Are you a foreign brand trying to break into the Chinese market? Then you need to understand user-generated content, or UGC: the videos, photos, reviews, and posts made by real people, not by your brand and not by an agency. In China, this one category of content decides more purchases than your ad budget ever will.
Philip Chen is the CEO of GMA. He has spent close to two decades building content and influencer campaigns for foreign brands entering China, and he has watched the rules of UGC change twice: once when Xiaohongshu turned “seeding” into a science, and again in 2026, when AI-generated content started showing up inside real UGC feeds.
This article used to be about hashtags and KOL photo shoots. It still covers those. But the real story in 2026 is what happens after the seeding: whether people actually buy, whether an AI wrote half of what you’re reading on Xiaohongshu, and how much of your budget should go to a nobody with 8,000 followers instead of a celebrity.
Key Takeaways
- UGC still drives Chinese purchase decisions, but the mechanism has a name now: 种草 (zhongcao, “planting grass,” meaning content that plants the idea to buy) followed by 拔草 (bacao, “pulling the grass,” meaning the actual purchase). A campaign that only plants and never pulls is wasted money.
- Reverse content, called 拔草 or 避雷 (“mine avoidance”), is now rewarded by the platforms. Honest, mixed, even slightly negative posts get more reach than they did three years ago, because Xiaohongshu’s algorithm and its users both got tired of fake perfection.
- KOC economics beat KOL economics for most foreign brands entering China. A wall of small, credible voices usually out-converts one expensive face.
- AI-generated content must now be labeled by law in China, and platforms are actively punishing accounts that skip the label. If your UGC strategy touches AI tools at any point, you need a policy for this, not a guess.
Why UGC Still Decides Who Chinese Consumers Trust
China has over a billion internet users, and most of them do not trust ads. They trust other people. That has not changed since this article was first written. What has changed is that we now have real numbers on why.
A 2026 iiMedia (艾媒咨询) consumer survey on China’s “seeding economy” found that 49.10% of consumers say they act on UGC because they already had a real need for the product, and 42.96% say they act because the content itself, photo or video, was well made and convincing. In the same survey, KOL and blogger review videos accounted for 61.49% of the content that actually triggered a purchase, ahead of ordinary-person posts in content communities at 48.62%.
Here’s the part most agencies won’t tell you: the same survey found that 54.39% of Chinese consumers who bought something after seeing it “planted” said the product fell short of what the content promised. Only 32.37% said it matched their expectations. That gap is why 拔草 content, real people posting “I regret this” or “it’s fine but not worth the price,” is not a threat to your brand. It’s proof the platform still has trust left to protect, and if you play it straight, that trust works in your favor.
The Xiaohongshu 种草 to 拔草 Cycle, Explained
If you only remember two Chinese words from this article, make them these. 种草 (zhongcao) means “planting grass”: a piece of content puts an idea into someone’s head. 拔草 (bacao) means “pulling grass”: the moment that idea turns into a purchase, or gets pulled out and discarded. Your whole UGC strategy on Xiaohongshu is really just a plan for how you move people from the first word to the second.
Xiaohongshu is not a small platform anymore. It now counts around 360 million monthly active users and close to 300 million searches a day, and in 2026 the platform moved to what it calls its CES 2.0 recommendation system. In plain terms: it now weighs deep engagement (saves, comments, follows, not just likes), how well a post matches what someone actually searched for, and how consistent a creator’s account is in its niche. Generic, one-off “look at this product” posts get buried. Consistent, specific accounts get pushed.
That matters for foreign brands because the platform got crowded fast. The number of merchants on Xiaohongshu grew roughly eightfold in two years, from around 750,000 to about 6 million, and the density of sponsored posts under a single keyword has tripled or worse. Standing out with one big KOL post is harder than it was. Standing out with a steady stream of smaller, honest, well-targeted posts is not.
One of my consultants put it this way in a client review last quarter: the algorithm now rewards the post that actually helps someone decide, not the post that just looks nice.
Foreign brands new to the platform should read our full guide to Xiaohongshu marketing in 2026 before setting a budget. The short version: plan for both halves of the cycle, the planting and the pulling, or you’re just paying for attention that goes nowhere.
Douyin: Cheap Reach Still Exists, If You Build the Right Challenge
Douyin’s version of UGC still runs on hashtag challenges, and in 2026 that format is holding up better than most marketers expected. A challenge that gets official platform backing (front page placement, trending page slots, creator matchmaking) still produces what a paid ad campaign cannot: people making content for you, for free, because it’s fun to join in.
