KOL vs KOC on Xiaohongshu: You Are Buying Index Coverage, Not Reach
Amateur mirror-selfie posts beat polished creator content on Xiaohongshu, and authenticity is not the reason. The reason is that Xiaohongshu is a search index with a feed attached, which changes what a note is worth and how a seeding budget should be built.
Annie Chan··15 min read
Every Western brand that has run a Xiaohongshu programme has noticed the same anomaly. You pay a well-known creator with half a million followers for a beautifully shot note, and it does fine. You pay forty ordinary users a few hundred yuan each for phone-camera photos taken in a bathroom mirror, and that collection of forty outperforms the creator, keeps outperforming for months, and shows up in your search traffic a year later.
The standard explanation is authenticity: Chinese consumers distrust polished advertising and trust peers. That explanation is comfortable, partially true, and useless as a planning input, because it does not tell you how many creators to book or which ones.
The actual mechanism is structural. Xiaohongshu is a search index that happens to have a feed attached. A KOL note is a broadcast asset that decays within days. A KOC note is inventory in a retrieval system that keeps serving it for as long as people keep asking the question it answers. You are not buying reach. You are buying coverage of queries.
Once you accept that, most of the standard advice about budget splits stops making sense, and a different decision rule replaces it.
The platform is a search engine that was mistaken for Instagram
Xiaohongshu's core consumer behaviour is not scrolling. It is typing. Users go to the app with a question, type it in, and read a stack of first-person notes from people who have already bought the thing. The feed exists, and it matters for discovery, but the commercially decisive surface is search.
Estimates of how dominant that behaviour is come mostly from agencies and platform marketing material rather than audited disclosure, and they cluster around 70 percent of monthly active users performing product searches. Treat that as a directional figure from interested parties, not a measured one. The behaviour itself is not in doubt, and any Western brand operator who has watched a Chinese consumer use the app for ten minutes has seen it.
The scale figures are similarly soft, and it is worth being explicit about why. Xiaohongshu is private and discloses selectively. Third-party compilations have put monthly active users at around 350 million at the end of 2024, 300 million in figures circulating through 2025, and above 400 million in company statements reported in 2026. Those numbers cannot all be measuring the same thing. Sacra's equity research estimates full-year 2025 revenue at close to 42 billion yuan, up roughly 40 percent, with advertising rather than e-commerce contributing the larger share at around 76 percent. That last detail is the one worth holding onto, because it tells you what the platform is optimising for and therefore what it will charge you for.
"A platform whose revenue is three-quarters advertising is a platform that will progressively make organic distribution harder. Plan your seeding as an asset that has to survive that."
Why the mirror selfie wins
Three reasons, in order of how much they matter, and only the third is about authenticity.
It matches the query, not the mood
A search result is judged against an intent. Someone typing a query about, say, whether a particular moisturiser works on sensitive skin in a dry northern winter is looking for a specific answer, and a note that names that exact situation in its title and first line will beat a beautiful note that does not, regardless of who posted it. KOC notes are written by people who had the specific problem, so they naturally use the specific language. KOL notes are written to a brief, and briefs are written in brand language.
It compounds instead of decaying
A creator note gets its distribution in the first 48 to 72 hours and then flattens. A note that ranks for a query gets a smaller daily number that does not stop. Over twelve months the areas under those two curves are not close, and the seeded note is also cheaper by an order of magnitude.
This is the single largest measurement error in Western Xiaohongshu reporting. Programmes are reviewed at 30 days, which is exactly the window in which the expensive broadcast asset looks best and the cheap index asset has barely started.
It reads as evidence rather than as advertising
This is where authenticity actually enters, and it is narrower than the usual claim. The reader is not naive. They assume a proportion of what they are reading is paid. What they are doing is triangulating: if nine unconnected-looking people describe the same product with roughly consistent detail and inconsistent phrasing, the claim survives. One immaculate note from a famous account does not survive that test, because it is a single source with an obvious incentive.
Which means the value of KOC content is not in any individual note. It is in the corpus. A single seeded note is worth close to nothing. Forty of them, covering overlapping queries with genuinely varied phrasing, are worth a great deal. Brands that seed twelve notes and conclude that seeding does not work have measured a corpus too small to do the job.
The decision rule nobody gives you
Agency guidance on this question converges on percentage splits: 60/40 in favour of KOLs for awareness budgets, 70/30 in favour of KOCs for conversion budgets, 60/70 percent to KOCs for new market entrants. You will find those exact ranges repeated across a dozen agency blogs, sourced to nothing.
The splits are not wrong so much as they are answering the wrong question. The real question is prior to budget allocation, and it has a binary answer.
Does the category word already exist in Chinese consumer vocabulary on this platform? If yes, your money goes into index coverage and KOCs do the work. If no, KOC volume is wasted until someone creates the word, and that job requires a KOL.
