Effective Ways for Marketers for YouTube Targeting

Effective Ways for Marketers for YouTube Targeting

  06:05    Silver Push

---

YouTube is the most popular video hosting website in the world with most extensive catalog of online videos. The number of monthly logged-in users on YouTube is about 2 billion. YouTube offers tremendous opportunities to advertisers.

YouTube provides a wide range of ad formats and varied targeting options to marketers, enabling them to effectively and easily reach their target audience. It is important for marketers to use the right YouTubetargeting options in order to drive success to their YouTube advertising campaigns. Using basic keyword and topic targeting at the start may not result in success. Below are discussed some of the hand-picked ways that you as a marketer can use for targeting on YouTube -

Custom Intent Audiences

This targeting option helps marketers in reaching new customers on YouTube on the basis of the keywords used by the users to search for products and services on Google.com. It is not necessary that these audiences have any previous interaction with your brand. These are built from users who have recently searched the keywords that were used by you for to creating your audience.

Some of the great custom intent audiences that you can test out for your video advertising campaigns on YouTube are -

·       Converting search queries - You can use your list of converting queries from your search campaigns to display video ads to users who have searched for these queries.

·       Converting keywords – These keywords are not the same as converting search queries. You can create a different audience based on these.

·       Competitor terms - These allow you to show your video ads to users that are actively searching for your competitors. 

·       Best-selling products - You can create an audience from your best-selling products. You can easily find these products from the sales report of your e-commerce platform. 

Life Events

Life Events can prove to be a great targeting option. It allows you to show ads to customers during life milestones such as starting a business, graduating from college, changing or starting a job, getting married, purchasing a house, retiring from job, etc.

These life events provide a great opportunity to brands, which offer products or services needed in these moments, to emotionally connect with consumers. For effective targeting via life events, your video creative should clearly show consumers how your brand can help them go through these milestones smoothly.

TrueView Discovery Ads

TrueView discovery ads appear on the search results and watch pages on YouTube. YouTube is a huge search engine, next only to Google.com. It allows users to conduct search for specific videos that they are interested in. TrueView discovery ads campaign is unique as a video advertising campaign as it is the only one in Google Ads that allows marketers to target just the YouTube's search results page.

Combining keyword targeting with TrueView discovery campaigns offers a great way to capitalize on user intent. By running TrueView discovery ad campaigns, marketers can take over the top spot of the search results page on YouTube.

Computer Vision-Powered Contextual Targeting

This is a highly effective YouTube targeting method that enables marketers to dramatically boost the performance of their YouTube video advertising campaigns. Computer vision enables AI advertising platforms to detect and understand contexts in online videos. Faces, emotions, logos, objects, activities and scenes in online videos can be detected with high accuracy.

By using computer vision powered contextual targeting, you can place your ad against the video content that is highly relevant to the ad, i.e. your ad is fully in line with the content the user is actively engaging with. As the ad shown matches the current interest of the user, the chances of user viewing or clicking the ad are very high.

AI contextual targeting powered by computer vision offers a very high degree of contextual relevance unmatched by other methods of contextual advertising such as keyword targeting, which fail to fully reflect the user’s current state of mind and cannot understand nuances in context.

The above-mentioned, hand-picked ways for YouTube targeting will help marketers in effectively achieving their YouTube advertising goals.    

Read More

Beyond Black and White: The True Color of Brand Safety

Beyond Black and White: The True Color of Brand Safety

  23:02    Silver Push

---

Over the past few years, a lot of brand safety issues have surfaced that have led marketers to review their brand safety measures. The current coronavirus crisis has intensified the brand safety woes of marketers, as most of the brands don't want ad adjacency to the content dealing with morbidity and mortality. 

Common brand safety methods used by marketers include blacklisting and whitelisting. Blacklisting involves avoiding placement of ads against content containing one or more blocked keywords. In case of video content, a blocked keyword is searched in topic, title, description and metadata.

Keyword-based blacklisting method is in reality not that effective as it seems to be. It is marred by under- and over-blocking of content. Research shows that because of the use of keyword blacklists, more than half of the safe stories published on the major news platforms are being incorrectly tagged as brand unsafe.

Keyword-based blacklisting method can lead to blocking of completely innocuous content. This is because it fails to comprehend the nuances in context, i.e. it is unable to understand the true context in which a keyword is used. For example, if "alcohol" is the blocked keyword, then the blacklisting method will not only tag a video featuring drunk and driving as unsafe, but will also tag a video featuring a recipe in which alcohol has been used as one of the ingredients, as unsafe.

