Most foreign brands run Xiaohongshu advertising the way they run Instagram advertising, and then wonder why the money disappears. RED is not a display network with a Chinese interface. It is a search and recommendation engine wrapped around user-generated reviews, and its ad system, Juguang (聚光), rewards a completely different kind of input.
I have watched a lot of budgets burn on this platform. Below are ten things I would tell any brand before it spends its first RMB, written for people who already run paid media and want the parts that are specific to RED.
1. Understand what you are actually buying
Juguang is Xiaohongshu’s self-serve advertising platform. It lets you promote notes, target by interest and keyword, and bid for placement in search results and the discovery feed.
The critical difference: on RED, an ad is a note. You are not inserting a banner into someone’s feed, you are boosting a piece of content that has to survive next to organic posts written by real users. If it reads like advertising, it dies, and you pay for the impressions anyway.
2. Fix your organic base before you spend
Advertising on RED amplifies what is already there. If your brand has thirty notes and no reviews, paid traffic arrives at an empty profile and leaves.
Before switching on Juguang, get a base of authentic content in place: brand notes, and more importantly notes from real users and small creators. Buyers who see your ad will search your brand name, and what they find at that moment decides the outcome. This is the single most common sequencing mistake I see.
3. Treat search as the main event
RED functions as a product search engine for a large share of its users. People type a problem, “sensitive skin serum”, “quiet luxury bag”, “office lunch ideas”, and read notes before buying.
Search ads on RED therefore capture intent that already exists, which usually makes them the more efficient half of a campaign. Build a keyword structure the way you would for a search engine: high-intent terms separately from broad category terms, with separate budgets and separate creative.
4. Separate seeding from conversion
Do not ask one campaign to do two jobs. Content seeding, getting your product discussed credibly, and conversion, driving traffic to a store or a lead form, need different notes, different keywords and different measurement.
Seeding is judged on saves, comments and search-volume lift for your brand name. Conversion is judged on cost per acquisition. Mixing them produces a campaign that looks mediocre on both metrics and tells you nothing.
5. Write for the platform, not for your brand book
RED notes have a recognisable shape: a first-person title with a hook, an image or short video that shows the product in real use, a body that reads like advice, and hashtags. Polished campaign imagery underperforms consistently.
The most useful test I know: would a real user post this without being paid? If the answer is no, the algorithm will reach the same conclusion.
6. Let the creators keep their voice
Creator collaboration on RED is regulated and must be declared through the platform’s official system. Work within it, because undeclared paid content is penalised.
Within those rules, the mistake is over-scripting. A brief that dictates every sentence produces a note that sounds like an ad, and RED audiences detect that instantly. Give creators the product, the key facts and the boundaries, then let them write. Volume of credible mid-size and small creators beats one expensive name in almost every test I have seen.
7. Sort the account setup before the campaign
This is the boring tip that saves the most money. A foreign brand needs the right account type, verified business documentation and a working payment route into Juguang. Trademark documentation matters if you want a verified brand account and the protections that come with it.
Brands routinely plan a launch, build the creative, then discover the account cannot be funded or verified in time. Start this two months before you intend to spend.
8. Measure the assisted effect, not just the last click
RED rarely gets credit in a last-click model, because the journey often ends on Tmall, Douyin or a WeChat mini-program days later. Judged on last click alone, RED will look like a poor performer and you will cut it, which is exactly the wrong move.
Track brand search volume, note saves, and store traffic in the days following campaign bursts. If your brand’s search volume on RED and on Baidu rises while a seeding campaign runs, the campaign is working even if the platform’s own conversion column looks thin.
9. Plan around the commercial calendar
618, Double 11 and the Chinese New Year period concentrate attention and raise auction prices. Two consequences: bid costs rise sharply during peaks, and the weeks before them are when seeding needs to be already in place.
The brands that do well are the ones whose reviews and notes exist before the shopping window opens. Starting to seed in the week of the festival means paying peak prices for content that has not had time to accumulate credibility.
10. Give it a proper test period
A two-week trial on RED tells you almost nothing. The platform’s value compounds: notes keep surfacing in search months after publication, and review volume builds trust slowly.
Budget for a minimum of three months, with a clear structure, and be prepared to iterate creative weekly. Brands that treat RED as a quarterly experiment usually stop just as the content library starts working for them.
How the budget usually breaks down
Brands often ask for a split before they ask for a strategy, so here is the shape I recommend for a first serious quarter on RED, with the caveat that category and ambition move it considerably.
Roughly half to creator content and seeding. This is the part that produces the asset base: notes, reviews and the search results people find when they check you out. It is the least glamorous line and the one that determines whether everything else works.
Around a third to Juguang, weighted toward search rather than feed at the start. Search captures existing intent and gives you cleaner signal about which keywords and which creative deserve more.
The remainder to production and iteration. Content on RED wears out faster than most teams expect, and you will want to refresh creative weekly rather than run one set for a quarter.
What I would avoid is the reverse profile: most of the budget in paid, a token amount in content. That produces high impression counts, thin engagement, and a brand page that looks abandoned to anyone who clicks through.
A note on what RED is not good for
Being useful means saying where it does not fit. If you sell a low-consideration commodity product with no visual or lifestyle dimension, or you have no ability to produce Chinese-language content continuously, RED will be expensive relative to the alternatives. Douyin or marketplace advertising will serve you better. RED rewards categories where people research and share: beauty, fashion, food, travel, home, wellness, parenting.
Written by Philip Chen, who leads GMA. His perspective comes from launches he has run personally, including several that went badly before they went well.
If you want a specialist read on your RED setup, our sister agency runs Juguang campaigns and creator programmes full time: RedNote ads management, a service page covering account verification, campaign structure and optimisation for foreign brands.
Also useful on this site: how an e-commerce-first launch works in Chinese beauty.
Sources
- Xiaohongshu on Wikipedia, background on the platform’s history, ownership and scale.
- Statista’s Xiaohongshu topic page, user numbers and demographic breakdowns updated through the year.
- Jing Daily, ongoing coverage of Chinese consumer and platform behaviour, useful for tracking how RED commerce is evolving.