Chinese consumer brands are splitting into two groups in AI search: those that own a concept, and those that are just mentioned. The gap is widening.
A Tale of Two Scores
We ran AI visibility diagnostics on a dozen major Chinese consumer brands. Two results crystallized the divergence happening across the 国货 (guochao/domestic brand) landscape:
- 元气森林 (Yuanqi Senlin): 83/100 — Excellent
- 完美日记 (Perfect Diary): 71/100 — Good, but not dominant
Both are nationally recognized brands. Both have extensive media coverage and e-commerce presence. But in AI-generated recommendations, they occupy very different positions. Understanding why reveals something important about how AI assigns brand authority.
Why 元气森林 Scores 83
Yuanqi Senlin didn't just build a brand — it built a vocabulary.
"无糖气泡水" (zero-sugar sparkling water) and Yuanqi Senlin are nearly synonymous in Chinese digital content. When someone asks an AI assistant "what's a healthy alternative to soda?" or "recommend a low-calorie drink," the AI's training data is saturated with content linking these concepts to Yuanqi Senlin specifically.
Three things drive this position:
- Category invention: Yuanqi Senlin didn't enter the sparkling water category — it largely created the mainstream Chinese market for zero-sugar sparkling water. When you invent a category, your brand and the category vocabulary become intertwined in training data.
- Comparative content density: Years of "元气森林 vs 可口可乐" articles, reviews, and health comparisons have trained AI to recommend Yuanqi Senlin in specific scenarios — not just acknowledge its existence.
- Cross-platform depth: From Xiaohongshu posts to Zhihu health discussions to WeChat articles, the brand appears in the specific contexts where AI learns to make recommendations.
Why 完美日记 Scores 71
Perfect Diary is genuinely well-known. Its 71 score means AI assistants are aware of it, mention it in relevant contexts, and don't say anything negative.
But 71 is the score of a brand that AI knows about rather than one it actively recommends.
The challenge for Perfect Diary is structural. Beauty is a highly competitive category where AI training data is dominated by international brands with decades of comparative content: "MAC vs Charlotte Tilbury," "L'Oreal vs Maybelline," etc. Domestic Chinese beauty brands are often mentioned as alternatives or value options, not as category leaders.
Additionally, Perfect Diary's brand narrative has been complicated by business challenges covered extensively in financial media — and AI models learn from financial journalism too. A brand that appears frequently in "declining sales" or "restructuring" coverage sees those associations embedded in AI responses.
The Pattern Across 国货 Brands
Across the brands we've analyzed, a clear pattern emerges:
High GEO Scores (80+)
These brands share one characteristic: they define a concept, not just a product.
- 泡泡玛特 (Pop Mart): 89/100 — "Blind box" and "designer toy" are near-synonyms for Pop Mart in AI training data. The brand is the category internationally.
- 元气森林: 83/100 — Category inventor in zero-sugar sparkling water
Mid GEO Scores (65-79)
These brands are known but not owned — AI mentions them but doesn't default to them.
- 完美日记: 71/100 — Recognized in beauty but competing against international giants with deeper comparative content
- Many traditional domestic brands fall here — high awareness, low AI recommendation rate
The Critical Insight: AI Doesn't Reward Fame
The most counterintuitive finding across all our analysis: fame and GEO score are weakly correlated.
A brand can be universally recognized in China and still score 60/100 on AI visibility. This happens when:
- The brand is famous but doesn't own specific comparative vocabulary
- Most mentions are neutral ("Brand X exists") rather than comparative ("Brand X is better for Y because Z")
- The brand's primary presence is in Chinese-language content, while many AI models still weight English content more heavily
That last point is particularly relevant for Chinese brands expanding globally. A domestic brand with massive Xiaohongshu presence may score lower than its international recognition would suggest, because AI training data across English, Japanese, and Korean contexts hasn't fully captured its position.
What the High Scorers Did Right (And How to Replicate It)
The brands scoring 80+ share actionable patterns:
- Own one specific word or phrase — Don't try to be everything. Be the definitive answer to one specific question ("what's the best zero-sugar sparkling water in China?").
- Generate comparative content — Encourage, seed, or enable third-party comparison content where your brand wins on specific dimensions. Reddit-style Q&A platforms, product review sites, and tech blogs in your category matter enormously.
- Build English-language presence deliberately — For brands with global ambitions, English-language coverage in authority publications has outsized AI training weight. A Bloomberg Businessweek feature on 元气森林 contributes more to its global AI score than a thousand domestic Weibo posts.
- Measure it — You cannot optimize what you cannot see. Tracking GEO score over time reveals whether your content strategy is moving the needle.
The Divergence Is Accelerating
As AI assistants handle more product discovery queries, the gap between high and low GEO scorers will compound. Brands that appear in AI recommendations will get more awareness, more trials, and more third-party content written about them — which further improves their AI visibility in a self-reinforcing cycle.
Brands that don't appear will find that their traditional SEO presence increasingly fails to convert, as users never reach their search results.
The window to build AI visibility is now, while the content ecosystem is still being formed. The brands that own concepts today will be very difficult to displace in two years.