Abstract
The ongoing integration of artificial intelligence is fundamentally transforming the way we access digital information. Search engines are increasingly evolving from mere intermediaries into direct providers of answers. For businesses, this technological development presents a challenge, as traditional search engine optimization strategies are becoming less effective. To maintain a presence in AI-generated search results, the discipline of Generative Engine Optimization is coming to the fore. However, as this field of research is still in its infancy, there is a knowledge gap regarding which factors actually increase the visibility of corporate websites in generative systems. This thesis addresses this issue and investigates how digital content must be prepared in order to be cited as a source by AI models.A two-stage research design is employed to provide well-founded answers to the research questions. In the first stage, a systematic literature review identifies the theoretically postulated success factors for generative search. Building on this, an exploratory and quasi-experimental study is conducted. In the study using the AI search engine Perplexity, the effect of two selected optimization measures is tested on real corporate websites from the construction industry. The websites are divided into test groups and specifically supplemented with quotations or external references. Visibility is assessed by comparing a first and a second measurement using the quantitative metric of the Position-Adjusted Word Count, as well as a supplementary subjective relevance assessment. An unchanged control group serves as a neutral reference value.
The theoretical analysis shows that, in addition to a technical foundation, a clear content structure and verifiable evidence of information are the key drivers of AI visibility. However, the empirical part of the study paints a more nuanced picture of the reality on the ground. The targeted and isolated implementation of the Quotation Addition and Cite Sources measures did not lead to any significant positive changes within the limited observation period. The measured values remained largely constant following the content-related interventions. The minor deviations primarily reflect the natural fluctuations of generative systems, as the results of the control group also indicate. The study thus demonstrates that selective text adjustments are insufficient to overcome the algorithms’ complex selection criteria in the short term. In practice, successful Generative Engine Optimization requires a holistic and long-term strategy built upon a strong foundation of traditional search engine optimization.
| Date of Award | 2026 |
|---|---|
| Original language | German (Austria) |
| Awarding Institution |
|
| Supervisor | Fabian Josef Stangl (Supervisor) |
Studyprogram
- Marketing and Digital Business
Cite this
- Standard