03:34    Silver Push

---

How Computer Vision is Transforming Marketing?

Computer vision is a powerful and revolutionary technology. Its use cases have emerged in various industries including retail, automotive, healthcare, defense, agriculture, banking and marketing.

Computer vision provides exciting capabilities to marketers. Computer vision enables computers to acquire visual understanding just like human beings use their eyes and brains to understand their surroundings. Computer vision works by scanning images and videos, and translating their contents into metadata. This data can be then organized and used by the marketers in varied ways.

Computer vision is transforming marketing in several ways. The following are some of the ways through which computer vision is reshaping the marketing landscape and providing awesome opportunities to marketers –

Content Generation Through Generative Adversarial Networks (GANs)

Creation of new content is one of the greatest challenges faced by online marketers. Generative Adversarial Network (GAN) is an approach to generative modelling, which implies that these networks can be used to generate or create new visual content.

A Generative Adversarial Network is trained using a generator model and discriminator model. The generator model learns to generate new examples, whereas the discriminator model learns to differentiate examples as either real or fake.

GANs can be used by marketers to create visual content such as videos, images and three-dimensional models. For example, GANs can be used to generate an image of a person in different poses. This is done by providing data on different poses to the system. GANs have also been used to generate realistic photos of fake fashion models. These examples show how GAN can help marketers easily create original visual content through computer vision.

Some of the other uses of GANs include generation of realistic photographs of human faces, creation of cartoon characters, image-to-image translation such as translation of photos of day to night or of summer to winter, translation of text to image, etc. Marketers can use GANs to create visual content at massive scale so as to fully meet their marketing requirements.

Product Search by Using Visual Similarity

Computer vision allows consumers to search a product without typing keywords or product-tags in the search bar. A buyer uploads an image of a product. The visual search technology shows him/her the same and similar products on the screen.

This computer vision powered visual search technology makes product finding easy and fast for consumers, and improves their shopping experience. This technology facilitates online shopping and leads to increase in sales.

Contextual Advertising

Computer vision-powered contextual advertising delivers very high degree of context relevance. Computer vision enables detection of contexts in video such as faces, emotions, logos, objects, scenes and activities with very high accuracy, allowing marketers to display in-video ads that are in line with the content the user is actively engaging with.

Computer vision powered contextual advertising is a boon for marketers as it allows them to achieve unprecedented reach and user engagement; not achievable by traditional methods of contextual advertising.

Understanding Images Through Screen Graphs

By using scene graphs, it is possible to retrieve images by describing their contents. For example, by using this technology, one can enter “a man wearing a black hat and sitting in a white car” in the search bar and all the relevant images will appear in the search results. Current image retrieval systems that do not deploy scene graphs cannot handle such complex queries.

Scene graphs represent objects, attributes of objects and relationships between the objects in images. The scene graph-based technology is a new field of applied computer vision. For marketers, this technology provides benefits such as contextual image generation and intelligent auto-captioning. Scene graphs enables marketers to make data-driven decisions around visual content.

Brand Safety

Computer vision-powered brand safety solutions enables marketers to effectively avoid ad placements against brand unsafe content. These solutions work by detecting harmful or unsuitable contexts in streaming videos. Computer vision offers a tailored approach to brands by enabling them to custom define unsuitable contexts, thus allowing them to run their ads in a truly brand suitable environment. Computer vision also overcomes the problems of content under- and over-blocking inherent in the traditional brand safety methods.

By integrating computer vision into their marketing efforts, marketers can easily generate visual content at scale, gain deep insights, enhance customer experiences, facilitate consumers’ buying, run highly successful campaigns, and ensure true brand suitability.

Read More

Using Computer Vision for Effective Visual Content Strategy

Using Computer Vision for Effective Visual Content Strategy

  02:46    Silver Push

---

Visual formats such as images and videos are embraced by people over just a plain piece of text. Images and video enable brands to bring life to their messages, making consumers better understand their products and services. For brands, an effective and strong visual content strategy drives engagement and sales.

Research shows that brands are using visual formats much more on their own platforms and their social media pages for conveying messages to consumers, but less frequently in display ads.

