Frost & Sullivan, under the guidance of the China Advertising Association, released '2026 China GEO Marketing Practices and Compliance Development Report'

AI has penetrated all aspects of consumer decision-making, and its role is most prominent when consumers compare and evaluate products. Survey data shows that 61.7% of respondents believe AI has played a significant or decisive role in the evaluation stage, with the highest proportion in the five stages of cognition, evaluation, purchase, sharing, and after-sales service. After entering the purchase stage, this proportion drops to 51.8%, and real-time information such as price, inventory, channels, and after-sales policies become the determining factors again.

AI is also changing consumers' views of brands. 80.8% of respondents said that AI has influenced their views or choices about brands: 39.8% became more confident or continued to purchase some large brands, 29.6% started to focus on new and niche brands, and 11.4% reduced or abandoned consideration of some brands. It can be seen that AI can both consolidate the trust accumulated by established brands and provide opportunities for new brands to be discovered. Whether a brand can be accurately understood and clearly compared by AI depends on whether it can enter consumers' shortlist.

Source: Orion, Frost & Sullivan

The changes in user behavior are already reflected in businesses' investments. Budgets originally used for SEO, content marketing, and brand public relations are gradually being reallocated to GEO. Frost & Sullivan's calculations show that the scale of the Chinese GEO market was approximately 700 million yuan in 2024, increased to 5.7 billion yuan in 2025, and is expected to reach approximately 13.7 billion yuan in 2026, and further increase to 48.8 billion yuan in 2029. Meanwhile, the global GEO market size is expected to grow from 47.5 billion yuan in 2025 to 42.79 billion yuan in 2029.

Source: Frost & Sullivan

Users recognize AI's capabilities, but they still retain concerns about the credibility of AI information. In a comparison of ten capabilities between AI tools and traditional search, AI has obvious advantages in complex demand processing, multi-modal processing, and personalized adaptation. The average score for complex demand processing is 4.59 points, while traditional search only gets 2.63 points. Traditional search remains leading in information credibility, timeliness, and accuracy of results. In terms of information credibility, traditional search scores 4.00 points, while AI scores 3.53 points. When choosing an AI tool, the accuracy of results is the most mentioned consideration, with a mention rate of 41.8%.

Therefore, many users are accustomed to using AI to integrate information first, and then verifying through search. This puts higher requirements on GEO services: the frequency of a brand's appearance in AI answers is just a starting point. Whether the information can be accurately understood, the reliability of the source, and whether the answers from different platforms and times are consistent determine the true value of the service.

Source: Orion, Frost & Sullivan

The report suggests that credible GEO should be based on verifiable facts and consist of four links: content, source, dissemination, and evaluation. Credible content is the starting point. Brands need to organize their official websites, product materials, parameters, and cases clearly to form a unified factual draft, ensuring that each key piece of information specifies the basis, scope, and source. Content generated by AI should be marked according to regulations. Credible sources answer the question of why AI accepts the information. Sources should be classified and graded based on qualifications, past authenticity, traceability, and external evaluations, and verified through multiple independent sources to avoid relying on a single channel.

Credible dissemination allows brands to use different expressions on different platforms, but the core facts, sources, and versions must be consistent. Information should be updated or withdrawn in time when it changes, and manipulation behaviors such as bot spam, machine mass distribution, and content poisoning should be avoided. Credible evaluation checks whether the previous three links are truly effective and reports the found problems to the updates of content, source, and dissemination, forming a continuously operating closed loop.

Source: Frost & Sullivan

Credible evaluation determines whether the effectiveness of GEO can be proven. Generative answers themselves are random; model version, question style, testing time, and region will affect the results. A screenshot or the number of mentions on a single platform cannot indicate the true effectiveness of the service. The report suggests measuring effectiveness from three aspects: whether the brand information itself is accurate, complete, and timely; whether the visibility of the brand in AI answers, recommended positions, cited sources, and cross-model performance have improved; and whether the information source and dissemination methods are compliant.

Evaluation methods also need to be standardized. Evaluation should design a set of questions around the brand, product, usage scenario, competitor comparison, and potential risks. Complete answers should be collected from multiple models and at multiple times, establishing a baseline, and then comparing the changes before and after optimization. The prompts used for testing, model version, answers, citations, and time should be fully recorded, including failed samples, to ensure that each round of results can be traced and reviewed. When analyzing effectiveness, it is also necessary to distinguish which changes come from optimization work, which come from model updates, public sentiment changes, or competitor actions, and avoid counting natural fluctuations of the platform as service results.

