Behind the independence of channels lies the differentiation of traffic value logic.

Recently, a new rate quietly appeared in the merchant backend of Douyin Laike.

For hotel orders transacted through the Doubao entry, the software service fee is 11.4%, and the payment processing fee is 0.6%, totaling 12%. Previously, these orders were settled at a unified rate of 8% within Douyin's natural traffic.

The 4-percentage-point increase sparked an uproar among hotel merchants. That evening, "Doubao-recommended hotels now charge fees" became a hot topic. Liu Xing, Doubao's PR head, responded on Weibo that night: The claim is inaccurate. The lifestyle services business does not offer paid promotions or charge advertising fees. Merchants cannot influence recommendations or rankings through payments; they only pay channel service fees after orders are transacted.

On its own, 12% is not high in the online travel industry. Traditional OTAs have long maintained a comprehensive rate of around 15%, which can reach 25% with premium placements and ad promotions. If Doubao currently only charges transaction service fees without an additional ad bidding system, it actually provides merchants with a transparent cost structure.

The real change is that this marks the first time in China's internet history that an AI dialogue entry has been explicitly priced as a fully independent transaction channel. It means orders generated through Doubao are no longer Subsidiary products (secondary products) of Douyin's main traffic—they now have independent settlement rules, rate tiers, and a complete commercial identity.

Many don't understand why Doubao's channel seems more expensive within the same Douyin ecosystem.

This isn't about pricing strategy tricks but rather reflects the differentiated traffic value stemming from evolving distribution formats—a universal trend as internet traffic shifts from broad attention to precise intent.

The same order, whether sourced from short video recommendations, search entries, or AI dialogue, carries different levels of user intent concentration and conversion efficiency, leading to varying commercial values.

01

Two Types of Traffic, Two Types of Business

According to Douyin Lifestyle Services' policy on specific channel service fees, Doubao is now officially classified as a "specific transaction channel" with independent rates by category.

Take the accommodation category as an example: the group-buying comprehensive rate is 8% for both Douyin's main site and Doushengsheng, while Doubao's channel rate is 12%. Merchants can still choose to operate solely on Douyin's main site and settle at the original standard. Previously, lifestyle service orders generated through Doubao were uniformly categorized under Douyin's natural traffic pool and settled at the main site's standard.

The reason for the split settlement is clear. With recent milestones like surpassing 380 million monthly active users, launching a professional subscription version, implementing transactional features, and integrating scenario capabilities, Doubao's role within ByteDance is shifting from a C-end user product driving traffic to a commercial facility with independent monetization capabilities.

The 12% vs. 8% difference between Doubao and the main site reflects more than just channel distinctions—it's not a crude rationale of "charging more for the same traffic through a different entry point."

Let's start with the difference in user intent concentration. The core traffic logic of Douyin's main site is interest-based recommendations, where users primarily consume content, and consumption needs are passively triggered. Many people are enticed by content and then place orders, representing a weak-intent, wide-funnel traffic pattern.

However, traffic in AI large model dialogues comes from users' active inquiries, where they seek solutions with clear plans, budget ranges, and preferences—a strong-intent, narrow-funnel pattern. The act of users actively asking questions already completes a round of demand screening, and merchants engage with an intention-filtered customer base. The commercial value of active demand is often higher than passive exposure.

On the other hand, if transaction links are embedded within content scenarios, users have more dispersed choices and may compare content from a dozen hotels before placing an order. This dilutes individual merchants' conversion and redemption efficiency.

In contrast, when interacting with AI, the model performs a centralized match based on all user constraints and ultimately presents only a few highly relevant options. This significantly shortens the user decision path, reduces drop-off points, lowers churn rates, and increases single-traffic conversion efficiency.

For merchants, while entering AI's recommendation pool is challenging, the probability of traffic converting into orders is higher than in general browsing scenarios—equivalent to the platform completing precise supply-demand matching in advance. Notably, Doubao adopts a post-transaction commission model, where merchants only incur fees upon actual order completion.

In simple terms, AI screens "potentially converting traffic" into "already converted traffic," and the premium can be understood as the price for risk assumption and effectiveness guarantees—mirroring the industry norm where CPS (Cost Per Sale) sharing ratios typically exceed ad placement conversion ratios in e-commerce.

It's worth noting that as an independent AI assistant entry, Doubao serves not only existing users within Douyin but also potentially covers users who previously didn't consume lifestyle services on Douyin. For example, users accustomed to using AI for travel planning and Strategy inquiry (guide searches)—whose original booking paths involved search engines and traditional OTAs—can now directly enter Douyin's transaction closed loop (closed loop) through Doubao.

For merchants, this represents incremental external traffic rather than a redistribution of existing traffic across different Douyin entries, effectively opening new customer acquisition channels. The value of incremental channels inherently exceeds the internal reallocation of inventory channels (existing channels).

02

Traffic Value Is Being Repriced

From an industry-wide perspective, this rate differential is a microcosm of traffic value differentiation.