This is exactly how Coach’s earlier #CoachFromAbove challenge worked on WeChat, and the format has only gotten more built out on Douyin since. Users post their own take, the best ones get featured on the brand account, and the brand gets both reach and a library of real content it did not have to shoot itself.
Douyin has also pushed into new UGC formats worth watching. A 2026 white paper from Aurora Mobile (极光大数据) on Douyin’s “world-building” feature counted more than two dozen user-made interactive “worlds” that had each passed a million visits by February 2026, farm-style, sightseeing, and two-player worlds being the most popular formats. It’s not a fit for every brand, but it shows how far Douyin users will go to build inside a platform when the tools are simple enough.
For brands that also touch AI tools in their Douyin content, from AI hosts to auto-generated product clips, it’s worth reading Douyin’s rules on AI digital humans and live streaming before you launch. The line between “made by a user” and “made by a bot for a user” is getting watched closely, and we’ll get to why in a moment.
KOC vs KOL: Where Your Budget Should Actually Go
KOL (Key Opinion Leader) means a real influencer with a large following and a media-style presence. KOC (Key Opinion Consumer) means an ordinary person, often with under 50,000 followers, whose posts read like a friend’s honest opinion rather than a campaign. In 2023 this article treated KOLs as the main lever. In 2026, for most foreign brands, that’s the wrong order.
Current Xiaohongshu pricing gives you a feel for the trade-off. A “waist-tier” creator with 5,000 to 10,000 followers typically charges 300 to 800 RMB per post. A niche KOC with 10,000 to 50,000 followers runs 800 to 2,000 RMB. Once you cross into KOL territory, mid-tier creators (50,000 to 100,000 followers) charge 3,000 to 8,000 RMB, and top-tier names (500,000 to 1,000,000 followers) run 8,000 to 50,000 RMB or more, with beauty and mother-and-baby niches often pricing 50 to 100% above that.
Run the math with cost-per-follower (price divided by follower count) and a healthy range sits between 0.03 and 0.2 RMB. Above 0.5, you’re usually paying for a name, not for conversion. For a new brand with a limited budget, thirty to fifty KOC posts across a niche audience will almost always out-convert three or four expensive KOL posts, because Xiaohongshu users have learned to spot a paid celebrity post from a mile away, and they trust the KOC post more precisely because it looks less produced.
That does not mean KOLs are dead. A well-chosen KOL still opens doors a hundred KOCs cannot, especially for brand awareness at launch. It means the order of operations flipped: use a small number of KOLs to open the door, then let a wide base of KOCs do the actual convincing.
When the “User” Generating the Content Is an AI
This is the part that did not exist when this article was first written. Since September 1, 2025, China has enforced the 人工智能生成合成内容标识办法, a rule co-issued by the Cyberspace Administration, the Ministry of Industry and Information Technology, the Ministry of Public Security, and the national broadcast regulator. It requires visible and hidden labels on AI-generated text, images, audio, video, and virtual scenes.
Xiaohongshu now asks creators to actively tag AI-assisted posts through a dedicated setting, and posts that skip the tag when they should have used it get quietly pushed down in search and recommendation. Douyin built a similar labeling system and, on top of it, added automatic detection that stamps a note on content the creator didn’t tag themselves. This is not a light-touch rule either: in a single enforcement sweep in February 2026, platforms suspended over 13,000 accounts and removed more than 543,000 pieces of unlabeled or misleading AI content.
For a foreign brand, here’s what that means in practice. If a creator you’re paying uses an AI tool to touch up a photo, write a script, or generate a voiceover, that post likely needs the label, and hiding it is a real risk to the account, not just a technicality. It also means “UGC” as a category is getting blurrier: a real person filming a real unboxing is still gold-standard trust, but a real person posting an AI-polished version of the same thing is a different, and now legally distinct, category. Brands that get ahead of this by asking creators upfront about their AI use, and by keeping some content clearly, verifiably human, will keep more trust than brands that get caught mixing the two without saying so. If your content strategy touches China’s AI tools anywhere in the chain, from DeepSeek integrations inside WeChat to AI hosts on Douyin, build the labeling question into your creator brief from day one.
The Old Playbook Still Has Its Place
None of this makes the classic UGC plays useless. Michael Kors’ “The Walk,” which asked customers to post photos wearing the brand, and Coach’s collaboration with actress Yang Mi, where fans submitted their own designs and the winning one got produced and sold, still work as a model: give people an easy, specific action, and reward the best ones publicly.