But what is causing marketers to give less preference to display ads when it comes to using highly effective content formats - images and videos - for communication with the consumers? Research shows that using their own platforms allow them to exercise more control over their visual content in comparison to putting it out on the uncontrolled internet in the form of ads. There is enormous competition and it is hard for marketers to ensure that they are reaching their targets and drawing user engagement.

Another reason that marketers cite is of brand safety. Enormous amount of content is uploaded on the internet on daily basis and marketers have no idea against what content their ads would get displayed. On their own platforms, whole content is under their control.

Research shows that when it comes to using visual content for increasing user engagement, raising brand awareness and generating revenue, marketers face the following issues – insufficient viewability, contextual irrelevance, and ineffective demographic targeting. Data privacy regulations such as the General Data Protection Regulation (GDPR) and the California Consumer Privacy Act (CCPA), along with the gradual phasing-out of third-party cookies in Chrome by Google, have made practices like demographic targeting all the more difficult. 

The problems that hinder the use of visual formats by marketers in display advertising, namely – insufficient control over ad placement, insufficient user engagement, brand unsafe environment and data privacy laws – can be resolved through contextual targeting.

Contextual targeting involves placement of an ad against the content that is relevant to the ad, i.e. the ad is in line with the content that the user is currently interested in. Contextually targeted ads readily capture the attention of users and increase their chances of viewing or clicking them, as it is likely that users are already interested in the products or services being advertised.

Keywords-based contextual advertising often delivers sub-optimal results as keywords fail to fully reflect the user’s current state of mind, while AI-powered solutions that utilize technologies such as NLP and semantic analysis fail to understand nuanced contexts and complex relationships that exist between words.

The true contextual targeting can only be achieved through computer vision. By leveraging computer vision, marketers can take control of their visual content strategy and use visual formats to run highly effective video advertising campaigns, without worrying about data privacy and brand safety issues.

Computer vision is an advanced technology that enables computers to understand images and videos. Computer vision uses deep learning to make computers learn how to detect patterns in images and streaming videos.

Computer vision powered contextual advertising technology works by accurately detecting contexts in streaming videos in order to display in-video ads that are in line with what the user is actively engaging with. Any content that is unsafe or unsuitable is contextually filtered out to provide true brand suitability.

Computer vision enables marketers to embrace contextual targeting and fully utilize their visual content for achieving their marketing goals.

Read More

Using Computer Vision for Effective Visual Content Strategy

Using Computer Vision for Effective Visual Content Strategy

  04:04    Silver Push

---

Visual formats such as images and videos are embraced by people over just a plain piece of text. Images and video enable brands to bring life to their messages, making consumers better understand their products and services. For brands, an effective and strong visual content strategy drives engagement and sales.

Research shows that brands are using visual formats much more on their own platforms and their social media pages for conveying messages to consumers, but less frequently in display ads.

But what is causing marketers to give less preference to display ads when it comes to using highly effective content formats - images and videos - for communication with the consumers? Research shows that using their own platforms allow them to exercise more control over their visual content in comparison to putting it out on the uncontrolled internet in the form of ads. There is enormous competition and it is hard for marketers to ensure that they are reaching their targets and drawing user engagement.

Another reason that marketers cite is of brand safety. Enormous amount of content is uploaded on the internet on daily basis and marketers have no idea against what content their ads would get displayed. On their own platforms, whole content is under their control.

Research shows that when it comes to using visual content for increasing user engagement, raising brand awareness and generating revenue, marketers face the following issues – insufficient viewability, contextual irrelevance, and ineffective demographic targeting. Data privacy regulations such as the General Data Protection Regulation (GDPR) and the California Consumer Privacy Act (CCPA), along with the gradual phasing-out of third-party cookies in Chrome by Google, have made practices like demographic targeting all the more difficult. 

The problems that hinder the use of visual formats by marketers in display advertising, namely – insufficient control over ad placement, insufficient user engagement, brand unsafe environment and data privacy laws – can be resolved through contextual targeting.

