Summarizing “Query Implied Generative Engine Optimization”: A Hamsterdam Research Post

By Ethan Lazuk

Welcome to another edition of Hamsterdam Research! 🐹

This is where we look at recent AI research papers to learn what they’re talking about and explore their hypothetical implications for the future of search and SEO/GEO strategies.

This time, we’ll look at research called, “Query Implied Generative Engine Optimization.”

It was submitted to arXiv on September 19, 2026, and its authors are Shilpa Ramakrishna and William B. Andreopoulos.

First off, why should SEO/GEO professionals care about this research?

This research challenges a basic assumption about content optimization, which is that you need to start with a query. Instead, the authors of QI-GEO ask whether the document itself can reveal the information needs it should be serving. Their system extracts the entities and relationships already present in a document, expands them through a knowledge graph, identifies relevant concepts the document is missing, and then makes localized additions to the document to close the gaps.

The paper identifies the fact that we can never know every query that might lead to a page as a weakness of query-dependent GEO. In other words, content creators rarely know the complete set of questions users may ask. QI-GEO attempts to approximate that unknown intent space from the document itself.

QI-GEO also offers a different way to think about content gaps. While traditional SEO can find gaps by comparing keywords, competitors, SERPs, or related questions, QI-GEO instead looks for semantic relationships that are strongly implied by the document but not actually covered. The system gives more weight to concepts supported by multiple entities in the document, then filters out information that is already expressed in different wording.

The authors’ results suggest their content additions can affect citation visibility, not just topical completeness. In the experiment, citation coverage increased from 65.7% to 72.6% in the single-query setting and from 71.3% to 77.2% across the multi-query setting. The paper also reports a 12.6% to 15.9% improvement in its position-adjusted visibility metric.

And what’s most interesting for GEO is that the improvements they saw weren’t limited to the exact wording of one query. Across five semantically related query formulations, 66% of documents had a net-positive mean gain, and 73.7% improved or tied on at least three of five phrasings.

For SEO and GEO practitioners, this research is an interesting shift from optimizing content around a list of queries to optimizing the semantic coverage of the document itself.

One caveat is that the experiment tests whether an already-retrieved document becomes more visible or more likely to be cited after optimization; it does not show that QI-GEO helps a page get retrieved in the first place.

Let’s start by reviewing the paper’s abstract.

Here’s the abstract, which is worth a read in full:

“The landscape of search has changed drastically with how people look for information online. Traditional search engines are being replaced by Generative Search Engines (GSEs), which use Large Language Models (LLMs) to generate natural language responses to user queries. For content creators, visibility is no longer solely determined by ranking in search results but by being cited within generated responses. But Generative Search Engines are black-boxes, leading to the emergence of Generative Engine Optimization (GEO), a set of techniques aimed at improving content visibility in generative search settings. Most existing approaches rely on the explicit queries or query derived signals to align content to better suit user needs. We propose Query Implied Generative Engine Optimization (QI-GEO) to infers user intent directly from the document. Our approach approximates document’s intent space and identifies content that may be missing yet relevant to answer potential user queries. Evaluation on GEO-Bench and Extended GEO-Bench demonstrated improvements across objective and subjective metrics. QI-GEO improved objective scores by up to 15.9% and subjective scores by up to 17.6%, while yielding nearly twice as many citation gains as citation losses. These results suggest that document-derived approximations of user intents can improve visibility without relying on explicit query inputs.”

Next, let’s break down the paper’s key vocabulary terms:

• Query-Implied Optimization (QI-GEO): A document-centric GEO method that infers potential user intent from the document itself rather than starting with known or generated queries.
• Generative Engine Optimization (GEO): Techniques designed to improve a document’s visibility and likelihood of being cited in generative search responses.
• Document intent space: The broader set of possible information needs implied by the topics, entities, and relationships contained within a document.
• Semantic expansion: Expanding the concepts found in a document to discover closely related information that may be relevant but missing.
• Knowledge graph: A structured network of entities and relationships used by QI-GEO to represent and expand the meaning of a document.
• Coverage map: A representation of the entity-relation pairs already covered by a document, used to distinguish existing information from genuine gaps.
• Informational gap: A relevant concept that is implied by the document’s semantic structure but is not currently covered in the content.
• Convergence-based scoring: A method for prioritizing gaps based on how many different entities in the document independently point toward the same missing concept.
• Fixed-retrieval setting: An evaluation setup where the candidate documents remain unchanged, allowing researchers to measure the effect of content edits without changes in retrieval.
• Position-Adjusted Word Count (PAWC): A GEO visibility metric that considers both how much of the generated answer is attributed to a source and where its citations appear.

