The AI Ads Machine
Will Artificial Intelligence remain the revenue engine of 21st century advertising?
Until the late 90s, ads had one purpose: to sell and create cravings for products people didn’t know they wanted. They appeared on TV, in magazines and on billboards, disrupting the shows they wanted to watch, the articles they were reading or simply their views. The internet led to a paradigm shift and put the customer in the center. With the rise of (social media) networks, multi-service-, “free” search engines, and shopping platforms, contextualized advertising transformed the business radically. The widely discussed downside to this is that there is no such thing as free lunch, and as a result, the customer’s data has become the product. At the same time, advertising has never been more “meaningful”, as the user indirectly influences the offers being displayed in a seamlessly integrated experience.
Leading up to our bi-annual rebalancing, we questioned the longevity of ads as a business against the backdrop of new GDPR regulations, and how much innovation is truly driving continued value creation at Alphabet, Twitter, Facebook and the likes.
The application of AI in advertising remains a compelling and powerful one. Google, Facebook and Amazon have invested heavily in complex AI systems — and turned them into revenue generators in no time. Was that it or are we seeing just the dawn of an era?
Advertising revenue of Google from 2001 to 2020 (in billion U.S. dollars)
To get closer to an answer, Singularity Think Tank expert Alexander Stumpfegger clusters and summarizes some of the main aspects of applied “AI in ads” for us:
Intelligent data collection and tracing these back to single individuals via tracking procedures: Today, due to privacy measures, “simple” tracking via cookie is no longer sufficient. A wide variety of data points must be collected and linked in order to identify users and compile a profile.
Mapping and ease of use for advertising customers: It’s easier than ever to book ads with either precisely or only roughly defined target groups. On the one hand, this requires good clustering of the users identified in the first step to enable detailed targeting, on the other hand, it requires intelligent presentation of the advertising.
Dynamic real-time analysis of content: Search queries on Google are initially unpredictable. At the moment of the search, the appropriate advertising must be displayed. On other websites outside their own platforms, the context and user must be reliably identified in order to choose the most appropriate advertising. This requires crawling, language processing and tracking technology.
Real-time success tracking: Campaigns are optimized on an ongoing basis. Based on the advertising/user matching as well as the dynamic price calculation (who bids how much for which “advertising space”, i.e. presentation in front of the desired target group in the desired context) and the goal of serving several advertising customers also proportionally, the advertising is presented dynamically. Feedback is evaluated directly (including how long someone was able to see the advertisement at all). Campaigns are thus directly optimized further and customer profiles are also sharpened.
The fully automated and intelligent setup provides advertising clients with unseen opportunities, while the platform itself sells without any sales efforts in the backend — the customer is doing everything by herself. Ad syndication allows to reach audiences even beyond the original platform’s user pool, which in fact means that a few platforms share and control the main bulk of the ad spend and space. Contextualized advertising in combination with seamless, often “one-click” payment funnels are a marketer’s dream. The time from first touchpoint to purchase has not only decreased tremendously, the whole customer journey has become trackable, measurable, and adjustable.
Are GDPR rules the end of targeted advertising?
While “the internet” seems to provide boundless opportunities when it comes to collecting and processing user data, new privacy rules try to restrict the plethora of possibilities. Does this cast a shadow over the Land of Cockaigne? Not really, says Stumpfegger: The development of intelligent methods for tracking users or simply deciding even faster which context-relevant ads can be displayed without much tracking are levering regulations out. Smart AI allows for compliance with tightened regulation by still providing relevant ads.
“Even Facebook recently seems more relaxed about Apple’s tightening of privacy rules, for example. Even the GDPR and the annoying cookie banners have not harmed the whole business,” comments Stumpfegger.
On the contrary: Examples like Duckduckgo show that even without user tracking, the smart algorithms of today manage to match advertising with search queries in real time without giving any leeway for privacy infringement.
Amazon takes AI-powered advertising even further. The company allows ads that direct the user to other providers’ offers outside their own network. What makes this still attractive for Amazon? More companies and more users mean more data, which in turn helps Amazon sell more itself. Facebook, however, provides highly targeted steering as the social media giant knows its user inside out. From gender to hobbies and dislikes to location, job, or circle of friends — advertising customers can tailor target groups to their product specifications or unique selling propositions and increase conversion rates tremendously.
“The most dominant player remains Google, though,” concludes Stumpfegger.
Through it’s massive search engine, multiple apps and interconnectivity, it has so many data points available that we’ve only seen a fraction of what’s actually possible with this — not only in the ads business.