AI in Marketing: With Learning Algorithms to New Target Groups
Although artificial intelligence does not replace human marketing teams, it does support employees in managing campaigns or content marketing. In the future, it could increasingly replace a popular means of brand expression: Emotions.
The announcement in March that Zalando had replaced 250 marketing employees with artificial intelligence went through the media landscape like an alarming wildfire. On closer inspection, however, the AI threat backdrop becomes somewhat more relative: Zalando decided to restructure a number of processes, for example, in order to localize marketing measures even more strongly and to distribute content production decentrally to the sales countries, which goes hand in hand with organic staff reductions.
Of course, it was smart for Zalando, as a technology leader, to adorn itself with AI innovation leadership in marketing. Ultimately, only part of the marketing will be replaced by algorithms.
Even if the news has far less impact than initially expected, AI will profoundly change marketing. A recent study by McKinsey shows that, in terms of the various operational functions in companies, the greatest potential for AI lies in marketing, with a potential for value creation of up to six trillion US dollars. Interestingly, this contrasts with a very small number of successful AI applications in marketing - especially in German companies.
Artificial intelligence in the sense of a completely autonomous, self-learning system is indeed not yet a practical reality. There are indeed impressive examples such as Google's AlphaGo, a program for the board game Go, which has been able to beat professional Go players several times and thus shows the increased potential of AI. But these programs are still based on a rather mechanistic approach, which "learns" diligently from large amounts of data. In a way, they are still laboratory situations that have little to do with corporate reality. The analogy of neural networks with the human brain is also misleading. Not only are we quantitatively miles away from the human brain's neuronal quantity structures and relationships, we still don't know exactly how the human brain works, but we want to recreate it. And yet, even if the AI approaches simulate a quasi-intelligence, they can sustainably support and change marketing. From an entrepreneurial point of view, the dimensions automation and augmentation ultimately count: How well can AI automate marketing processes and how well can AI support or optimize marketing decisions?
The AI Marketing Matrix
Most AI applications, like the Zalando example, relate to the automation of marketing functions and processes. In this context, systems also make simple decisions on their own. This usually involves substituting human activities with artificial intelligence in order to achieve cost and efficiency advantages. There are many automation applications that already have a high degree of maturity and practical application. These include, for example, marketing automation or real-time bidding. In contrast, augmentation applications are particularly concerned with the intelligent support and enrichment of complex and creative marketing tasks, which are currently still performed by human actors. Thus, the AI can automatically analyze competitors, target groups and trends. Marketing managers can use these insights to develop or adapt their strategy. This enriches decision-making processes with important information. However, the actual decision is not automated, but remains at the sovereignty of human actors. In the sense of augmentation, AI can also help marketers to manage the increasing complexity of channels and touchpoints. In this way, both the value contribution of a channel and the necessary interaction of the channels can be calculated to optimize conversion. On the basis of extensive customer journey data, the optimal media budget allocation over time can also be determined: When should Euro be invested in which channel? The final media plan, including the organizational distribution of roles, is then - at least today - created and evaluated by people. Due to the higher complexity and creativity of these tasks, both the degree of maturity and the distribution are significantly lower compared to the automation examples with AI.
However, there are also applications which, despite their high degree of maturity, are still used comparatively little in practice today. One area of application to which this phenomenon applies is the principle of look-a-likes, which can be used to identify and profile target groups. Artificial intelligence can collect more than 10,000 data points on the web to companies or consumers and identify and profile new target groups using so-called deep learning algorithms. In the B2C sector, for example, this can be implemented well with Facebook Custom Audiences.
AI also plays an increasingly important role in content marketing. Algorithms for the semantic conceptualization of content are helping here. For example, the word-to-vec algorithm automatically maps the content in the form of vectors that abstractly formalize the actual content. This representation is much more powerful than the typical index about content. On this basis, similar or complementary content can automatically be found on an analytical level. So-called "Long-Short-Term-Memory-Recurrent-Neural Networks" can also produce new content in the sense of predictive content by predicting the next obvious words and sentences based on words and including the temporal context. One example is the AI-supported creation of an issue of the British marketing magazine The Drum. Thousands of copies were printed of the edition, where the AI selected images, adapted text, and designed the pages. The AI was fed with data from the winners of the Golden Lion at the Cannes Lions International Festival of Creativity. The aim was not only to create the magazine, but also to create an artificial intelligence that meets the taste of the lifestyle audience. Robotic journalism is also becoming increasingly creative on the basis of such methods. Algorithms are able to automatically search the web for information, combine it and create a readable piece of content. Data-based reports in the fields of sports, weather or finance are already frequently generated automatically today.
This blog was translated from the following article: https://t3n.de/
************* Promotion *************
Campaigning in the age of algorithms
Secure your ticket now: http://l.ticketino.com/CSCH19
You like this content? Subscribe to your newsletter now: http://businesscampaigning.12all.com/f/1920