In the past, internet traffic pricing centered on attention duration as the core metric, with insufficient differentiation between general browsing and precise demand pricing. As AI intervene (intervenes) in distribution, the intent concentration and conversion efficiency of traffic are further quantified, widening the value gap between different traffic attributes.

An industry trend is emerging: broad-attention traffic will increasingly follow a cost-effectiveness route, while the value of precise-intent traffic will continue to rise. AI simply makes this differentiation clearer and more quantifiable.

Over the past few years, as AI search gained traction, the industry once believed "large models will replace search engines." But years later, replacement hasn't occurred; instead, search's form and value have continuously evolved.

Mainstream search engines have deeply integrated AI capabilities, shifting information presentation from "ten links" to "summarized answers + cited sources." Users no longer need to manually screen information—AI completes an initial round of Summarization (summarization) and judgment (judgment).

Accompanying these form changes is a shift in commercial value. In general browsing scenarios, user attention is diluted by vast content, causing the commercial value of individual information pieces to decline. In precise-intent scenarios, however, users arrive with clear demands, expecting direct solutions, and the commercial value of each effective interaction rises.

In short, opportunities in general browsing are weakening, while the value of precise demand is increasing.

Data supports this: over the past five years, the average annual price increase for keyword ads on China's mainstream search engines exceeded 15%, with even higher CPC (Cost Per Click) growth for in-platform search on e-commerce platforms. Meanwhile, information feed ad unit prices have plateaued.

This rise and fall reflect a revaluation of traffic value: when information supply is infinitely abundant, scarcity shifts from content to users' true intent.

AI's integration is accelerating this process. Traditional search returns ten results, splitting traffic across multiple sites; AI dialogue directly provides answers and recommendations, narrowing user choices and strengthening the Head Effect (winner-takes-all effect).

For merchants involved, the gap between being selected by AI and not being selected may be even larger than the difference between the first and second pages in traditional search—explaining the rapid rise of GEO (Generative Engine Optimization) as a new sector.

This trend is a global consensus.

For example, Amazon's in-platform search ad unit prices have risen far more than display ads, precisely because traffic with clear purchase intent is increasingly scarce. Meanwhile, Google Shopping's transaction-based commission model forms a clear hierarchy with traditional search ads. OpenAI's Instant Checkout similarly opts for a transaction-sharing model, settling rates only post-transaction.

Behind these concurrent choices lies a shared industry judgment: when AI can understand users' true intent and directly complete transactions, its commercial value is no longer calculated by exposure count but repriced based on transaction contribution.

For platforms, this opens a second growth curve beyond ad revenue. Ads expand cognitive boundaries, while AI fulfills explicit demands. Within platform ecosystems, content seeding, live-stream conversion, search fulfillment, and AI matching now form a complete consumer decision journey, with each link corresponding to a distinct commercial system.

03

The Value of AI Engines Will Continue to Expand

This model shift aligns with different evolutionary trajectories within Douyin's ecosystem.

Over the past few years, Douyin's e-commerce business has centered on interest-based commerce, stimulating potential consumption through short video and live-stream content recommendations—"goods finding people" as the main theme. After proposing full-domain interest-based commerce in 2023, scenarios like search and the mall, where "people find goods," have gained increasing weight.

Last year's white paper on search operations by Massive Engine (Juliang Engine) revealed that traffic from search channels has a 90% higher conversion efficiency than information feed-based general browsing recommendations. Over the past few years, a key evolutionary trajectory of Douyin e-commerce has been shifting from passive content-triggered interest seeding to accommodating more user-initiated explicit shopping demands.

While the overall transaction share of the Shelf yard (shelf field) has risen to 40-50%, most of these transactions are still driven by short videos and live streams, with purely user-initiated search orders remaining limited in proportion.

The addition of the AI dialogue entry represents the latest step in this evolutionary path.

Returning to Doubao, its pioneering implementation of independent rates in the hotel and travel category aligns highly with sector characteristics.

Hotel bookings are a typical multi-dimensional decision-making scenario, requiring simultaneous weighing of over a dozen variables like location, price, star rating, room type, and reviews. Traditional OTAs rely on filters and sorting, forcing users to repeatedly toggle and compare. AI dialogue can receive all preference conditions at once and directly provide optimal options, dramatically boosting decision efficiency.

Deloitte's 2026 Summer Travel Survey shows that the share of users planning trips with generative AI has risen from 15% in 2025 to 25%, reaching 36% among millennials. The high-value, long-decision-chain, and multi-matching-dimension nature of this sector makes AI's efficiency gains most perceptible and translatable into quantifiable commercial value.

Industry-level changes rarely arrive abruptly. Fifteen years ago, search engine keyword bidding transformed how SMEs acquired customers; ten years ago, information feed recommendations restructured content and e-commerce distribution logic. Today, AI dialogue entries are beginning to participate in transaction flows as independent channels, redefining the value benchmark for internet traffic.

Following this logic, it's easy to imagine that the commercial value of AI engines like Doubao will continue to grow.

— END —For more, click below to follow