AI at work: Automated ad placement with programmatic advertising
Digital advertising is becoming more and more programmatic. In programmatic advertising, AI delivers the data that make it possible to present the right content to each user at the right time. Will traditional media planning soon be obsolete?
Are we breaking new ground with programmatic advertising?
Programmatic advertising has been a buzzword in the German online marketing world for around ten years. However, many marketers in Germany still don’t have a clear idea of the basic concept of automated trading with ad space – and not everyone grasps the full potential harbored by the method when it comes to optimizing online marketing campaigns, both in B2C and B2B environments.
But the trend is clearly moving towards programmatic ad placement: in 2019, 65 percent of the expenditure for online display advertising and almost 50 percent of the total advertising expenditure in Germany was spent on programmatic campaigns. By next year, the share is expected to rise to 84.3 percent. This makes programmatic advertising the fastest growing segment in the display advertising market and describes a meteoric ascent.
The duopolists Facebook and Google in particular have benefited from the long-booming popularity of programmatic advertising. The two big players have the data base and provide the tools to display target group-specific advertising. So anyone who places ads via Facebook or the advertising services of Google Ads is already benefiting indirectly from the driving force behind these, AI.
How does programmatic advertising work?
The concept behind the method is easy to understand: ad space is automatically purchased via platforms and sold in a promotional process. Marketers and advertisers form the demand side, which purchases ad space in real time bidding (RTB). These spaces are made available by the sell side, in other words the providers of online ad space. The automated real time auctioning of ad space via programmatic advertising is increasingly replacing the traditional, manual purchasing of a specific ad space on a certain website.
Just why programmatic advertising is so unstoppable is obvious: the purchasing of an individualized ad space and the display of advertising content tailored to the personal visitor profile takes place in real time via the bidding process. The comparison of whether the website visitor is a person relevant to the target group, the sale of the ad space to the highest bidder and the display of the ad take place completely automatically while the browser loads the website.
There are no more lengthy selection, negotiation and purchase processes. All marketing tasks dealing with the planning and display of advertising content are integrated into the PA technology. For RTB, advertisers set a budget beforehand and specify how often the content is to be displayed per day.
Programmatic online advertising therefore offers advertisers a whole host of advantages:
- Automatic processes help in making efficient investments.
- The bidding process keeps costs low.
- Data-based targeting keeps wastage to a minimum.
- Ads are purchased automatically and without taking any time.
- The reach is immense via advertising networks worldwide.
- The campaign performance can be viewed in real time.
How does AI support programmatic advertising?
But what role does AI play? Artificial intelligence ensures that the desired target group is reached with this method of ad display – and more precisely than would ever be possible with even the most in-depth target-group analysis. Naturally, the more user data available to the programmatic advertising algorithms, the more intensively they work. To really reach the target group with creative advertising content, immense amounts of data are required.
Whether a target group or individuals in retargeting – to make sure your ads are displayed to the people you want to reach, AI delivers the data you need to know how and where you can reach these people. This data base is also used by the programmatic system to place bids on the right platform in real time bidding and thus ultimately to purchase the most suitable location to display the ad. Without AI there is no targeted ad placement, no target group-oriented positioning, high wastage and additional costs.
Highly developed and constantly improving algorithms use the complex data sets to make a prediction about the conversion rate or CTR. This predicted value is then translated by a bidding algorithm into a concrete bid price for a specific ad space. The comparison of various interested parties’ bids for ad space and the display of the winner’s ad take place in real time within 100 milliseconds while the website is loading for the visitor.
How effective is programmatic advertising and what data does it use?
The programmatic advertising system uses five different data types. The basic goal is to control campaigns optimally and to be able to optimize them in the process if necessary, to significantly increase the lead and conversion rate. These data are:
- Consumer data: Demographic, sociographic and psychographic data, browser cookies that provide information about purchases, shopping cart contents, website visits and the way the user is redirected to the website
- Contextual data: Time of day at the time of use, geographical position of the user, local weather conditions to display content appropriate to the situation
- Creative data: Preferences and ideas of the user, recognizable, for example, by the designs the user responds preferentially to, so they can be reached with adapted layouts and contents
- Campaign data: KPIs that provide information about the results of a campaign, for example visibility, impressions, CTR, conversion rate, consistency of consumer and contextual data for the display rate of content
- Cost data: Expenditure per advertising contact, summary of costs for data, programmatic system, bids, display and reporting to record all cost units
The data with which programmatic advertising operates are therefore only partly derived from an analysis of personal customer data. To gain a comprehensive understanding of both the behavior of the target group and the campaign’s cost drivers, the technology itself constantly accumulates data. In this way, optimization measures can be taken during a campaign if necessary.
