Human Pioneers
NEWS AGENCY

World Brand AI Visibility Index

World Brand AI Visibility Index

WBVI :2026

World Brand AI Visibility Index

World Brand AI Visibility Index

Compiling Organization: World Intelligence Organization (WIO), CNBNTV (China New Vision Internet Television Co., Ltd.)First Release Date: June 2026

Release Cycle: Quarterly data collection, annual comprehensive release
Language Versions: Chinese / English / French / Spanish

I. Introduction

1.1 Background

Generative artificial intelligence is reshaping the global information distribution landscape at an unprecedented pace. According to Statista, as of mid-2026, the global monthly active users of generative AI have exceeded 2 billion. Research by iResearch indicates that AI search is shifting from traditional search engines to answer-oriented models. Before making consumption decisions, users increasingly ask questions directly to AI large models such as ChatGPT, Gemini, Ernie Bot, and DeepSeek. Whether a brand appears in AI answers and how it is presented are becoming core variables determining its global market competitiveness.

In this context, a brand's "visibility" in AI large models—the degree to which it is mentioned, recommended, and accurately described—has formed a new dimension of brand digital assets. However, there is currently a lack of a systematic, objective, and cross-cultural evaluation tool globally to measure this new form of brand presence in the AI ecosystem.

1.2 Purpose

The World Brand AI Visibility Index (WBVI) aims to:

• Quantitatively assess the comprehensive visibility level of representative brands across major global industries in mainstream AI large models;
• Cross-culturally compare visibility differences and trends among brands from different countries, industries, and within the AI knowledge system;
• Dynamically track temporal changes in brand AI visibility, identifying cognitive risks and growth opportunities;
• Provide a public good by offering an open, transparent, and reproducible global benchmark dataset of brand AI visibility for academic research, business decision-making, investment evaluation, and brand strategy formulation.

1.3 Theoretical References

The compilation of this index references the following international standards and academic achievements:

• United Nations Statistics Division, "Handbook on Statistical Indicators" (2015)
• OECD, "Handbook on Constructing Composite Indicators" (2008)
• ISO 10668:2010, "Brand Valuation – Requirements for Monetary Brand Valuation"
• ISO/IEC 22989:2022, "Artificial Intelligence – Concepts and Terminology"
• Pang Pei (2025) "Media-type GEO" theory – revealing AI's preference mechanism for authoritative sources
• Pang Pei (2026) "AI Brand Equity (AIBE)" theory – constructing a dimensional framework for brands to build cognitive assets in the AI era
• Pang Pei (2026) "AI Influence Model" – explaining the deep mechanism by which AI recommendations become the dominant logic of influence

1.4 Relationship with GEO100 and CBVI

WBVI is the global index application of the GEO100 certification standard and the global expansion of CBVI (China Brand AI Visibility Index). The relationship among the three is as follows:

• GEO100: A certification standard for individual brands, providing tiered ratings (L1-L5)
• CBVI: A visibility index for Chinese brands, focusing on the performance of Chinese market brands in AI
• WBVI: A global version of the visibility index, covering brands from multiple countries worldwide, providing cross-cultural and cross-lingual comparable metrics

II. Definition and Scope

2.1 Index Definition

World Brand AI Visibility Index (WBVI) is a composite statistical index that comprehensively measures the level of recognition, mention, recommendation, and presentation of representative brands across major global industries in content generated by mainstream generative AI large models.

2.2 Core Construct: Brand AI Visibility

Brand AI Visibility comprises three levels:

• Presence Level: Whether the brand is included and recognized by AI – basic visibility
• Mention Level: The frequency, context, and language range in which the brand appears in AI responses – breadth of visibility
• Presentation Level: The accuracy of information, depth of description, and emotional tendency when the brand is mentioned by AI – quality of visibility

