CNN) was utiliised for classifying post-images according to the three identified image types. Hartmann and colleagues actually employed an existing generic CNN (ImageNet) model for visual recognition of objects in images, and updated (re-trained) it with some of their social-media-originated user-generated images to adapt it for more accurately classifying the images in their research dataset.
  • A natural language processing (NLP) methodology was applied for analysing text in comments of observing users (as well as captions added by senders to their images), aided by language models.
  • The researchers estimated regression models (linear or generalised-linear) with likes, comments, purchase intent, and net self-thoughts as dependent variables and the three image types as an independent variable, with additional controlling variables describing images (e.g., place and size of brand logos) and senders (e.g., frequency of posting, numbers of friends and followers). Analyses of post-images in Twitter and Instagram were performed separately.
  • A lab experiment helped in verifying (validating) the interpretation of expressions of purchase intention in comments, and also that their verbal predicates of intention correspond pretty well with stated purchase intention (i.e., based on a Likert-type scale), in addition to the experiment per se.

There is a certain tension between the goals and interests of companies and consumers with respect to images that the latter create and publish on social media. Companies are interested in having consumers shown closely together with their branded-products, and thereby encourage publishing the consumer selfie type of image. Consumers-users on social media are rather more interested in drawing positive attention through likes and comments towards themselves, more than towards the branded-products. They may also feel less comfortable about appearing as promoters (brand advocates, especially if paid, are the exception, but are more difficult for companies to reach). Yet, the research demonstrates that brand selfies can have an advantage over consumer selfies even from the perspective of the companies in attracting the users-observers of images to their brands. Therefore, companies need not push consumers in social media to share photos that show their faces when presenting the branded-product.

A few more factors are proposed hereby that could make the brand selfie images more inviting for consumers-observers to obtain the displayed branded-product for themselves. The sight of the hand holding the product may drive a stronger desire of consumers with greater need for touch to hold the product themselves — the brand selfie highlights the hand gesture compared with the packshot image and avoids ‘distractions’ in the consumer selfie (e.g., the sender’s face). It could be merely a body gesture of the user-sender like ‘handing over’ the branded-product (e.g., putting the product on one’s palm and stretching it forward) that is more inviting for his or her observing peers. From a graphic-visual perspective, the hand holding or grasping a branded-product may function as an arrow pointing at the product and drawing attention to it. The user-sender’s hand may also appear as an agent ‘serving’ the product and thus facilitating an urge of the observer to approach the branded product presented.

The research of Hartmann, Heitmann, Schamp & Netzer offers an intriguing insight: A brand selfie image with a hand ‘serving’ the branded-product can have greater impact in driving brand engagement than a packshot image with the product displayed alone or a consumer selfie image that draws more attention to the face of the sender presenting the product. Greater brand engagement may be delivered through driving more self-reference thoughts of the user-observer, particularly if connected with the focal branded-product. Additionally, the researchers point out that relying only on frequencies of likes and comments can provide a partial and misleading picture; marketers should look into the comments in social media to better understand if and how their brands gain engagement of consumers-users.

Ron Ventura, Ph.D. (Marketing)

Reference:

[*] The Power of Brand Selfies; Jochen Hartmann, Mark Heitmann, Christina Schamp, & Oded Netzer, 2021; Journal of Marketing Research, 58 (6), pp. 1159-1177. (Accessed for reading online at sagepub.com)

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.