AI constantly learns by comparing prediction models and probability calculations with the actual results of individual events and campaigns. The system adapts future predictions accordingly, so that the predictions and ad display accuracy become more and more accurate and precise. This ultimately makes personalized dynamic advertising possible: programmatic advertising gives you precise predictions as to which user you can ideally reach with which message, in which design, via which channel, at what time.
AI also plays a crucial role in the RTB process: the bidding algorithm not only uses the predicted value, but also takes a number of factors into account, such as campaign duration, campaign timing, statistics from previous bids, cost data or auction format. The AI constantly takes into account a vast amount of different data and information in real time and is thus able to react immediately to any market change. The AI operates proactively, systematically and adaptively.
Is the end of traditional media planning in sight?
Given the speed, the data processing capacity and automation provided by programmatic advertising, is traditional media planning passé and has digitalization taken over? Yes, AI will completely take over huge tasks:
The often lovingly outlined but hypothetical personas could increasingly give way to hard figures and measurable facts, as these provide in-depth and accurate insights into who each individual user actually is and how they behave in detail. Even if this change is not a radical turning point: practicing programmatic advertising also means throwing overboard some cherished marketing practices, assumptions and theories about target groups and user behavior.
Administrative and analytical processes that do not require creativity have cost media planners a lot of time up to now. These tasks can be fully automated thanks to AI. The vision is therefore initially to leave only strategic, administrative and recurring processes to AI, while giving marketers more space and a more solid information base to develop creative and high-quality content.
Is programmatic advertising doomed to disappear before it even really got going?
Since May 25, 2018, the rules of the GDPR have been binding for all EU states. Before the General Data Protection Regulation came into force, gloomy scenarios were conjured up from many sides: the end of AI in all marketing activities was said to be imminent. A technological Armageddon that would catapult digital advertising back into the Stone Age. Things turned out differently – luckily.
The GDPR did not deal a death blow to the programmatic market, instead it made sure that it became more transparent – a thoroughly positive trend. User figures and company investments show that, especially since 2018, confidence in AI-based advertising methods has grown rapidly up to the present day, and even previously doubtful companies are increasingly installing and accepting programmatic advertising as a central marketing tool.
But no sooner had the specter of the General Data Protection Regulation disappeared than a new catastrophe was looming: the ePrivacy Regulation would not only have a negative impact on the user experience on the Internet, but also pose a real threat to AI-based programmatic advertising. The death of cookies is prophesied.
Since the ruling by the European Court of Justice in October 2019 it has been clear that the setting of cookies will require the active and voluntary consent of the user for the time being. Personality profile-based online advertising is therefore prohibited until the user gives his or her consent or objection. This is a central question for programmatic advertising, which is based on this very data. At the moment the precise legal situation is still vague, if not completely unclear, in many points. Cookies for the shopping cart, login data and language settings, for example, are still permitted.
- Demand side platforms (DSP)
- Sell side platforms (SSP)
- Data management platforms (DMP)
- Customer journey mapping
- Frequency capping (determines the maximum number of times a certain ad may be shown to a user)
In the worst case scenario, there is a threat of a return to the cost-per-click model. However, it remains to be seen in what form and when the ePrivacy Regulation will finally take effect.
When drafting and adopting new data protection rules, it should never be forgotten that programmatic advertising makes a crucial contribution to a positive user experience on the net: users are increasingly seeing ads and advertising content relevant to their individual needs. Having to watch a commercial about the new Mercedes-Benz A-Class for the hundredth time as a passionate cyclist is really not in the spirit of user-friendliness. And for advertisers this means extremely high wastage and pointless expenditure.
However, potential alternatives to the cookie, which seek a compromise between data protection and personalization, are already in the starting blocks:
- Semantic targeting: The placement of advertising based on defined keywords
- DigiTrust universal ID token: Users control the information available about themselves and optionally release it to advertisers.
- Fingerprinting: Identification of the user by unique features such as browser navigation, operating system, color depth, plug-ins and fonts.
New data protection rules do not necessarily mean the end of programmatic advertising, but it must be in the mutual interest of consumers and businesses to receive data-based personalized advertising.