2.3 Coverage Scope

• Brand Sample: The initial phase covers no fewer than 800 representative brands from at least 20 major economies globally, spanning eight industry sectors: Technology, Automotive, Consumer Goods, Luxury Goods, Finance, Pharmaceuticals, Energy, and Industrial Manufacturing. Sample selection criteria are detailed in Chapter 5.
• Model Scope: Covers no fewer than 15 mainstream generative AI large models globally:
  • Global General-Purpose: OpenAI GPT-5, Google Gemini 2.0, Anthropic Claude 4, Meta Llama 4
  • China: Baidu ERNIE Bot 4.0, ByteDance Doubao, Alibaba Tongyi Qianwen 2.5, DeepSeek V3/R1
  • Europe: Mistral Large (France), Aleph Alpha (Germany)
  • Middle East: Falcon 3 (UAE)
  • East Asia: Naver HyperCLOVA X (South Korea), Rakuten AI (Japan)
• Language Scope: The initial phase covers seven languages: Chinese, English, French, Spanish, Arabic, Japanese, and Korean. Subsequent expansion will include Russian, German, Portuguese, and Hindi.
• Time Span: Using 2026 as the base year, published annually.

3. Index Architecture and Indicator System

3.1 Index Hierarchy Structure

WBVI adopts a four-level hierarchical structure:

• Overall Index (WBVI): Reflects the comprehensive level of brand AI visibility.
• Sub-Indices (3): Breadth Visibility Index, Quality Visibility Index, Stability Index.
• Sub-Dimensions (8): Each sub-index includes 2-3 sub-dimensions.
• Basic Indicators (18): Constitute the smallest unit of data collection.

3.2 Indicator Framework

Table 1 – WBVI Indicator System

Weight Determination Method: Sub-index and sub-dimension weights are determined using the Delphi method, inviting no fewer than 25 global experts in AI technology, brand management, and cross-cultural communication to participate in scoring. Weights are reviewed every two years.

3.3 Score Standardization

Raw values of each basic indicator are mapped to a 0–100 range using Min-Max Normalization:

X_normalized = (X_original - X_min) / (X_max - X_min) × 100

Where X_min and X_max are the minimum and maximum values of the indicator among all evaluated brands in the current period.

For cross-year comparisons, 2026 is used as the base year (WBVI=100), and subsequent years are linked using the chain index method.

IV. Data Collection and Processing Methods

4.1 Question Set Design

To comprehensively assess brand AI visibility, a multi-dimensional standardized question set is designed:

(I) Breadth Visibility Question Set (Approximately 80 Questions)

(II) Quality Visibility Question Set (Approximately 70 Questions)

(III) Stability Detection Question Set (Approximately 30 Fixed Tracking Questions)

30 core questions are selected from the above question sets as fixed tracking questions to be collected in each period for time-series stability analysis.

Each brand’s question set can be individually supplemented based on its industry and target market. The total question set shall include no fewer than 180 standard questions, with multilingual simultaneous translation, and no more than 20% updated annually.

4.2 Data Collection

• Tool: Queries are sent to target AI large models via standardized API interfaces, and complete response texts are automatically recorded.
• Frequency: Two rounds of data collection are conducted annually (mid-year and year-end) to reduce response fluctuations caused by model updates and current events.
• Replication: Each question is independently queried 3 times on each model (with intervals of no less than 24 hours), and the average of the 3 results is taken.
• Data Volume: Single-round collection volume = 15 models × 180 questions × 800 brands × 7 languages × 3 repetitions. Stratified sampling is used (approximately 40 core questions per brand), resulting in a total effective data volume of approximately 10 million response records per round.
• Ethical Compliance: All collection targets only publicly available information and does not involve trade secrets or user privacy data.

4.3 Data Processing

• Named Entity Recognition (NER): Multilingual NER models are used to extract brand names and related entities from response texts. Dedicated fine-tuned models are used for each language, with a manual spot-check rate of no less than 10%.
• First Mention Determination: The first industry-related brand name appearing in the AI response text is recorded.
• Information Accuracy Determination: A standard fact database for each brand is established (including brand full name, country of origin, main products, founding year, core technical parameters). AI responses are compared against this standard database. The fact database is updated quarterly.
• Source Authority Determination: Source URLs or publication names cited in AI responses are extracted and matched against a predefined "WBVI Authoritative Source Database," scored according to their level.
• Sentiment Analysis: A multilingual three-class sentiment model (positive/neutral/negative) is used, with dedicated fine-tuned models applied for each language.
• Cross-Language Calibration: Cross-language alignment analysis is performed on the performance of the same brand across different language AI models to eliminate language bias.

