# From Experience to Skill: Multi-Agent Generative Engine Optimization via Reusable Strategy Learning

Source: [From Experience to Skill: Multi-Agent Generative Engine Optimization via Reusable Strategy Learning](<https://www.sota2.com/research/paper/69e85bf67d0172140ec81283>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-23T05:00:12.895Z

Authors: Beining Wu, Fuyou Mao, Jiong Lin, Cheng Yang, Jiaxuan Lu, Yifu Guo, Siyu Zhang, Yifan Wu, Ying Huang, Fu Li

Published: 2026-04-21T14:39:24.000Z

## Abstract

Generative engines (GEs) are reshaping information access by replacing ranked links with citation-grounded answers, yet current Generative Engine Optimization (GEO) methods optimize each instance in isolation, unable to accumulate or transfer effective strategies across tasks and engines. We reframe GEO as a strategy learning problem and propose MAGEO, a multi-agent framework in which coordinated planning, editing, and fidelity-aware evaluation serve as the execution layer, while validated editing patterns are progressively distilled into reusable, engine-specific optimization skills. To enable controlled assessment, we introduce a Twin Branch Evaluation Protocol for causal attribution of content edits and DSV-CF, a dual-axis metric that unifies semantic visibility with attribution accuracy. We further release MSME-GEO-Bench, a multi-scenario, multi-engine benchmark grounded in real-world queries. Experiments on three mainstream engines show that MAGEO substantially outperforms heuristic baselines in both visibility and citation fidelity, with ablations confirming that engine-specific preference modeling and strategy reuse are central to these gains, suggesting a scalable learning-driven paradigm for trustworthy GEO. Code is available at https://github.com/Wu-beining/MAGEO

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Generative Engine Optimization on MSME-GEO-Bench](<https://www.sota2.com/research/sota/generative-engine-optimization-on-msme-geo-bench.md>): Generative Engine Optimization; MSME-GEO-Bench; WLV: 5.3.

- [Generative Engine Optimization on GEO-Bench](<https://www.sota2.com/research/sota/generative-engine-optimization-on-geo-bench.md>): Generative Engine Optimization; Geo-Bench; WLV: 4.81.

- [Generative Engine Optimization on MSME-GEO-Bench Qwen-3 Max (test)](<https://www.sota2.com/research/sota/generative-engine-optimization-on-msme-geo-bench-qwen-3-max-test.md>): Generative Engine Optimization; MSME-GEO-Bench Qwen-3 Max (test); WLV: 3.84.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2604.19516>)

[PDF](<https://arxiv.org/pdf/2604.19516.pdf>)