Contextual targeting involves placement of an ad against the content that is relevant to the ad, i.e. the ad is in line with the content that the user is currently interested in. Contextually targeted ads readily capture the attention of users and increase their chances of viewing or clicking them, as it is likely that users are already interested in the products or services being advertised.

Keywords-based contextual advertising often delivers sub-optimal results as keywords fail to fully reflect the user’s current state of mind, while AI-powered solutions that utilize technologies such as NLP and semantic analysis fail to understand nuanced contexts and complex relationships that exist between words.

The true contextual targeting can only be achieved through computer vision. By leveraging computer vision, marketers can take control of their visual content strategy and use visual formats to run highly effective video advertising campaigns, without worrying about data privacy and brand safety issues.

Computer vision is an advanced technology that enables computers to understand images and videos. Computer vision uses deep learning to make computers learn how to detect patterns in images and streaming videos.

Computer vision powered contextual advertising technology works by accurately detecting contexts in streaming videos in order to display in-video ads that are in line with what the user is actively engaging with. Any content that is unsafe or unsuitable is contextually filtered out to provide true brand suitability.

Computer vision enables marketers to embrace contextual targeting and fully utilize their visual content for achieving their marketing goals.

Read More

Sunlight x Mirrors – Video script

Sunlight x Mirrors – Video script

  04:00    Silver Push

---

Energetic start

Unilever recently launched Sunlight, a leading brand of laundry soap sold globally, in Egypt. With the promise of a refreshing scent that lasts up to 7 days.

However, Sunlight was a late entrant in the market, where laundry soap is a highly mature and competitive category. Resulting in intense competition from pre-established local brands, that were heavily advertising on traditional media.

Sunlight found Digital media to be a highly relevant yet underutilized channel for advertising laundry soaps in Egypt.

With nearly 35% of Sunlight’s target audience present on social platforms, Sunlight chose YouTube to create brand awareness within the modern Egyptian woman, seeking more than just clean clothes from her soap. 

Slight pause, energetic start

The central idea was to not only reach the most relevant audience, but also reach at a moment they were most likely to engage.

However, advertising campaigns on YouTube are deployed only basis past searches and keyword mapping, resulting in random ad placements, and ad waste.

Hence, Sunlight turned to a contextual advertising strategy, which could go beyond demographics and affinities, to effectively reach the target audience creating higher engagement.

Slight pause, energetic start

This is where MIRRORS by SilverPush brings in the magic touch of artificial intelligence. 

MIRRORS, the world’s first in-video context detection platform, uses deep learning to identify scenes, logos, faces, objects and actions in a streaming video, to serve ads only against contextually relevant videos.

Mirrors targeted videos featuring celebrities (pause), competing brands (pause), allied products (pause), and actions.

Sunlight precisely aligned brand communication with this content, enhancing customer experience and engagement in the moment.

Mirrors further optimized the campaign for most responsive audience buckets.

Slight pause, energetic start

The campaign showed unprecedented results, with overall 2.3M views, 52% higher view rates and 28% higher click through rates compared to generic targeting.

Real-time optimization for audience buckets created significant results for contexts like celebrity faces, mud and dirt, competition, and complementary products.

Beyond outreach, the campaign helped Sunlight gain 2% market share in a highly saturated market in its first month.

Ad Recall Increased by 31.5% & Brand Awareness by 10.3%. Overall Sunlight over achieved the targeted business results by 21%, marking a huge success.

Read More

Impact of Prevailing Notions and Culture on Brand Safety

Impact of Prevailing Notions and Culture on Brand Safety

  05:15    Silver Push

---

Brand safety is a relative concept that changes with time and depends on the prevailing notions and culture. Today, in the digital times, changes in people’s thinking and cultural norms occur at a much faster rate, and therefore, brand safety is much more frequently affected.

Brands today advertise against different types of websites such as news sites and video hosting platforms. The content on these sites reflect the current notions. This poses a serious challenge to brands, as brands do not know against which content their ads may appear. For example, an ad of a car having a huge diesel engine may appear on a news website or YouTube against a story or video talking about the negative effects of climate change on the environment. A decade ago, this might not have posed any problem for the car brand, but in today’s time, climate change is regarded as a grave issue and consumers will form a negative impression of the brand.  