The searched category
Suppose you sell a sunscreen. Chinese consumers search for sunscreen constantly, with enormous query variety: by skin type, by season, by whether it leaves a white cast, by whether it works under makeup, by whether it is safe during pregnancy, by city. The demand exists and is already expressed in queries. Your problem is that none of those queries currently return you.
In this situation, spending on a top creator buys you a spike of attention in a category where the consumer already knows what they want and will go and search for the best option, which is not necessarily you. Spending the same money across forty to eighty seeded notes that each answer a distinct query buys you presence at the moment of decision, repeatedly, for a year. The second is obviously better and it is not close.
The unsearched category
Now suppose you sell something for which no Chinese search demand exists, because the concept has no established consumer name. A specific supplement format, an appliance category that has not landed, a service model with no local analogue. Your product is the answer to a question nobody is asking yet.
Here KOC volume genuinely is wasted. You can seed two hundred notes and they will sit in an index that nothing queries. What you need first is demand creation, which is a broadcast job: a creator with reach and narrative credibility who can make a lot of people want a thing they had not previously considered, and in doing so establish the words those people will subsequently type.
The sequence is then: KOL to mint the category word, monitor search volume on that word until it moves, and only then pour money into KOC coverage of the query cloud that has formed around it. Doing those two in the wrong order, or simultaneously, is the most expensive common mistake in this channel.
How to tell which you are in
Do not guess and do not ask your agency, whose incentive points toward whichever service they sell. Run the queries yourself, or have someone run them, and look at what comes back.
Type your category word in Chinese. If the results are dense, recent, and competitive, you are in a searched category.
Look at the auto-complete suggestions. Long, specific suggestion tails mean an established query cloud. Sparse or unrelated suggestions mean the word is not doing work yet.
Check whether the notes that rank are answering a purchase question or explaining what the thing is. Explainer content ranking at the top is a signal the category is still being defined, which is a KOL problem.
Count the brands present. A searched category with no incumbent doing this well is the best possible position, and it will not last.
What you are actually buying, unit by unit
Rate cards for Xiaohongshu creator collaboration circulate widely and vary enormously. Agency-published ranges in 2025 and 2026 cluster around: accounts under 10,000 followers at roughly 80 to 300 yuan per note, 10,000 to 50,000 followers at roughly 500 to 2,000, mid-tier accounts from 50,000 to 500,000 at roughly 3,000 to 15,000, and accounts above 500,000 anywhere from 20,000 to 100,000.
Those are advertised ranges from agencies with an interest in the numbers, not observed transaction data, and the spread within each tier is wide enough that the midpoint is meaningless. Use them to understand the shape of the market, which is roughly two orders of magnitude between the bottom tier and the top, and negotiate from your own quotes.
The structural point survives the imprecision: at the top of that range, one note costs what forty at the bottom cost. The question is whether one broadcast is worth forty index entries, and in a searched category it very clearly is not.
The declared-collaboration layer
Xiaohongshu runs an official brand collaboration platform, Pugongying, through which paid partnerships are meant to be arranged and labelled. Notes routed through it carry a visible sponsorship marker and the platform takes a facilitation fee. Notes not routed through it, where a brand has paid a creator privately, are undeclared commercial content, and the platform actively suppresses distribution of undeclared commercial notes it detects.
This matters more than it first appears, because it puts a structural ceiling on how far polished paid KOL content can travel. The more a note looks and behaves like an advertisement, the more likely it is to be identified as one. Declared, it is labelled and readers discount it. Undeclared, it risks suppression and, under Chinese advertising law, exposes the brand. Neither branch is good.
KOC content does not escape the rules, and brands should not treat undeclared seeding as a loophole. But content that is genuinely a user describing genuine use sits in a different position from content that is an advertisement wearing a person's face, and the platform's enforcement is calibrated accordingly.
The liability nobody mentions
Here is the consequence of the index framing that I have never seen a Xiaohongshu agency put in a deck, presumably because it is bad for business.
If seeded notes are index inventory that persists, then bad seeded notes are a durable liability. A batch of forty notes written from a poor brief, using clumsy Chinese, making a claim your product does not support, or all repeating the same phrase in a way that reads as manufactured, does not fade away when the campaign ends. It stays in the index. It ranks for your brand name. It is the thing a prospective customer finds in eighteen months when they search for you, and there is no delete button that reaches other people's accounts.
You cannot remove it. You can only bury it, by producing enough better material to outrank it, which costs more than the original batch did.
Treat a seeding batch the way you would treat permanent copy on your own site, because functionally that is what it is. The reversibility you are used to in paid social does not exist here.
The practical implication is that the first batch should be small, slow and heavily reviewed, and it should be reviewed by someone who reads Chinese natively and is willing to reject work. Most Western brands invert this, running a large first batch because the unit cost is low and it feels like a cheap test. It is a cheap test that produces expensive permanent output.
How to structure the spend
Assuming a searched category, which is the more common case for consumer brands, here is a structure that reflects the argument rather than a generic percentage split.
