Add preferred source
Ahead of Double 11, ByteDance’s AI app Doubao has embedded Douyin livestream rooms into its product recommendations, letting users watch livestreams, claim coupons, place orders and pay inside the app without jumping to Douyin. The platform is also proactively pushing Douyin Mall spend-and-save coupons and e-commerce feature cards, breaking with the convention that conversational products do not initiate conversations. Doubao is also using platform subsidy coupons to lower prices on some goods. The move reflects commercial pressure: as of the first half of 2026, Doubao’s daily revenue was under 1 million yuan, with daily e-commerce transaction value of about 10 million yuan and annualized GMV of roughly 3.65 billion yuan (approximately $546.7 million). With more than 400 million monthly active users, Doubao also faces high computing costs and quarter-on-quarter declines in user time spent and usage frequency, making e-commerce revenue from merchants a more viable monetization path.
Key Elements
Ahead of this year’s Double 11 shopping festival, Doubao, the AI app owned by ByteDance, has been steadily adding e-commerce features. Doubao’s app has added Douyin livestream room entry points to the product recommendations it surfaces in chats, so users can watch livestreams, claim coupons and complete orders and payments inside Doubao without switching to Douyin. At the same time, Doubao has begun proactively pushing Douyin Mall spend-and-save coupons to users; some received large coupons such as “900 yuan off 7,000 yuan,” and the platform also pops up small cards introducing scenario-based e-commerce capabilities such as “Doubao places orders for me,” “check logistics” and “return pickup.”
As a conversational product, Doubao has consistently followed a passive interaction logic of “user asks, AI answers,” and initiating conversation is uncommon. Breaking that convention to drive sales is seen by the market as a signal that Doubao and the Douyin E-Commerce pipeline are further connected. Previously, Doubao had already embedded Douyin Mall product recommendations in its answers, covering some Douyin local services merchants, and launched a “Ask Doubao for Double 11” entry point with a large number of preset shopping dialogues to help users screen products.
In terms of timing, this year’s Douyin Double 11 campaign began on Oct. 15, and Doubao moved its “Ask Doubao before buying” entry point forward that same day, upgrading it three days later to a red-font “Ask Doubao for Double 11,” a marked increase in prominence. By contrast, during this year’s 618 shopping festival, the entry point was not changed from “Ask Doubao before buying” to “Ask Doubao for 6.18” until June 18 itself. With the earlier entry point, platform subsidies and embedded livestream rooms combined, Doubao’s investment in Double 11 is more aggressive than in previous shopping festivals.
On price, shopping inside Doubao is often cheaper than ordering directly on Douyin. In addition to the Double 11 spend-and-save coupons consistent with Douyin Mall, Doubao also issues extra surprise subsidy coupons from the Doubao platform. For example, a carton of East Buy pure milk priced at 113 yuan sells for 102 yuan in Doubao after a 10% platform subsidy coupon, while entering through the Douyin platform still shows the original price; a Snow Flying down jacket priced at 999 yuan costs 894 yuan in Doubao after coupons, versus 919 yuan on Douyin after spend-and-save coupons, and some users can also get a 30%-off Doubao subsidy coupon. That means Doubao is putting up real money of its own for subsidies.
The livestream feature currently centers mainly on transactions. In product recommendations, users can see livestream rooms that match the product and are currently live; swiping left browses more and swiping up and down switches between them, but the number is limited, with a single answer usually offering only five livestream rooms. After entering a livestream room, users can view the product showcase but cannot browse the account homepage or other short videos as they can on Douyin. The top sales leaderboard is consistent with Douyin’s, with Hui Tong Xing ranking first, and tapping it jumps to the corresponding page.
Behind Doubao’s e-commerce push is clear commercial pressure. According to LatePost, as of the first half of 2026, Doubao’s daily revenue was under 1 million yuan, mainly from e-commerce commissions, and it generated about 10 million yuan in e-commerce transaction value per day. On an annualized basis, that is roughly 3.65 billion yuan (approximately $546.7 million) in annualized GMV and about 365 million yuan (approximately $54.7 million) in annualized revenue; the figures are estimates, not official data. For a super app with more than 400 million monthly active users, that revenue scale is not high.