For brands, #metoo movement surfaced as a shift in the cultural norms that cast a huge impact on their advertising strategy. Brands refrained themselves from placing their ads next to the movement related content. Cultural brand safety norms vary from region to region. For example, brands in the United States are susceptible to opposing political views, while those in the Europe are vulnerable to fascist content. Thus, brands are required to devise their brand safety strategy according to the region in which they want to promote their products.

Research shows that consumers think that it is the responsibility of brands and their agencies to ensure their ads do not appear against harmful or inappropriate content. Brands should devise a proactive and elastic brand safety strategy that can keep them protected from changes in cultural norms that are bound to keep occurring.

Brands should continuously pay attention to real-world cultural, social and political events along with evolving behavior of consumers to keep themselves abreast of negative content and trends. This will save them from getting entangled in a brand safety crisis. Brands should pay attention to both local and global happenings in order to run successful cross-border campaigns.

A robust brand safety strategy would be that which can effectively deal with changing cultural norms, changing notions of people, and differences that exist across country lines. Brands should use a brand safety solution that works in real-time, provides a very high degree of context relevance while detecting harmful or inappropriate contexts, and offers custom controls.

Blanket exclusion methods based on keyword blacklists have limitations such as content under- and over-blocking, and lack the flexibility to offer custom controls to brands. AI-powered solutions have emerged that focus on providing high context relevance, but those based on machine learning, natural language processing, and semantic analysis are unable to truly understand the sub-text, nuanced contexts, and complex relationships between words.

True context relevance is only offered by brand safety solutions that make use of computer vision. Various contexts in online videos such as such as faces, objects, logos, on-screen text, emotions, scenes , and activities can be accurately detected by leveraging computer vision technology. By using a computer vision powered brand safety solution, brands can effectively get rid of harmful or unsuitable content, and ensure a true brand-suitable environment for achieving their video advertising goals.

To effectively deal with different notions and cultural norms in different times and regions, brands should adopt a fully context-relevant brand safety solution. 

Read More

New Research Highlights the Importance of Brand Safety

New Research Highlights the Importance of Brand Safety

  02:13    Silver Push

---

Placement of ads in a brand unsafe environment tarnishes a brand’s image, weakens consumer trust, and results in decrease in revenue. Ensuring brand safety is very important for brands, whether small or big. Like previous research, new research also fully backs this statement.

A recent survey conducted by GroupM, which included fourteen-thousand consumers in twenty-three countries, has shed light on the concerns that consumers have about digital marketing and advised important considerations for digital marketers. According to the survey, more than six in ten (64%) consumers would have a negative opinion of a brand displaying ads against inappropriate content. 37% respondents, i.e. over one-third of respondents, found digital ads to be highly intrusive.

75% of the survey respondents believed that the responsibility to stop harmful or inappropriate content from appearing rests with the digital platforms. They said that proactive steps should be taken by the marketers in order to make sure that parameters are set around ad placements for creating marketing effectiveness and affording protection to brand value.

The survey report found that the trust of consumers in digital marketing is less than expected. The trust factor is very important as brand value is directly correlated with the consumer trust. For brands, it means they should work on building a responsible digital marketing ecosystem that does not dampen consumer trust in brands. For keeping consumer trust intact, a brand should take measures to prevent ad placement against any type of harmful or unsuitable content.

The findings of the above survey are consistent with those of the survey conducted by the Trustworthy Accountability Group (TAG) and Brand Safety Institute (BSI) in 2019. The survey was conducted among the US consumers and included over one-thousand respondents. 90% of the respondents said that ensuring non-placement of ads against unsafe or inappropriate content is very or somewhat important for advertisers.

Over 80 percent said that they would reduce buying or would entirely stop purchasing a product, which they buy regularly, if in case, it is advertised against extreme or dangerous content. 90% respondents said that they would decrease their spending on the product advertised next to the content involving terrorist recruiting videos, while 67% said they would completely stop purchasing it. 70% respondents held advertisers responsible for ensuring ads do not run against unsafe or inappropriate content, while 68 percent held ad agency responsible.