The report also recommends regularly re-evaluating the evaluation results and introducing independent third-party re-verification. Facts missing, inconsistent statements, or abnormal citations found in the evaluation should be promptly fed back to adjustments of content and source. Only when the evaluation continues, can the results of GEO be settled into a stable and credible perception of the brand in AI answers.

Source: Frost & Sullivan

The advantages established by the number of channels and short-term visibility are weakening, and the gap between service providers will be more reflected in four capabilities: first, understanding how different AI platforms capture, cite, and recommend information, and designing a set of questions that are close to real needs based on industry experience; second, organizing brand information into structured content that AI can understand easily, and integrating authoritative, traceable external sources; third, continuously monitoring the answers and citations from multiple platforms to find the reasons for effect changes; fourth, establishing a review mechanism for content authenticity, advertising labels, AI-generated content labels, and data compliance to manage compliance risks.

Source: Frost & Sullivan

Credible GEO methods are being practiced in different types of enterprises. The report evaluates the applied cases from six dimensions and 24 secondary indicators: industry representation, product and service capabilities, cases and achievements, compliance and credibility, method innovation, promotion and demonstration. Cases and achievements are the key dimensions, and compliance and credibility are the basis of evaluation. The eight selected enterprises include central media, internet platforms, marketing technology enterprises, and professional service institutions. The cases cover fields such as liquor, jewelry, education, housekeeping, MATERNAL AND INFANT, fast-moving consumer goods, high-precision positioning, etc. The following introduces them in the order they are included in the report, without indicating a ranking.

Source: China Advertising Association, Frost & Sullivan

Xinhua Netis a central key online media hosted by Xinhua News Agency. Its GEO business is centered on the Xinhua GEO intelligent platform. Relying on a matrix of vertical intelligent agents and a dual-engine expert model, it provides services in six steps: intention analysis, data analysis, content structuring, semantic optimization, multi-platform adaptation, and monitoring and iteration. The content goes through three levels of review: machine screening, manual review, and brand confirmation. The source is traceable, and the process is documented. In the practice of a liquor brand, the project started in August 2026. After 75 days of optimization, the brand's mention rate in AI answers increased from 34.40% to 83.19%, the first mention rate increased from 3.15% to 34.44%, and it rose from fourth place to first place among 5 monitored brands, and the negative rate of brand terms decreased by 8.6 percentage points.

SohuIn October 2025, Sohu Youjian GEO was launched to provide AI growth services for generative AI search. This service uses a fully self-developed technology stack, building a credible knowledge base through multi-agent scheduling and knowledge graph construction. When organizing content, it adopts the order of verified facts, local authoritative materials, and publicly available search results. It forms a service closed loop through monitoring, diagnosis, optimization, delivery, and precipitation. In the service provided to a century-old jewelry brand, Sohu Youjian GEO locked 9 core scenarios such as marriage, gift-giving, and investment on 6 major AI platforms including Doubao and Qianwen. In 3 months, more than 25 original credible contents were released, the first-placement rate of the brand increased by more than 3 times, and the Top3 recommendation rate increased by more than 30 percentage points.

360 Zhijianis a credible GEO product solution launched by 360 Group for commercialization. Relying on the group's more than 20 years of search and technology security experience, it provides full-chain services from AI search exposure, lead acquisition, to consultation conversion. 360 Zhijian uses the customer-verified knowledge base as the main basis for content. It goes through three levels of review: manual review, system verification, and pre-release review. It also supports continuous monitoring and snapshot retention of more than 3,000 word items. In the service provided to a national extracurricular tutoring institution, the average ranking of the institution in the DeepSeek tutoring institution recommendation entries increased from ninth place to second place, the positive rating ratio increased from 63% to 100%, and the number of sources expanded from 6 to 22. In the overseas service provided to a smart security brand for the US market, the mention rate of the brand in ChatGPT target questions increased from 0% to 30 within 15 days.