Build a query map before you build a creator list. Enumerate the phrases a buyer in your category would actually type, in Chinese, grouped by intent: what is it, which one, is it right for me, is it real, where do I buy.
Assign notes to queries, not creators to budgets. Each query cluster needs enough independent notes to survive the reader's triangulation test. Five to ten per cluster is a working starting point.
Cast creators to fit the query, not the follower count. The right author for a query about sensitive skin is someone with sensitive skin, whatever their audience size.
Reserve a minority of budget for one or two mid-tier accounts whose job is not sales but supply: they generate the reference imagery, the phrasing and the credibility anchor that the seeded corpus can echo without looking coordinated.
Only then consider a head KOL, and only if you can articulate what they do that the corpus cannot. Launching a genuinely new category word is a real answer. Awareness is not.
Layer paid amplification onto notes that are already ranking organically, not onto notes you wish would rank. Amplifying a note the index has rejected is paying to fight the system that decides your long-term visibility.
Measure at 90 and 180 days, not at 30. If your reporting cadence cannot accommodate that, the reporting cadence will select for the wrong asset every single time.
Frequently asked questions
What is the difference between a KOL and a KOC on Xiaohongshu?
A KOL is a key opinion leader, an account with substantial followers whose content functions as a broadcast: a burst of attention within a few days. A KOC is a key opinion consumer, an ordinary user with a small audience whose note functions as a search result, retrieved by the platform for months or years when someone asks the question it answers. The distinction that matters is not follower count but whether you are buying a spike of attention or a durable position in the index.
Why does amateur-looking content perform better on Xiaohongshu?
Mostly because it matches the specific language of the search queries people type, and because readers judge product claims by triangulating across multiple independent-looking sources rather than trusting any single polished one. Authenticity is part of it, but the primary mechanism is retrieval: the note that uses the searcher's exact words wins, and ordinary users naturally use ordinary words.
How many KOC notes do I need for a Xiaohongshu launch?
The wrong way to ask this is as a total. Ask it per query cluster. Each distinct question you want to own needs enough notes that a reader scrolling the results sees consistency across independent sources, which in practice means at least five to ten. Multiply by the number of clusters your category actually generates. A dozen notes spread thinly across an entire category is the most common failure and it produces no measurable effect.
What does Xiaohongshu KOC seeding cost?
Published agency rate cards in 2025 and 2026 put small accounts under 10,000 followers at roughly 80 to 300 yuan per note and accounts in the 10,000 to 50,000 range at roughly 500 to 2,000 yuan, but these are advertised ranges from interested parties rather than observed market data, and the real spread is wide. Budget from your own quotes and expect the total programme cost to be dominated by Chinese-language editorial review rather than by creator fees.
Should I use the official Pugongying platform or work with creators directly?
Paid collaborations are expected to be declared through the official platform, which labels them as sponsored and applies a facilitation fee. Undeclared commercial content risks distribution suppression and creates exposure under Chinese advertising law. The choice is not really a choice for anything that is clearly an advertisement. Where genuine ambiguity exists, take advice locally rather than from a guide.
Can a foreign brand run Xiaohongshu seeding without a Chinese entity?
Content seeding is possible through agencies without a local entity, but a verified brand account, official collaboration tooling and any commerce function all involve qualification requirements that vary by category and change frequently. Confirm the current position before planning around it.
The summary judgement
The KOL versus KOC framing is borrowed from Western influencer marketing, where both are broadcast assets differing only in scale. On Xiaohongshu they are different asset classes with different lifespans, different measurement windows and different failure modes. Treating them as points on one spectrum is what produces the 60/40 splits, and the splits are why so many programmes produce a good first month and nothing durable.
"Ask what queries you own, not how many people you reached. The first question has an answer that is still true next year."
For the operating layer underneath this, including account structure, ad inventory and what the agency market actually charges, see the Xiaohongshu marketing guide and RedNote for business. If you are still deciding whether the platform is the right first move at all, China market entry strategy covers the prior question.
Share:
Free · 24 pages
Read Chapter 1 of the Playbook
AI Foundation Models — how DeepSeek and Qwen crossed a billion downloads while Western boards were counting Mag7 capex, and which layers of the AI stack the rotation has not reached. No excerpt tricks: it is the full chapter.
You will also get the weekly newsletter. Unsubscribe any time — the link is in every email.
The Book · $99
China Market Entry Playbook 2026
456 pages across 22 industry chapters — AI, semiconductors, EVs, payments, logistics, beauty, F&B, healthcare, pets, toys. Every chapter runs the same five questions: the anchor number, the rotation underneath it, which layers of the global stack it has reached, and what to do about it. English and Simplified Chinese editions included.
Ex-Transsion Global Digital Marketing Director · Effie Awards Greater China jury member · Guest lecturer, Peking University School of Economics · Writes Annie Chan Talk, your insider lens on China.