L’Oreal’s partnership with Zhang Ziyi and Puma’s collaborations with Liu Wen and Jackson Yee worked for the same reason: the face they chose actually meant something to the audience they wanted. That part of the formula has not changed, and it never will. What’s changed is everything downstream of the post: whether the algorithm shows it to the right person, whether it’s labeled correctly, and whether a KOC’s honest follow-up post confirms or kills what the campaign promised.
A Case in Point
Take Julien, who runs a small European home-fragrance brand that entered China through cross-border e-commerce in 2025. His first move was standard: he booked three mid-tier KOLs on Xiaohongshu for launch week. The posts looked good. Views were solid. Sales for the month barely moved.
The problem, once we dug into the account data, was simple: three polished posts from paid faces planted an idea, but nothing “pulled” it through. No one was searching the product name yet, so the posts weren’t being surfaced days later when a buyer was actually ready. We restructured the budget: the three KOL posts stayed, but we added forty KOC posts over six weeks from real candle and home-scent users, each one written in their own voice, some of them mildly critical about the price or the size of the bottle.
That mix is what made the difference. The honest, mixed reviews built the kind of trust that a run of identical five-star posts cannot, and Xiaohongshu’s search algorithm kept surfacing them for weeks because they matched real search terms people were typing. Search volume for the brand name grew close to six times over the following two months, and the store’s conversion rate on Xiaohongshu-driven traffic roughly doubled compared to the launch week. The KOLs opened the door. The KOCs, and their honesty, is what actually pulled people through it.
How to Run a UGC Campaign That Actually Works in China
- Budget for both halves of the cycle. Set aside spend for the seeding (KOL or KOC posts) and separately for what happens after, restocking, customer service, and review management, so the “pull” doesn’t stall.
- Weight your creator mix toward KOC. A rough starting split for a new brand: 70 to 80% of posts from KOC-tier creators, the rest from one or two KOLs to open visibility.
- Let a few honest, even lukewarm, reviews through. A feed that’s all five stars reads as fake to Chinese consumers now, and the algorithm has started favoring accounts with a mix.
- Verify who you’re paying. Ask for ID, location, and a look at recent post history before signing a creator, especially on smaller KOC deals where fraud is more common.
- Ask every creator about AI use, upfront. Put it in the brief and the contract. If a post needs the AI label, make sure it gets one.
- Track search volume for your brand name, not just views. On Xiaohongshu, that number tells you whether zhongcao is actually turning into bacao.
FAQ
How much does a real UGC campaign cost in China?
It depends entirely on your creator mix. A KOC-heavy campaign of thirty to fifty small creators can run from around 15,000 to 60,000 RMB total, while adding even one mid-tier KOL can add 3,000 to 8,000 RMB per post on its own. Most foreign brands entering China see the best return by starting KOC-heavy and adding KOL posts only once they know what content actually converts.
Do I need to label AI-generated content on Xiaohongshu or Douyin?
Yes. Since September 2025, China requires visible and hidden labels on AI-generated text, images, audio, and video, and both platforms enforce it. If a creator you’re working with edits a post with AI tools, generates a voiceover, or uses an AI avatar, the content likely needs a label. Skipping it can get the post, or the whole account, suppressed or removed.
Can I run a UGC campaign in China without a local entity?
Yes, for most product categories you can run creator campaigns and cross-border e-commerce sales without a Chinese entity, working through a platform-registered agency or a cross-border store. Where you’ll usually need local registration is if you want a fully domestic Tmall or JD flagship store, or a Chinese ICP-licensed website. An agency familiar with your category can tell you which path applies before you commit budget.
Building Trust Through UGC in China, With Us
We’ve run UGC and creator campaigns in China for over twenty years, long enough to have watched the KOC wave replace the KOL-only playbook, and the AI labeling rules land on top of both. We know which creators are worth a contract and which are inflating their numbers.
We build the whole cycle for you: KOC and KOL sourcing, content review, AI-labeling compliance, and the follow-up tracking that tells you whether the seeding actually turned into sales. If you want a UGC strategy that survives contact with the real Xiaohongshu algorithm, get in touch with our team and we’ll walk you through what a realistic budget looks like for your category.
Sources
- 艾媒咨询 (iiMedia Research), 2026年中国种草经济行业发展状况与消费行为调查报告 (2026 China Seeding Economy Industry and Consumer Behavior Survey), iimedia.cn
- 新华网 (Xinhuanet), 你刷的短视频，标注”AI生成”了吗？ (Is the short video you’re watching labeled “AI-generated”?), on the 人工智能生成合成内容标识办法, news.cn