Esta bien. Let’s take a deeper look at the research paper’s contents now.

If you want to follow along, you can grab a PDF or the HTML version on arXiv.

The paper has seven main sections.

We’ll focus on the most relevant ones below.

Introduction

“As users increasingly interact with search systems through complete questions rather than keywords, visibility now depends not only on ranking in search results but also on being selected and cited within generated answers,” write the authors. “As a result, traditional SEO practices alone may be insufficient to maintain visibility.”

“The foundational GEO work introduced the concept of visibility in generative outputs and proposed objective and subjective impression metrics for evaluating it,” they write. “It also demonstrated that even simple content modifications can significantly affect visibility.”

They continue: “most existing methods rely on explicit queries or query-derived signals to guide optimization. In practice, content creators rarely know the full range of questions users may ask about a topic, making query-dependent optimization difficult in black-box generative search systems.”

Next they introduce QI-GEO:

“Queries are often used as a proxy for user intent. In this work, we propose Query-Implied Optimization (QI-GEO), a framework that seeks to infer user intent directly from the document rather than relying on explicit query inputs. The key idea behind QI-GEO is that documents contain entities and relationships that capture their underlying semantics and reflect the information users may be interested in. By representing these relationships in a structured form and expanding them using external knowledge, QI-GEO approximates the broader semantic space surrounding a document without requiring query generation.

During this expansion process, concepts emerge that are closely related to the document’s topic but are not explicitly covered in its content. These concepts can be viewed as potential informational gaps that may limit a document’s ability to address diverse user information needs. QI-GEO leverages these gaps as opportunities for optimization. By identifying and incorporating relevant missing information, QI-GEO improves content coverage while grounding optimization in the document’s semantic structure rather than generated queries. This provides a more stable representation of user intent and helps improve visibility across unseen queries.”

Related Works

“GEO methods have evolved from rule-based rewriting to intent-aware and multi-query optimization,” they write.

Next they talk about current practices:

“User intent is a central component of GEO because visibility depends on how well a document aligns with user information needs. Early GEO approaches optimize content with respect to known queries. More recent methods generate related queries or query variants to approximate a broader intent space.”

“Although these methods differ in how intent is represented,” they write, “they generally treat queries as the primary mechanism for approximating user needs.”

“Since content creators rarely know the complete space of possible user queries, there remains a need for methods that can infer intent directly from the document itself,” they write. “This gap motivates the document-centric approach explored in this work.”

Methodology

They give an overview of QI-GEO, which is worth exploring in full, along with Figure 1 below:

“Query-Implied Optimization (QI-GEO) is a document-centric GEO framework that improves visibility in generative search engines without relying on explicit query inputs. Instead of optimizing content for known or generated queries, QI-GEO attempts to infer user intent directly from the document’s semantic structure. The central idea is that entities and relationships present in a document reflect the information needs associated with its topic. By expanding these relationships using external knowledge, QI-GEO approximates the broader semantic space surrounding the document and identifies concepts that are relevant but not explicitly covered.

Figure 1 illustrates the overall workflow of QI-GEO. The framework first extracts semantic relationships from the document and grounds them in a knowledge graph. The resulting graph is then expanded to discover related concepts and construct a coverage map of the document’s existing semantic content. Candidate informational gaps are ranked according to their relevance to the document’s topic, and the most promising gaps are converted into targeted edits. These edits are incorporated into the document to improve coverage of potential user information needs while preserving the original content and structure.”

The complete methodology is worth reading in full from the paper.

Results and Discussion

The results section is detailed and has valuable figures. That said, for the sake of brevity, I’ve summarized the key findings below:

• Citation coverage improved: QI-GEO increased the share of queries where the target document was cited by 10.5% in the single-query setting and 8.3% in the multi-query setting. Citation gains also outnumbered citation losses by roughly 2:1, suggesting the improvement was broadly distributed rather than driven by a few outliers.
• Source visibility increased: Using Position-Adjusted Word Count (PAWC), QI-GEO improved visibility by 12.6% for single queries and 15.9% for multi-query evaluation. The optimized documents were not just cited more often but they also tended to appear earlier and contribute more content to the generated answer.
• Performance was fairly consistent: QI-GEO achieved a 70% win-tie rate, meaning the optimized document performed at least as well as the original for about 7 in 10 queries. Downside risk was low, so when optimization hurt visibility, the losses were generally small.
• Subjective quality scores improved: An LLM judge rated optimized documents higher across factors such as relevance, influence, uniqueness, diversity, click likelihood, position, and volume. Scores improved by 17.6% in the single-query setting and 17.1% in the multi-query setting.
• Lower-visibility documents benefited the most: The biggest shift occurred among documents that originally received weak subjective visibility scores, with more of them moving into moderate and higher visibility ranges after optimization.
• Subjective results were also stable: QI-GEO reached an 81% win-tie rate on subjective scores, meaning more than 4 in 5 optimized documents were rated as good as or better than the originals.
• The gains generalized across different phrasings: Across five semantically related query variants, 66% of documents had a net-positive average gain, and 73.7% improved or tied on at least 3 of the 5 phrasings.
• Positive gains were larger than negative losses: When QI-GEO helped, the average gain was +0.1634 and when it hurt, the average loss was only -0.0629.
• Most documents benefited on at least one query variation: 87.5% of documents gained visibility on at least one phrasing of the same underlying information need.
• Overall takeaway: The results support the idea that improving a document based on its own semantic gaps can increase citation visibility and source prominence across multiple query formulations, without requiring explicit query targeting. The important caveat is that the study kept the retrieval pool fixed, so it tests post-retrieval visibility, not whether QI-GEO improves retrieval itself.

Conclusion and Future Work

“This paper introduced Query-Implied Optimization (QI-GEO), a query-independent framework for Generative Engine Optimization. Existing GEO methods typically rely on explicit queries or query-derived signals to guide document optimization. In contrast, QI-GEO models user intent directly from document semantics. By extracting entities and relationships, expanding them through a knowledge graph, and identifying informational gaps, QI-GEO discovers concepts that may be relevant to user information needs without requiring access to future queries.

Experiments on GEO-Bench and ExtendedGEOBench show that QI-GEO improves citation coverage, source visibility, and subjective impression scores compared to the original documents. The results further demonstrate that these improvements remain effective across semantically related query variations, suggesting that document-centric intent modeling can serve as a viable alternative to query-based optimization strategies.

Several directions remain for future work. First, richer knowledge graph expansion strategies could improve the quality and coverage of discovered informational gaps. Second, future systems could jointly model multiple inferred intents during optimization rather than treating candidate gaps independently. Finally, while this work uses a general-purpose knowledge graph, domain-specific knowledge graphs may provide more precise semantic relationships and enable stronger optimization in specialized domains such as law, finance, healthcare, and geospatial information. These directions offer promising opportunities for improving the effectiveness and generality of query-independent GEO systems.”

Knowing what we do now, why should SEO/GEO professionals care about “Query Implied Generative Engine Optimization”?

QI-GEO proposes a different way to think about optimization. Instead of starting with a query and adapting the content to it, it starts with the document to identify the information needs its own semantic structure implies.

Given that, this paper matters for a few reasons:

• It challenges query-first optimization: Most SEO and GEO workflows depend on known keywords, prompts, or generated query variants. QI-GEO tries to infer likely user intent directly from the entities and relationships already present in the document.
• It offers a new way to find content gaps: Rather than asking “What keywords am I missing?”, QI-GEO asks “What information is strongly implied by this document but not actually explained?” The method uses semantic expansion and a coverage map to identify those missing concepts.
• The identified gaps appear to matter for AI visibility: In the study, optimized documents saw citation coverage increase by 10.5% in the single-query setting and 8.3% in the multi-query setting. Source visibility also improved by 12.6% to 15.9%, depending on the evaluation setup.
• The gains were not tied to one exact query phrasing: Across five semantically related query variants, 66% of documents had a net-positive mean gain, and 73.7% improved or tied on at least three of five phrasings. That suggests the method may help content perform across a broader information need, not just a single prompt.
• It reframes topical completeness: For SEO/GEO practitioners, the practical idea is that a page may become more useful to generative systems by covering concepts that naturally belong within its semantic neighborhood, rather than simply expanding around a list of keywords or prompts.

Outro

QI-GEO is interesting because it suggests that GEO may not always have to begin with a query. By analyzing what a document already says, identifying semantically related information it’s missing, and filling in those gaps, the researchers were able to improve citation visibility across multiple query formulations.

For SEO/GEO practitioners, this research introduces a potentially useful new layer of content optimization: semantic gap analysis alongside, rather than instead of, traditional query research.

I hope you’ve enjoyed this edition of Hamsterdam Research! 🐹

Feel free to comment below or contact me with your feedback.

Stay tuned for another new article, hopefully next week, or check out related research posts below.

Until next time, enjoy the vibes:

Thanks for reading. Happy optimizing!