Weimai Xingqiis a GEO system developed by Weimai. It began providing services externally in December 2025, delivered as a SaaS system plus standardized operation services. Brand materials, scenario questions, content generation and review, channel release, and AI monitoring were all completed in the same system. Each piece of content could be traced to the corresponding knowledge source, version, and release record. Information not in the knowledge base was not written. From January to March 2026, Weimai Xingqi optimized GEO for Guangzhou household service chain brand Jinbei Household. Before the project started, users could not find this brand when asking about Guangzhou household on the AI platform. During the project period, its brand AI visibility increased from 0% to 79%, and its AI ranking increased from 99th to 1st place. The offline reputation was also available in AI search.

Titanium Magnesium AIis an AI marketing technology enterprise strategically invested by Focus Media. It has developed three engines: iris, Zhi Shu, and adaptive algorithms. GEO services cover intent recognition, knowledge graph, content creation, full-channel distribution, and AI search performance monitoring. Before content release, it must go through a five-level compliance review. Large and medium-sized brands use customized solutions, while small and individual merchants can use the standardized GEOForce intelligent agent. In the three-month service provided to a domestic high-end infant formula brand, the average recommendation rate of the brand in DeepSeek, Tencent Yuanbao, Doubao, and Kuaiku increased from about 20% to over 96%. The recommendation ranking remained in the top three of the same category, and the e-commerce sales increased by about 40% month-on-month.

Liuo DigitalAs a subsidiary of the A-share listed company Liuo Shares, its GEO product LEO ARO (Intelligent Response Optimization) connects precise AI deployment with generative engine optimization to form a capability chain. The service is divided into four stages: current status diagnosis, content infrastructure, information distribution, and performance monitoring. It builds a brand knowledge graph based on the customer's existing official website, press releases, white papers, and product documents. During ongoing operations for a beer consumer brand, the brand has a high visibility in AI responses, but niche products such as light dry, pure raw, and original juice have minimal exposure. Liuo Digital explores users' frequent questions on four major AI platforms, establishes dual-track dictionaries for products and brands, distributes content selectively, and continuously monitors to address the lack of exposure for niche products in AI Q&A.

Zhi Huowas established in 2011. It is an LANDSCAPING content integrated marketing service institution. Its GEO service uses the enterprise's official website as the core first-party source for the brand to the AI platform. It forms a complete path from credible content construction, brand knowledge base, official website AI transformation, authoritative source construction to continuous monitoring and review. It works with its self-developed Granking insight system, Pureploy G1 industry model, and content center. The content goes through five-level review, and effects are evaluated in six dimensions. From April to July 2026, Zhi Huo provided GEO optimization services for the satellite navigation enterprise Huace Navigation. It continued to operate on 952 keywords and 6 product lines on 5 major AI platforms. Huace Navigation's average ranking in DeepSeek increased from 26.5th to 11.6th, and in Doubao, it increased from 21.3th to 10.3th. All 6 product lines entered the top three of AI recommendations.

Deep AlignmentFocuses on GEO and AI information governance. Based on a full-stack self-developed three-layer AI native architecture, it provides three product lines: Noiseoff Brand GEO System, AI Data Link, and AI Official Website Link. It delivers in four stages: diagnosis, knowledge base construction, content generation and distribution, and public sentiment management and optimization. Content only calls for verified facts from the customer. After verification by the knowledge base and 30+ dimension checks, it is distributed. Effects are evaluated according to the KPIs agreed upon. In the annual GEO service provided to a global leading MATERNAL AND INFANT brand, Deep Alignment established a product-level AI monitoring and optimization matrix. The target scenario mention rate increased from 60% to 90%, the first-mention rate during the promotion period increased to 88%, and the public sentiment error rate decreased to 5%.

Frost & Sullivan's requirements for information quality are continuously increasing, and the competitive rules in the GEO industry are also changing accordingly. A brand's long-term value in AI solutions ultimately depends on three factors: whether the information is accurate, whether the source can be verified, and whether the results can be rechecked. This requires brands to prioritize factual accuracy over exposure volume, while service providers are responsible for the content and outcomes delivered. Platforms, industry organizations, and third-party institutions must work together to improve rules and evaluation mechanisms.

Frost & Sullivan will continue to monitor market trends and corporate practices in the GEO sector, working together with the China Advertising Association and industry stakeholders to promote credible GEO implementation, establishing a genuine, transparent, and responsible industry ecosystem, ensuring that brand information is more trustworthy, service outcomes are verifiable, and GEO creates long-term value for brands.