4.4 Outlier Handling

• When a single collection result deviates from the mean by more than 3 standard deviations, a supplementary query is conducted and the outlier is replaced.
• If a specific AI model experiences a systemic failure or regional unavailability, the weight of that model for that round is temporarily redistributed to other similar models, with a public explanation in the report.

V. Brand Sample Selection

5.1 Sample Selection Principles

• Global Representation: Covers major global economies, selected comprehensively based on GDP, brand international influence, and industry representation.
• Industry Diversity: Encompasses eight major industry sectors: Technology, Automotive, Consumer Goods, Luxury Goods, Finance, Pharmaceuticals, Energy, and Industrial Manufacturing.
• Brand Maturity: Prioritizes brands with global recognition or significant leading positions in regional markets.
• Data Measurability: Brands must have sufficient information in mainstream AI models to support data collection needs for standardized question sets.
• Dynamic Adjustment: The brand sample is reviewed every two years; brands that have been delisted or have significantly declined in influence are removed, and emerging representative brands are added.

5.2 Initial Coverage Regions and Sample Allocation

The complete list is available in Appendix A. The list is nominated by a global expert committee and finalized through multiple rounds of selection.

5.3 Dynamic Adjustment Mechanism

• New Inclusions: Brands whose influence has significantly increased and internationalization has notably improved during the previous evaluation cycle may be included in the next year's sample after review by the expert committee.
• Exit Mechanism: Brands that lose independence due to acquisition/merger, or whose market position has severely declined, may be removed from the sample. Historical data for exited brands will be retained.

VI. Index Calculation

6.1 Calculation Steps

Step 1: Standardization of Basic Indicators. Standardize the raw values of each brand for each basic indicator into a score of 0–100 using the Min-Max method.

Step 2: Calculation of Sub-dimension Scores. Each sub-dimension score = the arithmetic mean of the standardized scores of the basic indicators under that sub-dimension.

Step 3: Calculation of Component Indices. Each component index = the weighted arithmetic mean of the sub-dimension scores under that component.

Step 4: Calculation of the Total Index. WBVI = weighted arithmetic mean of the three component indices.

WBVI = 0.45 × S_Breadth + 0.40 × S_Quality + 0.15 × S_Stability

Step 5: Cross-Year Linking. Using 2026 as the base year (WBVI=100), subsequent years are calculated using the chain index method.

6.2 Published Content

• WBVI Global Overall Ranking: Comprehensive scores and rankings for 800 global brands.
• Regional Rankings: AI visibility rankings for brands in seven regions.
• Industry Rankings: AI visibility rankings for brands in eight industries.
• Special Rankings: Top 100 Quality Visibility, Top 100 Breadth Visibility, Top 100 Stability.
• Language Rankings: AI visibility rankings for brands in Chinese, English, and regional languages.
• Annual In-Depth Report: Trend analysis, cross-cultural comparisons, and typical case studies.

VII. Index Release and Dissemination

7.1 Release Cycle

• Annual Comprehensive Release: The WBVI global overall ranking and sub-rankings for the previous year are released in the first quarter of each year.
• Quarterly Brief: A dynamic tracking brief on AI visibility for key brands is released quarterly.

7.2 Release Channels

• Real-time updates on the official index website, providing visual charts and data downloads (open access).
• Joint release with major global financial, technology, and marketing media.
• Special releases at global platforms such as the World Economic Forum and the Cannes Lions International Festival of Creativity.
• Publication of methodological papers in academic journals.

7.3 Communication Ethics

• The index release is based on objective data and does not engage in subjective ranking hype.
• Brands have the right to know and query their own data.
• Clear statement: The WBVI measures a brand's visibility in AI-generated content and is not equivalent to a comprehensive evaluation of the brand's overall strength or market position.