Another problem with blacklisting is that universal blacklists cannot be created. They have to be regularly updated and modified according to the brands' requirements, current happenings and events, latest news, countries, languages and culture. There is also a requirement to tweak blacklists regularly on the basis of current safe content consumption patterns of consumers, so that increased reach for the advertising campaigns can be achieved. Overall, keyword-based blacklisting method is quite cumbersome to implement as it needs a lot of fine-tuning. With this method, content under- and over-blocking is a common problem, and this hinders marketers in getting optimal results from their advertising campaigns.

A whitelist enlists content that has been labeled as safe for ads to be placed against it. A whitelist provides a safe and trusted environment to brands to advertise within. Curating a whitelist for advertising on a video platform, for example for YouTube advertising, involves tagging unsafe content at the keyword, topic, video and channel levels. Video-level tagging helps brands to filter out unsafe videos from an otherwise safe channel; brands do not have to blacklist the entire channel just because of one or few unsafe videos.

Again, like keyword blacklists, whitelists need to be regularly updated, otherwise the campaigns will not witness an increase in reach, and brands will miss newer safe and engaging content for their ads; ads will keep displaying against the same video content enlisted in the static whitelist. 

Creation of whitelists is not an easy process; it requires a lot of curation by marketers, and is time-consuming and expensive. As the whitelisting method limits the number of videos against which ads can be placed, marketers are unable to take the full advantage of the true potential of huge video hosting platforms like YouTube. The campaign's reach gets reduced and the right audience does not get fully targeted.

The above-mentioned brand safety methods provide only suboptimal brand safety and have significant limitations. A highly effective way of ensuring brand suitability and safety is provided by contextual brand safety method that makes use of AI and computer vision. AI-powered brand safety platforms that deploy computer vision technology, provide high degree of context relevance unmatched by keyword-based methods.

Computer vision can accurately detect contexts in videos such as faces, objects, logos, on-screen text, emotions, scenes and activities. Thus, it can effectively detect unsafe or harmful contexts in videos without the risk of under- and over-blocking of content.

Amid the coronavirus pandemic, computer vision-powered brand safety platforms enable brands to selectively block ads against mortality-related coronavirus content, while allowing ad placement against positive coronavirus content. Thus, brands can safely capitalize on the news content; this is not possible with keyword-blacklists that fail to understand the true context in which the keyword "coronavirus" is being used.

By using AI-based contextual brand safety method, marketers can not only effectively block ad placement against recognized unsafe categories, but can also custom define unsuitable contexts that are unique to a brand. This helps them provide a fully suitable environment to brands for advertising.         

Computer vision enables marketers to go beyond blacklists and whitelists in order to achieve brand safety in its true color.    

Read More

Synergistic Approach to Visual Content Moderation Is Both Effective and Efficient

Synergistic Approach to Visual Content Moderation Is Both Effective and Efficient

  02:49    Silver Push

---

Enormous amount of content in the form of images, videos and text is posted on the world wide web on an hourly basis. As this content is posted by users around the globe, the nature of the content is highly heterogeneous.

User-generated content carries an immanent risk of being inappropriate, harmful, offensive, or dangerous. This content can be classified into the categories such as nudity, terrorism, hatred, child exploitation, violence, misinformation, etc. and requires strict moderation.

Content moderation is commonly achieved through human moderators. AI-based content moderation has also emerged and offers an automated way to filter out inappropriate content.

The enormous and heterogeneous user generated content cannot be moderated effectively and efficiently by using just one method of moderation - manual or automatic. The best approach is synergistic, i.e. using the combination of both human and AI moderation. Social media platforms are increasingly using the synergistic approach for achieving optimum level of content moderation.   

By using the synergistic approach for content classification and moderation, online platforms can enjoy the benefits of both human and AI moderation - the intelligence, wisdom and judgement of human beings, and the capability of AI-powered platforms to evaluate enormous amount of content in no time.         

AI content moderation platforms powered by computer vision makes image and video moderation highly efficient. Computer vision can detect faces, emotions, objects, logos, on-screen text, actions and scenes in the images and videos with high accuracy. Such platforms can determine whether the images or videos should be reviewed by a human content moderator or not. Thus, human moderators are saved from filtering out large volumes of content themselves; this also saves them from viewing mentally disturbing content in large quantities on a daily basis. They can look only at the images and videos flagged by the AI platform and make a publishing decision. The decision taken by the human moderator feeds back into the algorithm, but the reason for the decision does not.    

