The Role of Automated Tools in Digital Advertising

As internet platforms have explored ways to optimize the ad targeting and delivery process, they have increasingly adopted artificial intelligence and machine-learning tools. Given that advertising relies on making situation-based decisions to determine what content to show a user during a certain scenario, there is significant room for automation and and algorithmic decision-making to streamline this process. For example, machine learning systems can be used to optimize the ad delivery process by evaluating and extracting insights from datasets to predict future user behavior and target ads to these users based on these predictions. In addition, automated tools can be used to deliver ad campaigns to users quickly and precisely, based on a range of granular factors such as content format and timing.¹

Generally, the advertising operations of internet platforms can be categorized into two components: ad creation and ad delivery. Ad creation is the process by which an advertiser submits an ad to a platform. For most internet platforms, there are three key stages to the ad creation process:²

1. Producing the ad creative: Advertisers develop content such as the headline, text, media (e.g. images or video), and the website that an ad should link to. This content is collectively known as the ad creative.³
2. Audience selection and targeting: Advertisers select which segments of a platform’s users they want to target with their ad.⁴
3. Bidding strategy: Advertisers specify how much they are willing to pay to have their ads shown to users. There are a number of metrics that can be used when making and assessing a bid, including per-impression or per-click bids. Advertisers can also place a maximum cap on their bid and allow the platform to bid for them.⁵

Once an advertiser has submitted all of this information to a platform, the platform may review the ad, either using automated review, human review, or a combination of the two, to ensure it is in compliance with policies such as its ad content policies, ad targeting policies, and its general platform content policies. It is important to note, however, that not all internet platforms have instituted such a review process, raising concerns that platforms may be allowed to run discriminatory and harmful ads. Once an ad is ready to be run, it will be moved to the ad delivery phase.⁶ Ad delivery is the process by which an advertising platform shows ads to users. Most ad delivery systems are automated in part or in full. For every ad slot that is available, an ad platform will host an automated auction to determine which ads should be shown to a specific user. A platform will deliver an ad to a specific set of users based on a range of factors, including the advertiser’s budget and how an ad campaign is performing⁷ (as measured by metrics such as engagement or impressions).

Although this seems relatively straightforward, there are some caveats to this process.⁸ First, platforms generally avoid showing ads from the same advertiser to a user in quick succession. As a result, a platform’s ad delivery system may disregard bids for advertisers who recently won an auction for the same user. Second, platforms assert that one of the benefits of targeted advertising is that it provides users with relevant content that enhances their user experience. Although relevance is a rather subjective notion, some platforms try to quantify it by calculating a relevance score when considering which ads to show a user. However, since the ad delivery decision-making process typically involves algorithmic decision-making, ad platforms do not solely rely on a bid to determine the winner of an auction. This process permits ads with cheaper bids, but higher relevance scores, to win against ads with a higher bid in an auction. This is because platforms benefit financially from delivering relevant content, including ads, to users, as it increases the likelihood users will continue using a platform.⁹ Third, platforms and advertisers may prefer to evenly spread an advertiser’s budget over a certain period of time, rather than spending it all at once. This introduces further caveats related to which ads should be considered by platforms for which auctions. It also demonstrates that internet platforms consider more than just an advertiser’s bid when determining which ads to deliver.¹⁰

Once an advertiser enters the ad delivery phase, they will typically receive data from the internet platform detailing how their ad is performing. This data can include demographic and geographic information on the users who are viewing the ads, as well as some information on the users who actually click or engage with the ad.¹¹

Examples of how automated tools can be used during the ad targeting and delivery phases include:

1. Dynamic Creative Optimization: Internet platforms are using automated tools to help advertisers identify which users are most likely to react to a certain ad. During the Dynamic Creative Optimization process, advertisers can tailor ad content for particular users. They can do this by, for example, using A/B testing¹² to evaluate and collect data on how a user reacts to an advertised price, an ad delivered at a certain time, or an ad delivered in a certain format. During these trials, the advertising system captures engagement statistics. These can be used to inform the future efforts of advertisers and the internet platform. For example, this information can inform the internet platform’s algorithms on what the most successful targeting strategies are. The more resources that an advertiser spends on such tests, the more refined and effective their campaigns are likely to become.¹³
2. Digital Ad Mediation: Digital ad mediation is the process of connecting mobile publishers or consumer engagement platforms (such as the Twitter mobile application) with a brand in real time in order to deliver the most relevant ad to a user. Given the volume of content and the number of users on mobile applications, the digital ad mediation process is most effective when internet platforms deploy it using automated tools.¹⁴
3. Social Media Management Software: Social media management companies are increasingly using automated tools to produce, target, and deliver social media campaigns. Leading social media management providers such as Hootsuite are harnessing the power of automated tools to gather detailed information about users’ reactions and sentiments related to certain brands. When an event that is relevant to a target user occurs, these tools use these sentiment insights to immediately dispatch an ad that may be relevant to them.¹⁵ For example, luxury fashion house Givenchy could establish a contingency to initiate promoted ads for their products anytime their spokesperson Ariana Grande appears on television or in a Givenchy-related hashtag. Further, social media management companies are also increasingly adopting machine learning algorithms into their client workflows in order to fuel recommendations regarding audience segmentation, content delivery, and more. These insights aim to significantly expand the reach and influence of brands.¹⁶