VIII. Quality Control

8.1 Data Quality Assurance

• Reliability Test: At least 10% of samples from each collection round are manually reviewed. For sentiment classification, Cohen's Kappa ≥ 0.80 is required; for information accuracy determination, human-machine consistency ≥ 90% is required.
• Stability Test: Monitor baseline fluctuations of the same question with the same model over different time periods; abnormal fluctuations trigger a re-check.
• Fact Database Update: The standard fact database for each brand is updated quarterly.
• Multilingual Cross-Validation: Monitor the consistency of core facts for the same brand across AI models in different languages; calibrate promptly if systematic bias is detected.

8.2 Index Revision Policy

• Regular Revision: Review the indicator system, weights, and brand sample every two years; revisions are announced at least 60 days in advance.
• Major Revision: Initiate a temporary revision procedure with full explanation when there is a fundamental change in the technical architecture of AI large models.

8.3 Independence Statement

The index compilation institution is independent of any evaluated brand or AI model provider. It does not accept targeted funding to influence specific brand rankings. An independence statement and funding source report are published annually.

IX. Interpretation and Usage Guide

9.1 Key Points for Index Interpretation

• The WBVI measures a brand's "visibility" in AI large models, which is not equivalent to the brand's awareness or favorability among consumers. The three are related but not identical.
• High breadth does not equal high favorability—it must be assessed in conjunction with quality visibility scores. High mention rates accompanied by low information accuracy or negative sentiment may indicate cognitive risks for the brand.
• Stability reflects a brand's AI visibility resilience against model updates and market changes, serving as an important indicator of brand AI asset quality.

9.2 Usage Scenarios

X. Limitations

• Model Representativeness Limitation: The index only reflects performance within the tested models and cannot cover all AI models.
• Response Randomness: Generative AI outputs have inherent randomness. Although multiple sampling reduces its impact, it remains a source of measurement error.
• Language Coverage Limitation: Brands not covered in the initial language set may be systematically underestimated due to insufficient corpus.
• Time Sensitivity: AI models iterate rapidly. The index reflects the cross-sectional state within the release cycle, and trend analysis requires multi-period data.
• Industry Differences: B2B brands typically have lower visibility in consumer-oriented questions. Cross-industry comparisons need to consider industry characteristics.

XI. Appendix

Appendix A: WBVI Inaugural Brand Sample List (Excerpt)

(Approximately 800 brand names and their countries, listed by region and industry, omitted.)

Appendix B: Standardized Question Set (Full English Version Example)

(At least 180 standardized questions listed by breadth, quality, and stability, omitted.)

Appendix C: Brand Standard Fact Database Field Description

Appendix D: WBVI Authoritative Source Library (Excerpt)

(Refer to GEO100:2026 Appendix B, omitted.)

Appendix E: Delphi Method Weight Determination Process

(Detailed description of global expert selection criteria, scoring rounds, Kendall's W consistency test method, omitted.)

Standard References

[1]United Nations Statistics Division. (2015). Handbook on Statistical Indicators.

[2]OECD.(2008). Handbook on Constructing Composite Indicators: Methodology and User Guide.
[3]ISO 10668:2010. Brand valuation – Requirements for monetary brand valuation.

[4]ISO/IEC 22989:2022. Artificial intelligence – Concepts and terminology.

[5] Pang Pei. (2025). Media-type GEO: A New Paradigm for Building Brand Trust Assets in the AI Era

[6] Pang Pei. (2025). Trust Verification, Cognitive Parasitism, and Dynamic Decay: Theoretical Extension and Model Deepening of AI Brand Equity (AIBE).

[7] Pang Pei. (2026). From Mind Share to Cognitive Agency: Theoretical Construction and Dominant Logic Research of the AI Influence Model.

[8] Statista. (2026). Global generative AI user statistics.

Compiling Organization
World Intelligence Organization (WIO), China New Vision Internet Television Co., Ltd. (CNBNTV)Release Statement
This index is compiled in accordance with internationally accepted statistical standards. Copyright reserved. Reproduction must cite the source. The index results do not constitute a comprehensive evaluation of any brand's value, nor are they the sole basis for investment or business decisions. The compiling organization publishes an annual independence statement.