Research & Case Studies — Seeger & Partner | Political AI Strategy

AI Visibility and Election Outcomes: Baden-Württemberg State Election 2026

Case Study | March 2026

The 2026 Baden-Württemberg state election was decided by just 27,000 votes — a margin of 0.5 percentage points between The Greens and CDU. This case study examines the correlation between candidate visibility in AI systems and actual election results, using real-time GEO monitoring data from Peec AI across ChatGPT, Gemini, Claude, and Perplexity.

Key findings from this study:

How AI visibility metrics mapped to actual vote share across all major candidates. The two candidates with near-identical AI visibility (approximately 40%) produced near-identical election results (approximately 30%). In a race decided by 0.5 percentage points, the candidate with the higher sentiment score (52 vs 49) won.

Why sentiment score may have been the tiebreaker in the closest race in Baden-Württemberg state history. When visibility is equal, AI narrative framing becomes the differentiator.

The SPD collapse: with just 16% AI visibility and 8% Share of Voice, the SPD Spitzenkandidat was functionally invisible in AI systems. The result was the worst SPD performance in any German state election in history at 5.5%.

The source ecosystem: SWR, Welt, Tagesschau, YouTube, and Wikipedia were the most-cited domains across 945 total AI citations. Campaigns that maintain editorial presence on these platforms shape the AI narrative.

Tags: GEO Monitoring, BW State Election, Visibility vs. Outcome, Real-Time Data, Peec AI, ChatGPT, Gemini, Claude, Perplexity

Generative Engine Optimization: The Methodology Behind AI Visibility

Methodology | March 2026

How generative AI systems retrieve, rank, and surface brand information — and what organizations must do to remain visible in an AI-mediated information landscape. As more people turn to ChatGPT, Gemini, and other AI platforms for answers, the way brands, candidates, and institutions are represented is increasingly shaped by systems they do not control. This study provides a strategic framework for political campaigns and organizations navigating this new reality.

Key findings from this study:

Market dynamics: ChatGPT is the second-largest search engine globally. Gemini has surged to approximately 20% of ChatGPT sessions in the US and EU. Google AI Overviews appear on 15-45% of all search queries. Zero-click searches have risen from 60% to 65% of all Google queries.

Traffic impact: When AI Overviews appear, organic click-through rates drop by approximately 38%. Approximately 46-48% of consumers already trust AI-generated answers. Click-based attribution models fundamentally break for AI-driven discovery.

The four layers of GEO: Model training data (low influence), Knowledge graph layer (low influence), Grounding via own properties (high influence through SEO), Grounding via third-party sources (high influence through digital PR). LLMs exhibit strong recency bias — recently published content is cited far more frequently.

Source ecosystem: News websites account for over 20% of cited sources. LinkedIn is a top-20 source for B2B queries. YouTube is a top source for all major LLMs except ChatGPT. Reddit is dominant in English-language US queries.

Measurement framework: Core metrics include Visibility/Share of Voice, Competitive Position, Sentiment in Context, and Source Attribution. Organizations must adopt always-on digital PR, define topic-based prompt sets, invest in cross-functional coordination, move beyond click-based attribution, and prioritize platform-specific source strategies.

Tags: GEO Framework, Source Ecosystem, AI Search, Measurement, ChatGPT, Gemini, Perplexity, Claude, LLM Optimization