To prevent placement of ads against unsafe content, majority of brands make use of traditional brand safety solutions such as keyword blocking and whitelisted channels. These solutions often limit the reach of the campaigns and hampers monetization.

AI-powered brand safety solutions have appeared in the market, but those dependent on machine learning, NLP and semantic analysis fall flat when it comes to comprehending the sub-text, nuanced contexts and complex relationships words have in written or spoken language.

The innovative AI-powered brand safety solutions that make use of computer vision offer unparalleled context relevance and overcome the limitations of other brand safety methods such as content under and over-blocking. Computer vision enables accurate detection of contexts in online videos such as faces, on-screen text, emotions, logos, objects, scenes and activities. Computer vision powered brand safety solutions contextually filter out harmful or unsuitable content, providing a truly suitable environment to brands for video advertising.

With research continuously backing the importance of brand safety, adopting an effective brand safety strategy is a must for brands.

Read More

Mirrors Safe Goes Beyond Safety to Offer Brand Suitability

Mirrors Safe Goes Beyond Safety to Offer Brand Suitability

  01:22    Silver Push

---

Programmatic advertising has made brands vulnerable to the risk of damage to their image. Placement of ads against unsuitable or harmful content can negatively impact the perception of a brand in the minds of consumers, which in turn can lead to decrease in sales.

Silverpush’s Mirror Safe is not only a highly effective brand safety solution, but a full-fledged context relevant brand suitability platform. It allows brands to run their video advertising campaigns in the most brand suitable environment without killing reach.

Mirrors Safe, powered by AI and computer vision, is trained on millions of pieces of visual content. It accurately detects brand unsafe contexts in video, preventing placement of ads against such content. It effectively overcomes the limitations of keyword and NLP-based blanket exclusion technologies such as content under- and over-blocking.

Mirrors Safe offers a tailored approach to brands by allowing them to custom define the harmful contexts unique to them. This brand-specific approach helps brands to move beyond just brand safety to a truly brand suitable environment.

Features of Mirrors Safe

Features of Mirrors Safe can be summarized below –

Contextual filtration of unsafe content

Mirrors Safe accurately detects contexts like faces, objects, on-screen text, logos, emotions, activities and scenes within a streaming video, and contextually filters out harmful content across an extensive set of brand unsafe categories. It allows exclusion of unsuitable contexts custom defined by a brand.

Brand suitability score

Mirrors Safe makes use of an advanced algorithm for the calculation of a comprehensive brand suitability score. This comprehensive score takes into account five parameters. This score measures safety and suitability of the content, page and channel. The five parameters are -  

•                Engagement: likes, dislikes and participation that the content generates

•                Safety: exclusion through in-video context detection, on-screen text, and audio sentiment analysis

•                Influence: organic influence that channel/page/content creates

•                Relevance: how relevant is the content in terms of its peer channel/page category

•                Momentum: consistency that channel/page maintains or grows in terms of engagement

Complete control over ad placement

Mirrors Safe ensures absolute brand safety and suitability by predicting and controlling each and every video ad placement before serving an ad impression.

Thorough analysis

Mirrors Safe allows marketers to have an accurate pre, mid and post campaign analysis. It provides sequence depth charts so that marketers can view predetermined parameters in the dashboard.

Mirrors Safe’s post-campaign analysis provides marketers useful information beyond where the in-video ads were placed. Mirrors Safe enables marketers to measure performance of their past video advertising campaigns by identifying ad placements across harmful or unsuitable content.

By identifying content that is brand safe in their current blocklists, marketers can enhance the performance of their campaigns by improving accuracy and reach.

Mirrors Safe optimizes advertising campaigns in real-time, and ensures safe and measurable ad placements. By using this brand suitability platform, marketers are saved from compromising between performance of campaigns and brand safety. They can achieve maximum reach while having absolute control over ad placements.

With no content under- and over-blocking along with custom defining of harmful contexts unique to each brand, Mirrors Safe offers an exceptional brand suitability platform to marketers.      

Read More