Skip to main navigation Skip to search Skip to main content

How AI can boost qualitative and quantitative research methodologies in B2B: The case of Voestalpine in the Railway Sector.

  • Aleksei Bespechnyi

    Student thesis: Bachelor's Thesis

    Abstract

    In this Bachelor thesis the aim was to investigate the potential of artificial intelligence technologies for improving the qualitative and quantitative research methodologies in Business-to-Business (B2B) markets, more specifically with the help of the two methods of sentiment analysis and topic modeling. The study fills this gap, focusing on the dilemma of the growing amount of unstructured digital information and the still mostly traditional methods of market research in capital-intensive sectors such as the European railway infrastructure sector as illustrated in the case of Voestalpine Railway Systems.

    The big question for research was: Is it possible to systematize the use of AI methods in the monitoring of digital reputation and detection of strategic trends in professional online communities in B2B markets? Three sub-questions related to relevant data sources (RQ1), the dominant themes and the dynamics of sentiment (RQ2), as well as strategic action insights for Voestalpine (RQ3) were identified.

    The text corpus consisted of 570 text units, in particular corporate press releases, LinkedIn posts, and employee reviews in Kununu and Glassdoor (owned and earned media) from the time period of 2022-2024. Multilingual sentiment analysis was done with a multilingual Transformer model (Barbieri et al., 2021), and topic modeling with the BERTopic framework (Grootendorst, 2022).

    In external reputation, Key findings indicate that the reputation of Voestalpine Railway Systems is good. The focus of Corporate Communication is positive with regards to railway systems, digital monitoring solutions (Zentrak), turnouts and Sustainability initiatives. Employee reviews show slightly negative sentiment on salary, management and working conditions. There was low negative sentiment overall in the corpus, with a score of 4.6% which shows the corporation has a very stable reputation, however there was a gap between the external image of innovation and the internal perception of them as an employer.

    The thesis concludes that traditional research methods are not efficient to provide for B2B companies the insights that are required in real time and from multiple stakeholders, where AI-based research methods can do so. The study suggests creating a sustainable employer branding strategy for Voestalpine Railway Systems, to redefine company communication on digital railway solutions and sustainability, and to continuously apply AI-based digital listening.

    While the study has its merits, it also has certain limitations such as the small size of the data set, the use of publicly accessible data and the difficulty with which a sentiment classification is possible in very technical industrial settings using a transformer model. Further work could combine multimodal data sets and build on larger datasets from industry.

    The work offers a viable example of the application of AI in heavy industry B2B environments, and shows how to reproduce the digital reputation analysis and trend modeling framework.

    Keywords: B2B marketing, sentiment analysis, topic modeling, digital reputation, railway industry, artificial intelligence, Voestalpine
    Date of Award2026
    Original languageEnglish
    Awarding Institution
    • FH OÖ
    SupervisorRomana Forster-Gartlehner (Supervisor)

    Studyprogram

    • Global Sales and Marketing

    Cite this

    '