Brand images online

August 2018

Anyone wanting to market a new product to the general public 100 years ago would have had only a few media channels at their disposal: The radio or television had yet to be invented, meaning that billboards, signs or posters stuck to the cylindrical advertising columns – such as those invented by Ernst Litfaß – were the only options available. In a matter of a few decades, the opportunities for advertising products had been hugely expanded. In addition to classic advertising channels such as TV, radio and print, companies are now using the Internet: TV ads can also be found on YouTube, with the links posted on Twitter or Facebook and enriched with additional, background information. And another thing that’s changed: Users of social media channels have long since moved on from being mere recipients. In fact, they are now active participants in the world of advertising. They hold their purchases up to the camera or pose with products, and in doing so they contribute – in an ideal world – to increasing brand recognition and projecting a positive image of the brand. New methods can now be applied to ascertain exactly where and in what context brand images are being shared on social media platforms. 

Just imagine the following scenario: A company launches a new product on the market. The production processes have run smoothly to completion, the shelves have been well-stocked with the product. But the management do not know where the product is being advertised or marketed. They are also clueless as to which newspapers will contain announcements about the product and what these ads actually look like. Admittedly, this scenario is far away from reality. However, in relation to user-generated content such as images on social media platforms, it certainly strikes a chord. Companies are unable to effectively track when their brands appear in photos uploaded to the internet, let alone exert an influence over how the brand is framed or staged in these images. There is, however, one advantage to this: Images uploaded by private internet users represent a kind of free publicity for marketing the brand – provided, of course, that the product is not being showcased by an official brand ambassador or influencer – and is placed in a neutral or, even better, positive context. The downside is that nobody can know precisely how widespread the presence of a brand actually is online. However, one thing is for sure: With the number of image-sharing platforms on the rise and smartphones becoming increasingly widespread, ever more brand images will be found online posted by the users themselves.

Dealing with the flood of images: Trained algorithms can help

The question is: Who is supposed to maintain an overview of all these images? Who is supposed to analyze whether a brand is viewed as being suitable or worthy of being photographed? Information on this is essential to properly estimate brand perception and to assess the penetrative power of new products or campaigns. A manual image analysis of user-generated material on social media would be a mammoth task, perhaps even never-ending. In contrast, an automated image analysis approach developed by GfK Verein and the University of Augsburg delivers speedy results. The basis for this is artificial intelligence (AI), developed with the help of IT systems to replicate human thought and behavioral patterns. These computer-supported systems are initially “padded out” with training images, which have been manually annotated in advance. This means that they have been enriched with metadata to supplement the image content. Following a test phase, the computer algorithm trained in this way is then applied to a vast array of images, for example from social media channels – and can then independently filter out those images where certain content is visible. The system is not only able to recognize logos and products, but can also recognize human emotions or scenes in the images.

Comparing brands: Which chocolate brand is photographed most frequently?

The GfK Verein recently applied this approach in a collaborative case study with a student from the Friedrich-Alexander University Erlangen-Nuremberg. As part of this study, user-generated images relating to two different chocolate brands were compared. Which brand was most often the subject of user photos? What was the purpose of these images for the users? To answer these questions, nearly 50,000 images over the course of almost three years (May 2012 to October 2015) were analyzed using the algorithm trained to recognize the chocolate products of these two brands  (For more information on the methodology, please see: Workingpaper ANALYZING THE TEMPORAL DEVELOPMENT OF BRAND-RELATED SOCIAL MEDIA PHOTOS).

One finding from the study: German-speaking social media users active on image-sharing platforms uploaded significantly more photos of both brands in 2015 than three years previously.  In this regard, the number of photos posted during the observation period rose appreciably for both Brand A and Brand B. This is above all most likely due to the fact that a growing number of people are using image-sharing portals overall, with smartphones making this increasingly convenient as well. No matter whether you’re travelling on the subway, in a lecture at university or during a break from work – you can quickly take a snap with your smartphone camera and share it with the online community with minimal effort. 