AI makes content moderation much easier for human moderators. By considering a number of factors, an advanced AI content moderation algorithm can calculate a relative risk score to determine if a user's post should be posted immediately after creation, reviewed before posting, or should not be posted. This relative score can then be used by human moderators while making a publishing decision.

Although AI content classification and moderation enables online platforms to hire less number of human moderators, the need for human moderation will always remain and is indispensable. Without human moderators, accurate content moderation is not possible. Only human content moderators can make decisions that lie in the gray areas of decision-making, view a user's content from a subjective perspective, understand cultural context of content, etc.

Armed with a computer vision powered video and image moderation platform, human content moderators easily identify and filter out inappropriate visual content from large volumes of user generated content posted on online platforms.

By following a synergistic approach, which involves using both AI and human moderation, online platforms dealing with loads of user generated content can achieve efficient and effective content moderation.

Read More

Which Is Better for Your Business – Manual or AI Visual Content Moderation?

Which Is Better for Your Business – Manual or AI Visual Content Moderation?

  02:55    Silver Push

---

Visual content moderation is important for businesses or brands, especially if they have to deal with a lot of user-generated visual content. Any association with inappropriate content can damage their reputation, weaken consumer trust and result in decrease in sales.

Traditionally, visual content classification and moderation has been done manually. But with the advent of AI, automated content moderation platforms have emerged. These platforms make use of computer vision and provide an effective way for image and video classification and moderation.       

Whether a brand or business should moderate visual content manually, use AI-powered automated content moderation or augment manual moderation with an automated one, depends on a number of factors. These factors are discussed here below –

Source of content

In order to build brand recognition and consumer trust, more and more brands are now allowing user-generated content on their own platforms. However, user-generated content is potentially risky and can include inappropriate matter that can be highly damaging for the brands. Although brands can dictate their content posting guidelines to users, they do not have actual control over what a user is posting. Moderating such content is a must for brands. As there are high chances of user-generated visual content being inappropriate or unsuitable, brands should opt for computer vision-powered video and image classification and moderation platform.

If in case, most of a brand’s visual content is not user-generated, but is sourced internally or from highly trust-worthy third parties, then for such a brand, video and image moderation can be performed manually by hiring human content moderators and there is a lesser need for an automated system.

Volume of content

For brands that have to deal with a good volume of visual content, especially user-generated content, manual moderation does not work effectively and efficiently. They should make use of computer vision-powered image and video moderation platforms.

AI-powered systems can tackle enormous content volume with a high degree of accuracy. Computer vision technology effectively classifies and tags visual content at scale. Such automated systems are not plagued by human errors, can work continuously unlike human beings, and their algorithms get self-trained from the data they handle.

Nature of content 

An automated AI content moderation platform can effectively filter out content such as “not safe for work” images and videos, and other forms of inappropriate, offensive or dangerous content, but it falls short when it comes to filtering out misinformation. Here, human intervention from human content moderators is required.

User-generated visual content can be highly mentally disturbing for human content moderators. Filtering out such content through automated computer vision powered content classification and moderation platform is the best way to prevent ill effects on mental health.  

Hiring a large number of human moderators is quite expensive and may not be feasible for businesses with small budgets. Also, in most of the cases, as discussed above, manual moderation is less effective than computer vision powered visual content moderation.

For brands or businesses that have to handle a large amount of user-generated visual content, computer vision-based content moderation is much better than manual moderation in terms of accuracy, effectiveness and efficiency.

Read More

Our New research shows that 90% marketers consider brand safety a serious problem.

Our New research shows that 90% marketers consider brand safety a serious problem.

  02:38    Silver Push

---

Brand marketers and Agency Heads across Southeast Asia believe ad placements across harmful content damage brand perception and result in revenue loss.

The brand safety crisis, that first caught the attention of advertisers in a major way back in 2017, is even more real today. With millions of pieces of user generated visual content added to video sharing platforms daily, brand safety has taken centre stage in the advertising world.

Like every crisis, this has also resulted in practical and workable solutions that have provided a semblance of control to advertisers in varying degrees. However, some of the most widely used brand safety measures including blocklists, whitelisted channels/pages, third-party measurement and brand safety specialists, bring along their own set of efficiencies and pitfalls. A debate that gained more weight recently as Coronavirus topped keyword blocklists, squeezing ad revenues and killing brand reach.

