During the National Day holiday, "the first batch of people burned by AI travel guides" trended across Chinese social media: AI-generated itineraries ignored real-time factors such as crowd levels, weather and business hours, and tourists who followed the recommendations arrived to empty venues. Around the same time, 36Kr reported on October 6 that leading AI assistants have entered an "order-grabbing" phase — Doubao has started charging merchants channel fees on hotel orders completed through AI recommendations, and launched a CNY 68/month pro subscription.
Taken together, the two events point to one trend: AI search is moving from "giving answers" to "generating transactions." For brands, this means an AI answer is no longer just an impression — it can become an actual order-distribution gateway. Drawing on the latest data from CNNIC, Xsignal and others, this article breaks down the signals behind AI search commercialization and the strategy brands should adopt.
Three signals: AI search is becoming a transaction gateway
Signal one: AI assistants are competing for orders. According to 36Kr's October 6 report, Doubao now collects channel fees on hotel orders and offers a CNY 68/month pro plan for consumers. From free Q&A to commission-sharing and subscriptions, AI assistants have, for the first time, tied their business model directly to deal completion.
Signal two: Office agents are exploding, and a B2B gateway has taken shape. The AI app monthly report by Xsignal (republished by Sina Finance on September 30) shows Doubao's MAU reached 553 million in August (up 4.6% month over month). The office-agent segment surged across the board: Tencent WorkBuddy surpassed 10 million MAU (+72.9%), Baidu Dazi reached 7.879 million (+203.7%), and Qwen Office hit 1.217 million (+298.2%). Enterprise customers are getting used to "letting AI pick suppliers," making AI Q&A a new B2B customer-acquisition channel.
Signal three: AI answers now directly shape real consumer decisions. The National Day AI-travel backlash (reported by Sohu and PEdaily on October 3) proves that users really do book hotels and choose restaurants based on AI recommendations. When AI information is inaccurate, both brands and users lose. Whoever can continuously supply accurate, up-to-date information to AI engines gains a decisive edge in the "Q&A-to-purchase" chain.
The data backdrop: a 700-million-user AI decision arena
CNNIC's Report on the Development of Generative AI Applications (2026), released September 29, shows that by the first half of 2026, China's generative AI users exceeded 700 million, with adoption above 50%; 76% of users ask AI to answer questions. Intelligent computing power reached 2185 EFLOPS, up 177% year over year, and all six of the most-cited mainstream LLMs were built by Chinese teams.
AI Q&A is now one of the main highways for brands to reach customers — and commercialization has given that highway, for the first time, a "deal" attribute. For SMBs, missing AI search may mean missing both exposure and orders at once.
Traditional search versus AI search: the commercial logic
| Dimension | Traditional search (Baidu/Google) | AI search (Doubao/Yuanbao/Qwen) |
|---|---|---|
| Traffic gateway | Search box + keyword ad slots | Conversational Q&A + agent recommendations |
| Monetization | Bidding rankings, CPC ad fees | Transaction channel fees, subscriptions, referral commissions |
| Content requirements | Keyword density + backlink weight | Source credibility + citable paragraphs + real-time accuracy |
| Brand actions | Bid on keywords, buy ads, chase rankings | Build sources, optimize AI answers, feed transactional data |
| Performance metrics | Click-through rate, conversion rate | AI mention rate, recommendation rate, Q&A-driven deals |
Four steps to win the Q&A-to-purchase loop
Step one: structure transactional information so AI can cite it directly. Decision drivers — price ranges, opening hours, booking methods, stock status — should live on your website in FAQs, comparison tables and structured data. The root cause of the National Day AI-guide failures was missing real-time information; brands that fill that gap are far more likely to be cited as trustworthy sources.
Step two: cover the three question types that precede a purchase. Organize content around comparison queries ("how to choose between A and B"), FAQ queries ("how much / how long / how to contact"), and case queries ("who used it and what happened"). Turn every question on the customer's decision path into an extractable answer paragraph.
Step three: stay on the white-hat source-building path. On September 30, Shandong Province released the Application Specification for Brand Generative Engine Optimization (T/SDCBD 0007-2026), effective October 1; Google also updated its search quality guidelines on October 2, requiring human review of AI-generated content. GEO oversight is tightening, and black-hat tactics such as bulk article planting and fabricated sources face increasing cleanup risk.
Step four: build your baseline before you spend. Use free monitoring tools such as Lens GEO to track your mention rate and recommendation changes across Doubao, Yuanbao and Qwen. Establish your own baseline first, then invest systematically in content optimization — and treat any vendor-claimed conversion figures with caution until verified.
FAQ
Q1: Doubao charges channel fees on hotel orders. Do brands outside the hotel business need to care? Yes. Channel fees show AI assistants are evolving toward transaction commissions, and hotels are just the first stop; similar models will soon reach local services, e-commerce and B2B. Treat AI answers as a commercial channel now, not just an exposure slot.
Q2: Is traditional SEO still relevant after AI search commercialization? Absolutely. SEO builds the source foundation and GEO opens the AI gateway; the two are complementary. Site structure and content quality decide whether AI engines cite you, and citation decides whether users ever see you.
Q3: How can a brand tell whether AI engines are recommending it? Regularly query Doubao, Yuanbao and Qwen with "brand keyword + business keyword" prompts, and log whether AI mentions your brand, which source it cites and which competitor it recommends. A weekly or monthly mention-rate report gives you a comparable baseline.
Q4: Does a corporate website need a redesign for AI search? Not necessarily a redesign, but the content needs rework: add FAQs, comparison tables and structured data so every page carries answer-ready paragraphs, and fill in transactional details such as pricing, opening hours and booking methods.
Q5: Is it too late to start GEO in 2026? It is not too late, but competition is accelerating. With 700 million generative AI users and commission models taking hold, 2026 marks the first year of AI search commercialization. The earlier you build your source ecosystem and monitoring baseline, the stronger your first-mover position.
Final thoughts
AI search commercialization is one of the most certain traffic shifts of the second half of 2026. For brands, the core move is simple: turn your website and content into a source that AI engines trust, cite and can transact on. Shanghai Zheming Information Technology Co., Ltd. continuously tracks engine dynamics across Doubao, Yuanbao, Qwen and DeepSeek, offering end-to-end GEO services from source building and content optimization to performance monitoring. Quotations are subject to actual pricing. Further reading: 2026 AI Search Trends: From Conversational Tool to Decision Gateway, Doubao Search Service Opens Its API: GEO for the AI Search Gateway, Local Business GEO: How Regional Brands Win AI Search Exposure, GEO Group Standards: Seizing the Compliance Advantage in AI Search, Measuring GEO Results: Quantifying Brand Performance in AI Search.