Last updated on June 9th, 2018 at 12:43 pm
NEW YORK, NY — June 9, 2017. Scientists propose a “visual listening in” approach to measuring how brands are portrayed on social media, by mining visual content posted by users.
With the rapidly growing amount of visual brand-related content consumers create on
social media, images are a promising tools for marketers, designers and brand managers to track their brands‘ performance.
The study, Visual Listening In: Extracting Brand Image Portrayed on Social Media, published in SSRN by Professor Natalie Mizik of University of Washington, Liu Liu and Daria Dzyabura of NYU Stern School of Business, proposes an approach to leveraging image data by extracting scores of brand perceptual attributes expressed in the images.
Although text-mining approaches have gained popularity in leveraging user generated content for brand monitoring, image-mining approaches are still relatively new. A new scientific paper bridges the image-processing research with the branding domain by proposing an approach to online brand monitoring and market intelligence through consumer-generated images. This approach enables managers to monitor how their brands are portrayed on image-based social platforms by mining consumer-created brand images.
Academic researchers and practitioners have developed tools for social media monitoring,
to listen to consumers’ online brand conversations, and to gain insights on how consumers
perceive brands relative to their competitors’. The focus so far has been on text content.
However, given that images are on their way to surpassing text as the medium of choice for
social conversations, and the rich information of the consumption experience and feeling
conveyed through images, including visual content as part of firms’ social media efforts is
important to get a more complete understanding of brand-related online conversations.
Brand managers have long recognized the importance of creating, managing, and measuring brand image. In the first decade of the 21st century, a profound shift has occurred not only in how individuals consume information, but also in the very origins of the information itself. Corporations and the traditional media have lost their monopoly on information creation and communications channels.
Much brand-related content is now created and spread through Twitter postings, user discussion forums, social networking sites, and blogs. Firms now operate in an environment in which customers actively participate in shaping brand perceptions and co-creating brands and brand identities. With the wider, more egalitarian distribution model, monitoring how a brand is portrayed on social media is essential to effective brand management.
The study focus on consumer-created visual content, which is on the rise. With the proliferation of camera phones, good data plans, and image-based social media platforms, photo taking and sharing has become an important part of consumers’ social lives. Images are becoming an increasingly prevalent form of online conversations. Instagram users have shared over 300 billion photos to date, and share an average of 70 million photos daily.
In these shared photos, consumers often tag brands, resulting in a large volume of photos depicting brand. For example, a search of hashtag #nike in Instagram returns over 52 million photos tagged with Nike. Through these images, consumers communicate about brands with each other. By mentioning brands in their social media posts, consumers link the brand with context, feelings, and consumption experiences.
From visual content we are able to recover different dimensions of brand image, specifically those related to the nonfunctional attributes of the brand. Images non-verbally convey a lot of information about the usage situation, the setting, mood, and feeling associated with the product. For example, although a consumer is unlikely to write that he is feeling rugged today wearing his Levi‘s jeans, a photograph tagged with Levi‘s may look rugged.
A new study is related to three streams of research in the marketing and computer science
fields: user generated content (UGC), visual marketing, and machine learning methods
Mining UGC for online marketing intelligence is becoming increasingly popular in both
the marketing and computer science fields, as well as among practitioners. In the
marketing science discipline, Professor Odet Netzer proposed an approach to help firms understand market structure and monitor the topics discussed in relation to their brand by mining brand associations from consumer-generated content on forums. Dr. Culotta measured consumers‘ brand perceptions by mining the brands‘ social connections through social networks on Twitter. Several scientific papers have investigated the relationship between consumer-generated content online, for example, product reviews and ratings, and sales. The focus so far has been on text.
Leveraging image content is an important contribution to this research. By focusing only on text content, a significant portion of online conversations is not being “heard.” First, many consumers, particularly millenials, prefer visual-based online communication, such as Snapchat or Instagram, over text. As a result, analysis focusing only on text ignores many conversations. Second, for some consumer categories, including beverages and apparel, visual content is more easily available than text content.
Scientific paper also contributes to the visual marketing discipline, which has studied how
consumers perceive different visual stimuli. For decades, companies have used visual stimuli to shape consumer brand perceptions through brand logos, advertisements, retail store decoration, and, more recently, social media.
Marketing and branding theory tells us firms use brand elements, such as advertising, social media, and product packaging, to create a brand image in consumers’ minds. However, of course, the images from brands’ official accounts capture only part of the firm’s positioning efforts. For example, consider a beverage that, by the nature of the product itself, is perceived as very healthy, such as juice or bottled water. The firm may focus its visual marketing efforts on making the product look fun or even glamorous, rather than healthy.
The image-mining methods presented in this scientific paper, provide a first step in analyzing rich image data generated by consumers and firms. Future research can extend the application to analyzing how images affect consumer search behavior, learning consumer preference of product design, designing Ads targeting strategy based on consumer-posted images on social media, and so on. Given visual content is a ubiquitous part of modern life and affects consumers‘ decision making in multiple stages, being able to capture and incorporate visual content into marketing models is important.
Data sourced from Visual Listening In: Extracting Brand Image Portrayed on Social Media,Social Science Research Network; additional content by NicheHunt.com staff