In In an attempt to understand how leading marketers and brands perceive and mitigate brand safety risks, we surveyed 160+ agency heads, business leads in media and brand marketers in Southeast Asia.

This survey report highlights some of the brands’ biggest challenges with available brand safety measures and a pulse on the growing importance of and readiness for brand suitability. Key highlights from the report include:

·       Video platforms offer more brand safety controls, but continue to remain brand unsafe, with Tik Tok leading followed by Facebook and YouTube. This was further solidified with another research when earlier this year Silverpush analysed ~15 million videos across video sharing and hosting platforms in SEA, and found nearly 8–9% of all content as brand unsafe: featuring violence, smoking, adult, and extremist content. Which means that 1 in every 10 video ad placements can potentially be across harmful and damaging content.

•         ~90% industry professionals believe unsafe exposure impacts brand perception negatively, and 62% believe the extent of this damage is highly negative.
•      ~60% respondents believe brand safety risks can result into revenue loss ranging from reduced buying to complete boycott of the brand
•       Blocklists and whitelists remain top brand safety measures. NLP based technologies and in-video context detection are emerging.
•       However, 60% said that using current brand safety measures result in inability to reach specific audience
•      ~63% industry professionals stated lack of customized exclusion filters that can meet unique brand needs as the most pressing brand safety challenge, highlighting the importance of brand suitability.

The report further talks about how challenges of the current brand safety measures resulted in killing reach and monetization during COVID-19. And further highlights the growing importance of brand suitability, solutions brand and agencies seek, and the emergence of AI powered context detection technology.

Access the full report here.

Read More

Contextual Targeting Enables Marketers to Deal with Unpredictable Consumer Behavior

Contextual Targeting Enables Marketers to Deal with Unpredictable Consumer Behavior

  03:17    Silver Push

---

Amid the coronavirus pandemic, marketers are witnessing a dramatic shift in the behavior of the consumers. Consumers are not behaving in the way that marketers have expected them to do. Their behavior has become unpredictable, inconsistent and erratic.

For example, consumers have stockpiled grocery items in their pantries during the lockdown period to the extent that demand has surpassed supply. Consumers have either made a large number of visits to grocery stores or frequently procured essential items from e-commerce websites. 

Marketers have observed that consumers are showing less loyalty to brands, as they are filling up their pantries by buying products they require from any brand. They just want to make sure that they have enough goods to meet their requirements for a long time.  

This new behavior pattern exhibited by consumers has a good amount of deviation from the normal consumer behavior that the marketers are accustomed to. Before the coronavirus pandemic, marketers could easily predict consumers’ behavior and show them ads on the basis of the behavioral data tracked and collected by them.

This form of advertising, known as behavioral advertising, makes use of third-party cookies and collects user data such as websites visited, webpages viewed, time spent on website/web pages, visit frequency, clicked links, products viewed, purchase history, etc. This data helps marketers to create rich profile of consumers. But in the difficult times such as the coronavirus pandemic, when consumers do not show consistent behavior, the third-party cookies and consumer profiles created by marketers fail to predict what a consumer will be interested in buying next.

Marketers also rely on geo-targeting for serving ads to users. Geo-targeting refers to the practice of delivering ads to consumers on the basis of their geographical locations. It is often used by marketers for advertising to local prospects and help local businesses that depend on foot traffic such as restaurants and brick-and-mortar stores to increase sales. But during the coronavirus crisis, geo-targeting is also not delivering success to marketers as people are refraining from going out of their homes except for essential items.

Thus, consumers’ behavioral and geographical data, which is considered highly valuable by marketers under normal circumstances, loses its importance during the times of a crisis, as its use fails marketers in achieving targeted results.  

In this scenario, it is the contextual targeting that acts as the savior for marketers. Contextual targeting has emerged as a very effective way of advertising. Contextual targeting involves placement of ads on the basis of the content the user is actively engaging with and has nothing to do with users’ past behavior, purchasing habits, and their locations. It makes use of the context of the digital content that a user is consuming rather than his or her data profile.

Traditional contextual targeting based on keywords and topics has produced results less than optimal as it involves contextual fails. For example, if an ad of a burger appears against the content talking about the harmful effects of fast food, then it will severely harm the image of the burger brand.

