Buying decisions are moving into AI. Someone has a need, asks Doubao, DeepSeek or Qwen, gets two or three brand names back, and places the order. If your name is not in that answer, the awareness budget you spent upstream leaks straight away.
This is the most overlooked form of loss in search today: the brand simply disappears from the AI's answer. As a generative engine optimization (GEO) provider, New Galaxy AI (盖立克思) uses three real client engagements to show what "getting into the AI's answer" actually involves.
Case 1: Antioxidant Supplement Brand — Cross-Platform Mention Rate 3% → Top of Category
Problems and challenges
- The issue was not a lack of interest. People were convinced by the brand's content, then asked an AI, and were pointed somewhere else.
- Ingredient terms were both the entry point and the most crowded ground — however users searched, this brand did not surface.
- Questions like "is it safe" and "can I take it long term" were being answered by the AI on the user's behalf, with no input from the brand.
What we did
- Built a content matrix on sources AI is willing to trust: antioxidant mechanisms, ingredient traceability, clinical data — evidence-backed material instead of slogans.
- Restructured content around "health question → ingredient mechanism + brand answer" rather than inserting keywords.
- Distributed content deliberately, then tracked how each of the three platforms actually answered, and revised the content to match the AI's own framing.
Results
- Cross-platform brand mention rate: 3% → top of category
- Share of relevant user questions where the brand appeared in the Top 3: 60%+
Client quote
"Before we started, our brand was nearly impossible to find across the three major AI platforms — our mention rate was 3%. What hurt most was not the missing exposure: it was that a customer would be convinced by our content, then ask an AI, and the AI would name someone else. After the GEO work, we appear in the Top 3 for more than half of relevant questions, and when AI mentions us it cites our ingredients and testing data — not a throwaway 'this brand also makes it.'"
Case 2: Business Publishing Brand — AI Mention Rate 0 → 89%, 16 Recommendation Slots Won
Problems and challenges
- Twenty-plus years of accumulated content did not count for anything automatically — when users asked relevant questions, the brand was largely absent.
- The content sat in book form: AI could not parse it, retrieve it, or cite it.
- Entry points around corporate biography publishing, finance-IP publishing and business case books had been claimed first by faster-moving brands.
- Despite deeper expertise, it trailed newer brands with less content but a structure better suited to how AI retrieves information.
What we did
- Broke existing books and case studies into small units, each answering one specific question, so AI could use them directly.
- Built topic series on platforms AI cites frequently — Zhihu, Sohu, Yidian — across 30 scenarios users genuinely ask about.
- Gradually made corporate biography, finance IP and business case questions recurring fixtures in AI answers.
Results
- Mention rate across six platforms: 0 → 89%
- 1,700+ scenario-specific pieces, covering 30 question scenarios
- 16 AI recommendation slots won
Client quote
"We have been publishing for over twenty years and we are not short of corporate biographies or finance-IP case studies. But before this, users would ask a relevant question and the AI either skipped us or recommended newer names. After the GEO work, when users ask Doubao or DeepSeek about corporate biography or finance-IP publishing, the AI brings up our brand and our content. Going from almost zero to 89% across six platforms was unthinkable for us before."
Case 3: Domestic Salmon Brand — Coverage on All Six Platforms, No.1 Slot for the Core Question
Problems and challenges
- The product met the standard, but AI was repeating that it "cannot be eaten raw," and consumers were talked out of buying.
- There was too little authoritative, positive material — even when AI was willing to say something favourable, it had nothing to cite.
- Trust was being eroded bit by bit by inaccurate claims, and the longer it dragged on, the harder it was to reverse.
What we did
- Anchored the content on the national standard GB 10136 and produced a set of plain-language explainers on the raw-consumption standard, making the case for domestic salmon being safe to eat raw, thoroughly.
- Distributed this material across high-authority platforms that large models crawl frequently, so that when AI answers raw-consumption questions, it reaches for this body of standard-based explanation first.
Results
- Full coverage across six AI platforms (6/6)
- No.1 recommendation slot for the core question "Can domestic salmon be eaten raw?"