Cost pressure is also significant. Volcano Engine disclosed that as of June 2026, daily token calls on the Doubao large model exceeded 180 trillion, more than 10 times growth over the past year. Expanding user scale means computing consumption keeps climbing, while commercialization revenue cannot yet cover the related costs.
Organizationally, in September 2025, ByteDance commercialization veteran Zhao Qi took over Doubao, and a month after his arrival Doubao began preliminary integration with Douyin E-Commerce, after which commercialization moves clearly accelerated. This year, Doubao has successively pushed forward integration with Douyin E-Commerce, launched a subscription paid service and added travel services, among other monetization methods. The subscription service launched in May this year and has not disclosed operating data. Morgan Stanley estimates, using U.S. AI chat apps as a reference and factoring in Chinese consumers’ willingness to pay, that Doubao’s annualized subscription revenue ranges from $101 million to $1.5 billion. However, three years after adopting subscriptions, ChatGPT’s paid conversion rate is only 6.1%, and with no obvious technical barriers and peers generally offering free services, the conversion outlook for subscriptions is uncertain.
More noteworthy is the change in user stickiness. QuestMobile data show that as of June 2026, Doubao’s average monthly time spent per user was 144 minutes, falling to 133 minutes in September, down 7.8% quarter on quarter; average monthly usage frequency per user fell 6.1% quarter on quarter. With both time spent and frequency weakening, continuing to aggressively push paid subscriptions may not be the optimal choice. By contrast, generating revenue from merchants through e-commerce does not require consumers to pay directly for AI, and is closer to ByteDance’s long-accumulated commercialization experience.
The two products’ user bases overlap heavily. Based on QuestMobile data, 90% of Doubao users in May this year also used Douyin. Douyin has 1 billion monthly active users, and according to a securities industry expert cited by Huxiu, its GMV in the first half of 2026 was about 2.4 trillion yuan. Doubao handles the entry point and Douyin handles the commercial ecosystem, with the underlying logic still the traffic monetization ByteDance knows well.
However, the two traffic logics differ. Douyin relies on algorithmic proactive recommendations, stimulating user interest and purchases while they scroll short videos; Doubao is in a passive mode, requiring users to ask a question before product recommendations appear, and is not a platform that can continuously push products. Proactively pushing coupons and shopping feature cards may be exactly a test of user reaction.
This path also has boundary issues. Mini programs are likewise passive and require users to search before they appear, but merchants have their own brand entry points and a relatively complete commercial ecosystem, whereas merchants lack a “place of business” in Doubao’s recommendations, and Doubao will find it hard in the short term to build a mini-program ecosystem like WeChat’s. In addition, once proactive platform recommendations are tied to commercial interests, they may touch on the controversy that “AI should not have advertising.” OpenAI launched an advertising business early this year, with annualized revenue topping $1 billion in 200 days, but it has also been attacked by peers positioning themselves as “ad-free.”
Notably, large models moving into e-commerce has become an industry-wide move. Ahead of Double 11, Qwen Office launched an e-commerce Agent, offering cross-platform business dashboards, trend insights, hit-material generation and cross-platform ROI optimization to all e-commerce platforms and merchants; the “Qwen AI shopping assistant” inside the Taobao app has also been fully integrated with Taobao. E-commerce observers believe large models are stringing search, recommendation, price comparison, decision-making, ordering and even after-sales into a closed loop with conversation as the entry point, making “complete a transaction in one sentence” possible for the first time.
For the market, Doubao bringing Douyin’s core livestream selling scenario into a conversational product means AI applications are evolving from tools into transaction entry points. If the model works, it will provide a reusable monetization path for China’s conversational AI; if results fall short of expectations, it will also draw a commercialization boundary for the industry. The contest between entry points and shelves has only just begun, and Double 11 is merely the first observation window for this experiment.
