Marketing on steroids: fewer cookies, more generative AI

Author: Kerstin Lomb

Kerstin Lomb bei ihrem Vortrag am IBM iX Stand auf der OMR

The upcoming ban on third-party cookies is forcing companies to rethink data-driven marketing. While the sector is still concerning itself with access to customers’ data, endeavours such as ChatGPT, DALL-E and other generative AI concepts are changing the rules of the game. After all, it has probably never been easier to take customer retention and data analysis to the next level.

When ChatGPT was launched in November 2022, the reaction was one of great astonishment, disbelief and – in many sectors – dismay. Generative AI and tools such as ChatGPT, DALL-E and Midjourney show just what is possible when a text-based question-and-answer schema is combined with creative abilities. For digital marketing, generative AI offers an entirely new way of gaining access to the minds and wishes of customers.

The evolution of revolution – what is generative AI?

Generative artificial intelligence combines machine learning, deep learning, large language models and automation within artificial neural networks. Generative AI systems are able to use existing content to create new content, ranging from text, images, videos and audio files through to data analyses from various different aspects.

These results are based on prompts – specific requests given to the AI. These prompts trigger the evolutionary optimisation process.

  1. Existing datasets are analysed based on their similarities
  2. These similarities are used as a basis for creating the new content
  3. The main features of the result are based on the prompt; the result also relies on the rules that the AI extrapolates from the similarities

When analysing a prompt such as “Paint a picture like Picasso’s Guernica”, the AI will usually look at the painter’s style, the style of the picture, the wider context in terms of art history, and so on. Two models are used to drive the evolution of generative AI:

  • Generative adversarial networks (GANs) create visual and multimedia content using fewer data points from images and text
  • Transformer-based models (or generative pre-trained language models) are text-focused and generate new text-based content using the internet

Four AI paths, one destination – personalisation at the highest level

When it comes to digital marketing, content production is only the second building block within an AI-driven data workflow. Data is analysed based on prompts, and patterns and rules are then established and applied to new contexts. These contexts form the basis for forecasts relating to trends, risks and opportunities. Depending on the analysis, data pool and prompt, the AI theoretically has infinite amounts of data at its disposal for bringing the forecast as close as possible to p=1 probability. Both AI products – content and forecasts – are currently being tested out and developed as part of campaigns and marketing initiatives. At this moment in time, there are four key concepts:

  1. AI as an experience – user inputs transform branded content into AI works
  2. AI as a personal shopper – generalist traders offer AI-based assistants to help with personalised navigation through the product range
  3. AI as a customer experience designer – AI-induced personalisation of websites and customer platforms
  4. AI as a marketeer – data analyses and forecasts of customer behaviour and requirements for various use cases

When broken down to the essentials, we can see that AI already provides marketing workflows with far-reaching optimisation potential for the customer experience at a fundamental level:

  1. SEO for digital campaigns: relevant keywords, content structures, headlines and more
  2. Formation of opinions: text content with a positive, neutral or negative tone, with analyses of mindsets from customer feedback as a basis
  3. Content: customer emails, posts, blog articles, storytelling, product descriptions, product images, other images, etc.
  4. Virtual dressing room: more intricate applications and representations than with photo tools alone
  5. Audio-visual demos: products and services in action – according to the customer’s specific needs
  6. TV spots: customised creation of TV spots and other advertising materials

Considering AI from a more in-depth perspective, however, it is a strategic tool above all else; a tool that analyses the wishes, requirements and challenges of target groups, compares these to the competition, and uses this to generate approaches and content for personalised campaigns.

Limiting limitlessness – challenges when dealing with AI

In AI’s abilities, we also see the crux of the matter: it uses personal data to a much greater extent – and how it uses this data has not yet been mapped sufficiently. For this reason, OpenAI concepts (with ChatGPT a particularly prominent example) should be used with caution. Customers also want to know when their data is used by AI – whether it’s as an author of texts or a generator of images. Companies must communicate this in an open and honest way, and with “good justification”. Ultimately, AI is a strong tool, but it is not all-powerful – and it will always be just a tool. Marketing will continue to be about human-to-human communication. It will only be the workflows that will be changed by the introduction of AI. It is exactly these workflows, however, that offer a wide array of new challenges that companies must address.


Technology doesn’t limit us as people – only our creativity! Generative AI is a powerful tool for designing innovative and inspiring marketing campaigns.

What’s up?

Link copied Link copied?

This may interest you too