Summary of Key Points
Since August 10th, DouBao, the AI assistant under ByteDance, has started charging a comprehensive fee of 12% on hotel bookings that are directed through it to TikTok (11.4% for software service fees + 0.6% for payment processing fees). This marks the first time that AI conversation traffic has been priced separately. The fee is lower than that of traditional online travel agencies (OTAs) such as Ctrip (15%-22%) but higher than the natural traffic on TikTok itself, indicating that the commercialization of AI traffic has entered a critical phase. Behind this move is ByteDance's desire to make DouBao a core business on par with TikTok, while also alleviating the pressure of computing costs. However, there are challenges to consider, such as merchants' concerns about costs, users' habits of comparing prices, and the need to ensure the fairness of recommendations. Additionally, this initiative provides a reference for pricing AI traffic in the industry.
I. DouBao's Commission: Is 12% High or Low?
The commission charged by DouBao on hotel bookings is 12%, which consists of 11.4% for software service fees (covering the platform's technology and traffic costs) and 0.6% for payment processing fees. In comparison:
- Cheaper than Traditional OTAs: Platforms like Ctrip charge between 15%-22%, so DouBao’s 12% is significantly lower, making it more cost-effective for merchants.
- More Expensive than TikTok's Natural Traffic: Previously, DouBao’s orders were combined with TikTok’s natural traffic, resulting in a lower fee. Now, since AI can more accurately match user needs (for example, if you ask for a hotel suitable for couples near Wangfujing in Beijing, DouBao will directly recommend the right one), a “efficiency premium” is charged.
- Different Fees for Other Categories: Fees vary for different categories; for instance, pregnancy and maternity care services are 17.4%, and fitness services are 8.4%. The difficulty of matching costs and needs varies across industries, leading to differing fees.
II. Why Is ByteDance Hurrying to Price DouBao’s Traffic?
DouBao is currently a major player in the AI application space, with 382 million daily active users and handling 180 trillion conversations per day (a huge amount of chat requests). However, its daily revenue is less than one million, mainly coming from e-commerce commissions. The cost of computing resources alone amounts to tens of millions per day, creating a significant financial burden for the company.
ByteDance CEO Liang Rubo has made it clear that he wants DouBao to become a core business on par with TikTok, with potential branches in e-commerce and lifestyle services. Therefore, this pricing move is not aimed at immediate profit but rather at establishing the value of AI-generated traffic to both the market and merchants.
III. This Is a “Pricing Benchmark” for the AI Industry
DouBao’s 12% fee sets a new standard for the entire AI industry, indicating that the commercial value of AI conversation traffic is higher than that of ordinary information flow recommendations (such as ads on TikTok). Other platforms have quickly followed suit:
- Alibaba Qianwen: Launched a paid professional user plan on August 10th (basic functions are free, with AI video generation charged per use), and the platform has been opened to third parties for enterprise-level intelligent services.
- Industry Impact: In the future, when other AI assistants like Baidu WenXinYiYan and Tencent Hunyuan enter the local lifestyle and e-commerce markets, DouBao’s 12% fee will serve as an important reference, pushing the industry from conceptual hype to actual profit generation.
IV. Will Merchants and Users Accept This?
There are three main challenges:
1. Merchants’ Cost Concerns: Although the fee is lower than that of OTAs, merchants may pass on the extra cost to consumers. For example, the price of hotels recommended by DouBao through Wangfujing Peninsula Hotel was found to be higher than both Ctrip’s and the hotel’s own website. Merchants are also concerned that if AI doesn’t bring new orders but merely draws customers from Ctrip, the 12% commission could increase their costs.
2. User Trust: 66.2% of users verify recommendations with other platforms after receiving them through DouBao; only 15.2% make a purchase directly. Since users are accustomed to comparing prices, if DouBao’s prices are higher, they may opt out.
3. Fairness of Recommendations: DouBao claims that merchants cannot pay to influence recommendation rankings, but how can this be guaranteed? For instance, could some merchants use fake reviews to manipulate the system? ByteDance is already addressing issues like “GEO poisoning” (false recommendations), but as more categories (such as flights and taxis) are added, the balance of interests will become more complex.
V. Could This Model Be Applied to Other Industries?
Industry experts, such as Bao Zelin from Hangzhou Songyun Digital Technology, believe that after the hotel industry succeeds, this model could be extended to sectors like catering and culture and tourism. AI can streamline the decision-making process for users (for example, asking for delicious Hangzhou-style dishes near West Lake in Beijing and getting direct recommendations and bookings). However, this would require balancing merchants’ ability to bear the costs while maintaining user trust in AI recommendations—preventing AI from becoming a “marketing machine.”
In summary, DouBao’s commission charge is an important step towards the commercialization of AI. Whether this approach will be successful depends on whether merchants can actually generate new orders, whether users are willing to pay for the services, and whether ByteDance can ensure the fairness of recommendations. This is not only a test for ByteDance but also a crucial trial for the entire AI industry.