But this is not the case now; the advent of new technology has changed contextual targeting radically. AI-powered contextual advertising that makes use of computer vision has emerged as the smartest and the most effective way of using context for targeting audience. Through computer vision, in-video contexts such as faces, emotions, objects, logos, actions and scenes are detected with high accuracy, enabling marketers to serve ads on the basis of what the user is interested in at the moment. With computer vision powered contextual advertising, the chances of user clicking or viewing the ads are very bright. 

During the coronavirus pandemic, not only the consumers’ behavior has become erratic, they are also overwhelmed with the coronavirus content. They do not want to see brand messages appearing next to mortality-related coronavirus content. For marketers, it is a challenge to distinguish between safe and unsafe coronavirus content. Computer vision powered contextual advertising not only provides highly context relevant ad placement, but also accurately filters out harmful, unsafe or unsuitable content such as mortality-related coronavirus news. Apart from the recognized unsafe categories, marketers can custom define unsuitable contexts unique to each brand. Thus, computer vision-based targeting ensures true brand suitability.   

In crisis such as coronavirus pandemic, when consumers’ past behavior data becomes useless for marketers, computer vision-powered contextual targeting serves as the most effective way to serve ads in a truly brand suitable environment.   

Read More

Achieving Intelligent Content Moderation Through Computer Vision

Achieving Intelligent Content Moderation Through Computer Vision

  04:30    Silver Push

---

The quantity and diversity of content on the internet are increasing dramatically. Loads of content in the form of videos, images and text is daily added to the internet by the users around the world.

Online platforms such as social media sites, e-commerce sites, dating and matrimonial sites, online communities, chat rooms, forums, etc. make maximal use of the user-generated content.

For brands, user-generated content is an important tool for building recognition and trust, as content about brands generated by consumers is deemed highly trustworthy by other consumers. For brands, user-generated content can come in the form of social media posts, blog comments, posts on forums, feedback, testimonials, etc. User-generated content serve as a form of brand promotion by consumers themselves. Majority of consumers take user-generated content into account when judging the quality of a brand and while making a purchasing decision.

Although the user-generated content is the fuel that powers the internet, it carries an inherent risk of being inappropriate, objectionable, harmful, dangerous, or offensive. Such a content can hurt the sentiments of people, cause people to form negative opinions, promote terrorism, damage reputation of brands, and can have other dire consequences.

An effective solution to this problem is provided by content moderation. Content moderation involves scrutinizing and filtering the user-generated content according to the guidelines adopted by a web platform.

Although manual content moderation works fine when the volume of content to be moderated is small, the process becomes highly cumbersome when large volume of content loaded on a daily basis is to be handled. Moreover, manual text, image or video moderation is plagued by human errors.

Automated AI-powered content moderation platforms can easily handle very large volume of content with high accuracy. These automated platforms make use of computer vision for content classification at scale, significantly reducing human intervention. Computer vision detects faces, emotions, objects, on-screen texts, logos, activities, and scenes in images and videos with high accuracy.

Computer vision powered content moderation platforms involve building of image classification models for image moderation and video classification models for video moderation.

For image classification, several category-specific models are built. The classification component of each category model gives a binary prediction, i.e. it predicts whether an image should be classified as belonging to that category or not.

A category-specific model is trained on a very large number of images. Training, validation and test datasets with positive and negative examples for that category are deployed. For the purpose of training, labeled images are obtained from the following sources – collection of internally labeled images, search engine results for relevant queries with filter set to show only images labeled for reuse and modification, and synthesized images.

For testing these category-specific models, both offline and online cross-validation are performed. Offline testing is performed against a cleanly labeled and reliable dataset.

For evaluation of the performance of the classification models, the following metrics are considered - classification accuracy, precision, recall rate, false discovery rate, and false rejection rate. The recall rate denotes the accuracy in detecting images that should be flagged while the false rejection rate gives the percentage of the images that have been falsely rejected.

To make the final decision on an image, an ensemble model based on the aggregated predictions of different category-specific models is used. An image gets the final approval only after it has been approved by all of the category-specific models.

Based on the customers’ requirements, AI-powered content moderation solutions make use of custom-trained and tailor-made content classification models. This allows the use of customer-specific categories and tags.

Using computer vision technology, effective content moderation can be achieved at scale with minimal human intervention, while avoiding under- and over-moderation of content. 

Read More