Aside from the fact that both chocolate brands are evidently being photographed more regularly on social media channels, there are also some differences in user posting behavior. In this respect, clear seasonal peaks have emerged over time for Brand A: At Easter, for example, a disproportionately high volume of images is uploaded online showing the product together with or without the users themselves, as well as potentially featuring traditional Easter time greetings to their families, friends and acquaintances. The situation is similar for Christmas and Valentine’s Day, when messages of love and affection are sent. The photos posted of Brand B remained relatively constant in terms of volume over the observed time frame – even during summer, when the hotter weather generally dampens our appetite for chocolate somewhat! In both cases these are similar products, so what is the cause of these differences? The answer can be found by taking a closer look at the range of products offered by both brands. While Brand A sells seasonal products such as chocolate Easter bunnies, chocolate Santas and special Valentine’s Day items in addition to its regular range of chocolate bars, Brand B restricts itself to classic chocolate bar variants.

The product line of brand B hardly varies over the course of the year – except for special summer and winter editions. The differing strategies used by the companies in terms of product development is evidently also reflected in the posting behavior of social media communities. Brand A has repeatedly enjoyed success in generating lots of consumer attention on particular occasions during the year. In contrast, Brand B does not achieve these peak values – but consumers instead appear to be more consistently conscious of the brand over the course of the year. That also includes the summer, a season in which chocolate sales tend to decline.

New function for brand images?

Do you remember the bright, gaudy 'Prilblumen' (flower) stickers that decorated so many cupboard doors in Germany in the 1970s? The self-adhesive transfers, which were included with washing up liquid, formed part of the “Happy Kitchen” advertising campaign and quickly attained cult status. This kind of advertising material is commonplace for many brands. For example the cuddly toy version of the Bärenmarke bear, Nivea beach balls or the Coca Cola Christmas truck. These products ensure that brands receive additional attention and often serve as a good reason for a photograph – as long as they are designed creatively. They can either be snapped together with the user in a photograph or be pictured alone. This applies to both of the chocolate brands investigated too: It is not only the product with the respective logo that is featured in pictures and selfies of Instagram users, but various advertising materials pertaining to the two brands are also visible. In general, user-generated brand images also fulfil a new function these days: They now not only serve as recommendations for delicious chocolate that friends and followers might enjoy, but also provide – particularly in the case of creative advertising materials – completely new content for posting and often show the consumers’ own lifestyles in a good light. Classic advertising campaigns, for example in the form of posters, sponsoring or brand events, often only influence the volume of images to a comparatively minor degree. However, when the traditional advertising techniques are successfully paired with social media channels, the effect can be far more increased. 

Influence on sales figures?

It was difficult for the case study to quantify the extent to which posting activities had a short-term influence on product sales. However, data obtained from the GfK Household panel was able to prove that sales figures rose in parallel with the number of photos posted. Nevertheless, in the majority of cases, this happened concurrently. For this reason, it cannot be stated with certainty whether the products were more frequently purchased on account of increased posting frequency, or if the products were photographed more often because they were simply bought in greater volume. However, one thing is for sure: The posts themselves often represent an affinity for the brand in question (see also: “Brand images: I post, therefore I love“). Additionally, the more that user-generated brand images are circulated online, the greater the recognition of that brand and – with positive photos – this also improves the brand image. Obviously, both aspects favorably influence sales.  

In future, it will probably be worthwhile to look more closely at the social media channels and observe where brand images are uploaded, as well as the context framing these posts. As more and more videos and audio data are created and posted just as rapidly as photos taken with smartphone cameras, it is feasible to think that
analyzing user-generated content could be expanded to include audio and video clips uploaded to the Internet, thereby allowing companies to control the image and recognition of their own brands in real time. After all, to quote the brand expert Prof. Karsten Kilian from the University of Applied Sciences Würzburg-Schweinfurt: “Weak brands make customer ads, while strong brands use their customers as advertising.” And in each case, companies must have profound knowledge of these adverts. 

­­­­­­­­­­­­­­­­­­Source: GfK Verein, ANALYZING THE TEMPORAL DEVELOPMENT OF BRAND-RELATED SOCIAL MEDIA PHOTOS; A case study from the confectionary industry Carolin Kaiser, Lara Enzingmueller, Rene Schallner WORKING PAPER /// NO. 6 / 2018  

Responsible for the article and contact person for queries about Compact: Claudia Gaspar. (e-mail to