Client quote
"Now that AI mentions us, consumers ask the AI directly whether this salmon can be eaten raw, and the AI answers on its own. Our support team reports fewer questions on this, and order volume has gone up."
Three Kinds of Invisibility in AI Answers
Read together, the three cases are three distinct kinds of invisibility: one where the brand cannot be found however hard you search, one where AI cannot read the content, and one where a good product is buried under inaccurate claims. The routes differ; the goal is the same — get AI to list you accurately and credibly when it answers.
Type of invisibility | How it shows up | Root cause | Approach | Result |
Unfindable | AI names other brands for relevant questions | Entry points occupied; no source AI considers citable | Scenario-based Q&A content matrix, targeted distribution, revision against AI's actual framing | Mention rate 3% → top of category; 60%+ Top 3 share |
Unreadable | Content is substantial but AI cannot retrieve or cite it | Content exists as whole books or long pieces; wrong granularity | Break into units that each answer one question; multi-platform topic coverage | Mention rate 0 → 89%; 1,700+ pieces; 16 recommendation slots |
Overshadowed | Inaccurate claims dominate the AI's answer | No authoritative, positive material to cite | Anchor on national standards, publish explainers on high-authority sources | 6/6 platform coverage; No.1 for the core question |
About New Galaxy AI
New Galaxy AI is a full-service generative engine optimization (GEO) provider built on proprietary technology, with end-to-end capability in overseas brand marketing.
On the technology side, seven core algorithms run underneath: content structuring (GStruct), semantic enhancement (GSemantic), entity recognition (GEntity), authority assessment (GTrust), question mining (GQuery), content quality scoring (GScore) and knowledge graph construction (GGraph). These are combined with five proprietary models and Prism, a multi-platform citation monitoring engine, connecting generation, citation, monitoring and optimization into one pipeline. Compute runs on a self-built A100×8 server cluster: both training and semantic parsing happen on our own machines, and data does not leave them.
In terms of scale, New Galaxy AI has served 300+ companies across 12 industries and 28 sub-sectors, with direct access to 75,000+ publishing channels including 296 overseas news outlets. Coverage extends to nine AI platforms: Doubao, DeepSeek, Kimi, Qwen, Yuanbao, Wenxin Yiyan, plus ChatGPT, Gemini and Google AI Overview.
For overseas work, cross-lingual semantic distillation supports 30+ languages and covers ChatGPT, Gemini and Google AI Overview, helping brands localize content and extend their source footprint abroad.
Delivery follows a six-step loop: diagnosis → semantic insight → knowledge graph → content matrix → monitoring → iteration. It starts with a multi-platform visibility check-up for the brand, moves to structured content built around the questions users actually ask and full-channel distribution, and closes with Prism tracking results and attribution, so it is clear where the budget went.
On compliance, optimization stays within regulatory boundaries, and every piece of content passes both fact-check and compliance review. Data is stored encrypted on self-built servers and is visible only to the service team. Results are backed by a commitment: if the agreed brand AI mention rate is not reached within the agreed period, the service period is extended free of charge for an equal duration; if the target is still not met after the extension, the corresponding service fee is refunded in proportion to the shortfall.
On credentials, New Galaxy AI is a council member of the China Advertising Association (membership No. BM13132) and a council member of the AI Marketing Application Working Committee of the China Advertising Association of Commerce (CAAC). It has taken part in drafting multiple GEO group standards, including the T/CAACCHINA 001-2026 to 005-2026 series and T/CAPT 026-2026, Generative Engine Optimization (GEO) — Specification for Trustworthy Information Dissemination and Information Ecosystem Governance. It completed a USD ten-million-class angel round in January 2026 and holds five software copyrights.
Book Your Brand's AI Visibility Diagnostic
AI search has become a brand's new storefront. Want to know whether your name still appears in answers across the major AI platforms — and whether what AI says about you is accurate? New Galaxy AI provides an AI visibility diagnostic.
If brand visibility in AI search matters to you, get in touch with New Galaxy AI and book your diagnostic.