Two source families, one table—read carefully.

Dimension	36Kr / XinEntropy	Huxiu / JingShang	How to read
GMV growth	2025 YoY ~+72%	2024 ~400B → 2025 ~850B yuan	Fast growth, tiny base vs Douyin/Taobao
Ads / revenue	Ads ~32B, 76%	Revenue ~42B, ads 32B = 76%	Ad share aligns
High-ticket	>¥3,000 ~+47% vs Douyin	—	Decision quality, not all SKUs
On-site conversion	"Last mile" gap	Estimated 0.7%–1.2%	Seeding ≠ checkout

Douyin 2025 e-commerce GMV is publicly cited around 4.4 trillion yuan; Huxiu puts Taobao/Tmall near 10 trillion. Xiaohongshu's 850 billion (Huxiu) can grow fast and still be an order of magnitude smaller—better high-ticket conversion ≠ same SKU volume as on Douyin.

Strong seeding: decide here, pay elsewhere

Xiaohongshu's mood is "research first, buy second"—unlike Douyin's entertainment browse or Kuaishou's trust trade. The +47% line describes decision quality, not total traffic.

Huxiu's on-platform conversion estimate is colder: 0.7%–1.2%, far below traditional e-commerce. Users still "look here, buy elsewhere"—seeding on Xiaohongshu, checkout on Taobao, JD, brand mini-programs, or offline.

For brands:

• Content and ads buy search and saves; in-app GMV may not move in proportion;
• Scale often needs off-platform capture—longer ops chain;
• 76% ad share (both sources) means monetization still leans on reach and exposure, not a closed commerce loop eating the full transaction chain.

Heavy ad mix: fast growth, "traffic platform" shape

2025 ads ~32 billion yuan, 76% of revenue (36Kr / Huxiu). Huxiu adds the share rose from ~65% two years earlier—commerce has not yet replaced ads as the revenue engine.

Merchant reality:

• Seeding, search placement, and feeds all cost budget—CAC is explicit;
• In downturns, brands cut marketing first—ad-heavy platforms feel it upstream;
• >60% merchants losing money or barely breaking even (36Kr) coexists with "great seeding"—exposure and decision value ≠ shop profit, especially when mid-tier catalogs homogenize and users "watch but don't buy."

Huxiu describes Xiaohongshu commerce as buyer taste + content driven—headline live sellers move millions, but catalog depth is thin; ten livestreams later, users see the same brands. Viral stories are not the median.

Last mile: IPO story vs operating gap

36Kr: to be a real e-commerce platform—not a "seeding router"—Xiaohongshu still lacks the last mile: seed, decide, buy, repurchase inside the app. Huxiu's 0.7%–1.2% conversion quantifies the gap.

The platform is patching the loop—buyer livestreams, RedShop overseas, World Cup breakout. Breakout has cost: 36Kr cites media reports of nearly 1 billion yuan in World Cup rights; female users' daily time down 7%, note-to-product click conversion down 11% (Huxiu also discusses dilution of the core female experience). Trading scale for mindshare can erode the "authentic share" atmosphere high-ticket conversion depends on.

Huxiu cites AiMedia: 63% of users who quit rank "too much false information" first. Once soft ads and fake seeding wear trust down, the high-ticket edge shrinks—a longer ceiling than GMV growth rate.

How merchants should read "high AOV, hard scale"

Not "avoid Xiaohongshu"—don't paste Douyin-style GMV targets on it.

1. Role: taste, reputation, long decision chains above ¥3,000; national volume and low-price runs hit pool size and on-site conversion first.
2. Split metrics: seeding (search, saves, off-site traffic) vs in-app GMV—don't let one viral note carry whole ROI.
3. Cost structure: in a 76% ad-revenue platform, free traffic is bonus, not floor—price "decision-scene exposure."
4. Loop strategy: if checkout is mostly off-platform, align Xiaohongshu KPIs with Tmall/private domain—or content teams sweat while revenue lands elsewhere.

Xiaohongshu's high-ticket reputation field: sharp on decision, dull on scale. Categories with the cited +47% edge deserve heavy content; behind 72% growth and 850B GMV (Huxiu) sit 76% ad dependence, 0.7%–1.2% on-site conversion, and >60% merchants on thin margin—scaling usually means accepting strong seeding and weak closure, with a second pot ready off-platform.

Sources

• Three content-commerce platforms in 2026 (36Kr / XinEntropy)
• Xiaohongshu seeding economics at the IPO door (Huxiu)