This is a paper from the Sourso team. We wrote it up finding by finding in the research section. Read the write-up →
VoteGEO-Bench: Measuring the Impact of Generative Engine Optimization on Electoral Visibility was submitted to arXiv on 2026-10-15 by Shiqi Zhang, Xiaokui Xiao. Read it on arXiv or as a PDF.
Why it matters for getting cited
- Generative engines frequently draw citations from lower-cost owned channels and open platforms, with some pages maintaining visibility across days.
- Rewriting text without designating an entity rarely lifts that entity’s visibility, while target-specified rewriting achieves exposure gains more frequently.
- Evaluating 11 rewriting strategies across 4 categories reveals that structural reformatting and argumentation outperform evidence addition or vocabulary edits.
- Exposure gains from the two strongest strategies remain positive across evaluated personas without being consistently increased or decreased by persona conditioning.
- Explicit partisan voter identities can create baseline exposure gaps that exceed the gain achieved from rewriting one webpage.
The main takeaway. Structuring web content cleanly and strengthening arguments with comparative distinctions improves an entity’s exposure in AI answers, surpassing the impact of superficial wording changes or added citations.
The abstract, as written
Large language models (LLMs) are increasingly used for election information, yet little is known about how external web content shapes the political information they present. Generative engine optimization (GEO) rewrites web content to improve its visibility in generated responses. To investigate its potential risks in voting scenarios (VoteGEO), we introduce VoteGEO-Bench, a benchmark covering the 2024 U.S. presidential election and the 2026 Japanese House of Representatives election, combining election-related queries, historical webpages, and voter personas. We first analyze sources cited by generative engines and find that many citations come from lower-cost owned and open-platform sources, while some webpages remain visible across days. We then evaluate 11 rewriting strategies in 4 categories, examining target specification, effectiveness across engines and elections, and the role of voter personas. Rewriting without specifying a target rarely increases target exposure, whereas target-specified rewriting more often does. Strategies that improve page organization or strengthen a position through argumentation yield greater gains than adding supporting evidence or changing wording. For the two strongest strategies, gains remain broadly positive across evaluated U.S. personas, but persona conditioning shows no consistent effect on these gains. Explicit partisan identities can create baseline exposure gaps larger than the gain from rewriting one webpage. When partisanship is implicit in the voter profile, baseline shifts depend on how strongly the engine responds to that signal.
Key terms
- Generative Engine Optimization (GEO): The practice of modifying web content to increase its visibility and citation frequency inside answers produced by generative artificial intelligence engines.
- Query cell: A test unit combining an election-related search query, a specific generative engine, and the date the query was executed.
- Target-specified rewriting: A content rewriting method where the instruction explicitly names the candidate or party whose visibility should be increased.
- Share of Voice (SOV): A metric calculating the proportion of text mentions an entity receives compared to all entity mentions within an engine output.
- Voter persona: A user profile prepended to a generation prompt to simulate distinct voter demographics, partisan stances, or issue priorities.
From the paper
Generative engine optimization studies usually freeze the retrieved document pool to isolate generation changes from retrieval changes.
Existing GEO research typically separates retrieval from generation and focuses on the generation stage by fixing the retrieved webpage pool and studying how rewriting one webpage changes its visibility in the generated response ( Aggarwal et al., 2024 ; Puerto et al., 2025 ; Pfrommer et al., 2024 ; Tang et al., 2025 ; Wu et al., 2025 ; Chen et al., 2025 ) .
Specifying an entity is crucial because general rewrites fail to guide which entity the engine should highlight.
Target-unspecified rewriting rarely increases target exposure, whereas target-specified rewriting more often does so. Notably, strategies that improve page organization or strengthen a position through argumentation show stronger gains than those that add supporting evidence or mainly change wording.
The authors observe that the offline setup avoided real-world indexing risks while leaving dynamic crawler interactions untested.
Our rewriting experiments use fixed historical webpage pools with web retrieval disabled, leaving the effects of rewriting on live retrieval and citation selection unexamined. A live end-to-end evaluation would typically require publishing rewritten political content for crawling and indexing by search and generative engines.
What the authors say it means
- Content creators who optimize multi-topic pages must explicitly direct their rewrites toward a specific entity to influence AI responses.
- AI engines rely heavily on text hierarchy and argumentative comparisons when synthesizing answers from source pages.
- Model personalization by user demographic or party background shifts baseline visibility more than individual page optimization.
What to do with it
- Organize web pages with clear headings and bulleted structures. Supported by: Strategies improving page organization showed consistent gains across all tested generative engines and election settings.
- State clear comparative distinctions against alternative positions directly on key landing pages. Supported by: Comparative argumentation ranked among the top two rewriting methods across engine and election evaluations.
- Name the exact entity to emphasize when using automated content rewriting tools. Supported by: Target-specified rewriting produced exposure gains while target-unspecified rewriting rarely increased target visibility.
- Focus optimization efforts on owned channels and persistent discussion threads. Supported by: Engines repeatedly cited low-cost owned platforms and open community sources over consecutive days.
Caveat. The benchmark evaluated rewriting effects offline on fixed sets of retrieved pages, without testing how modified pages influence real-time web crawler retrieval or live citation selection.
In short
This research demonstrates that generative engines favor well-structured pages and clear comparative arguments over mere stylistic tweaks. While individual page optimization consistently improves visibility across diverse voter profiles, pre-existing model biases toward user personas can exert an even larger effect on final responses. Website publishers seeking AI visibility should focus on structural clarity and explicit entity targeting.
Written from the paper's own text and checked by a person before publishing. Quoted passages are verbatim; numbers are the authors' own. For the method and full results, read the paper.