The invisible shareholdersWhy AI visibility is becoming a key success factor in financial communications

Published in the October 2026 edition of GoingPublic (in German)

Many companies still view the annual report primarily as a publication. Yet the real challenge has long since shifted to another level. The annual report is evolving from a document into an information system. Its role goes beyond meeting regulatory requirements and informing investors. It is increasingly becoming a data source for search engines, AI platforms, financial portals and digital research processes.

Traditional investor relations has long assumed that analysts, investors and journalists consume corporate information directly. Artificial intelligence is fundamentally changing this logic. A growing number of market participants use AI systems to research, structure and analyse information. Studies presented during a webinar on AI visibility point to a clear trend: AI is increasingly becoming part of professional research and investment processes.

This raises a new question: Can AI systems actually find, interpret and accurately reproduce the information a company publishes? Or do they rely on external platforms, third-party sources or outdated information?

This question was the focus of a study of 50 listed companies in the German-speaking DACH region. The analysis covered 3,000 AI queries on key financial and ESG metrics, including revenue, EBITDA, dividends, employee numbers and Scope 1 emissions.

The results reveal a clear pattern. AI systems strongly favour structured HTML content over traditional PDF documents. HTML sources were preferred in 78 per cent of the cases analysed, while PDFs served as the primary source in only 22 per cent. In Switzerland, the share of HTML sources was as high as 81 per cent.

Even more striking is the question of data sovereignty. Where companies use modern HTML-based reporting solutions, AI systems are far more likely to draw on the companies’ own original sources. By contrast, when reporting relies predominantly on PDFs, AI platforms are more likely to use information from third-party providers and may therefore process inaccurate information.

For investor relations professionals, this represents a paradigm shift. Publishing information remains essential, but it is now equally important to ensure that a company’s own information actually serves as a reference source for AI.

The study also found that artificial intelligence makes errors significantly more often when working with PDF-based reports. Complex table structures, footnotes, currency conversions and metrics based on company-specific definitions, such as non-aligned financial and calendar years, are particularly problematic.

Across all companies examined, the measured accuracy was only around two thirds of responses. For complex metrics such as EBITDA or Scope 1 emissions, accuracy rates were at times considerably lower.

There are many reasons for this. AI systems may confuse subsidiaries with the group, misinterpret financial years or turn to external financial portals when they cannot find unambiguous information on the company’s website. The analysis identified examples of misinterpretation involving companies such as ABB, Sonova, UBS and Zurich Insurance.

The issue becomes particularly critical when AI systems provide answers that appear plausible but are factually incorrect. For listed companies, this can create a significant reputational risk and undermine trust.

The debate is often reduced to the question of «HTML or PDF?». In reality, this perspective is too narrow. An HTML report is not automatically fit for AI. The structure, semantics and clarity of the published information are just as important.

Four levels must work together to achieve strong AI visibility:

Only when these factors work together can AI systems interpret and reference information accurately.

At the same time, it is becoming clear that the supposed conflict between people and machines does not exist. In future, financial communications must serve both audiences simultaneously. Investors, analysts and other stakeholders need content that is clear, visually engaging and compelling. AI systems, meanwhile, need structured, machine-readable information that can be interpreted unambiguously.

The future-ready solution is therefore not to produce separate reports for people and machines, but to adopt a single-source publishing model. A central data and content base feeds multiple output channels: interactive HTML reports for people and machines, accessible PDFs for archiving and download, and XBRL data for regulatory and machine use.

For CFOs, investor relations teams and corporate communications professionals, this leads to a clear recommendation. AI visibility should become an integral part of the governance and quality assurance of corporate reporting. In particular, companies should:

The next stage in the evolution of financial communications is not simply digitalisation, but visibility. In future, companies will compete not only for investors’ attention, but also for the attention of the systems that prepare information for those investors. The annual report is therefore becoming a central information source for AI-enabled capital markets communications. Companies that publish their data in a structured, machine-readable and unambiguous form not only strengthen regulatory assurance, but also gain a strategic advantage in visibility, data sovereignty and trust.

How accurately do models such as GPT-4o, Claude and Gemini assess your company’s metrics? On the ir-monitor.ai platform, investor relations professionals can have their company’s AI visibility evaluated:

Materials from the webinar «AI Visibility in Financial Communications — What Do ChatGPT & Co. Say About Your Company, and Is It Accurate?» are available in German on request from marketing@mmssolutions.io.