Measuring KOL ROI in China: Metrics That Matter Beyond “Likes” and “Views”

 

Picture this. A global beauty or luxury brand reviews its latest campaign in China. The report looks impressive, with millions of views. High engagement on both Douyin and Xiaohongshu, but sales barely move.

This gap between visibility and actual revenue is a common issue for Western brands entering China. The problem is not the campaign. It is how performance is measured.

 

While “likes” and “views” look good on paper, they don’t accurately reflect the real-world impact of the campaign in the Chinese market. Likewise, while those numbers can be extremely high, they typically fail to translate into meaningful business results.

China’s digital ecosystem operates differently. It is highly commercial, algorithm-driven, and filled with inflated engagement signals. To succeed, brands need to shift their focus away from surface metrics and towards what actually drives growth: audience intent, search behaviour, and conversion.

This guide breaks down how to measure KOL (Key Opinion Leader) performance using metrics that reflect real ROI (Return on Investment) across China’s key platforms.

The Trap of Vanity Metrics in the Chinese Market

Before building a reliable measurement framework, it is important to understand why traditional metrics fail in China.

In Western platforms, a like or view usually reflects some level of real interest. In China, that assumption does not hold.

One major issue is the presence of “Shuijun” (Water Army), an automated click farm that inflates likes, views, and comments. Many KOLs and agencies face pressure to deliver strong numbers, which leads to artificial engagement. If success is measured using these metrics, it unintentionally rewards fake performance.

The second issue lies in how algorithms work. The entertainment/engagement model used by platforms such as Douyin does not prioritise a user’s purchasing intent; therefore, one video can receive millions of views, purely based on its entertainment value, with no indication that the viewer has an interest in buying the product.

This creates what many marketers call “zombie traffic.” Users will view, click, and scroll through content, then leave without recalling the brand or making a purchase.

Due to this issue, metrics such as Cost Per View (CPV) are also misleading. A campaign could produce very low CPV numbers and reach large audiences; however, if there was no purchasing intent from the audience, the total return on investment (ROI) would essentially be nothing.

The focus should shift from how many people saw the content to who saw it and what they did next.

Redefining ROI: The O-5A Audience Transition Model

To measure KOL effectiveness in China, marketers need to move beyond single-post performance and focus on how audiences progress towards purchase.

This is where the O-5A model, developed by Ocean Engine, becomes essential. It shifts the focus from isolated metrics to brand asset accumulation over time.

In China’s crowded market, consumers rarely buy after a single exposure. Instead, they move through multiple touchpoints before making a decision.

• A1 (Aware): The user sees your content.
• A2 (Appeal): The user shows initial interest.
• A3 (Ask): The user actively searches, saves, or investigates.
• A4 (Act): The user makes a purchase.
• A5 (Advocate): The user shares feedback or creates content.

Most brands focus too heavily on A1. That is where vanity metrics live.

The real measure of ROI is how effectively a KOL moves users into A3 (Ask) and A4 (Act).

A campaign with millions of views but minimal search activity signals a weak impact. Conversely, a smaller KOL who drives strong search behaviour and purchase intent delivers far more impact.

What you need to understand is that it’s not about how many people view your content. It is about how many of those viewers take the desired next step.

Your measurement framework should prioritise audience transition rates, especially movement into high-intent stages.

Deep Dive: Measuring Xiaohongshu (Little Red Book) ROI

Xiaohongshu (RED) is one of the most important platforms for building trust in China. It works less like a social feed and more like a search engine where users research products before buying.

For beauty, fashion, and luxury brands, this is where “Zhongcao” happens, the moment users develop real interest in a product.

Because of this, ROI on Xiaohongshu is not measured by reach. It is measured by intent and influence on decision-making.

The Superiority of Saves and Collections

On Xiaohongshu, likes do not carry much weight.

A user might like a post because it looks good. That does not mean they plan to buy anything.

A save, on the other hand, signals intent.

Users typically save information so they can refer back to it at some point in time, primarily before purchasing an item. This behaviour is common for products with longer consideration periods, such as beauty items and clothing.

For this particular example, your best indicator would be the save-to-view ratio.

• High saves → strong product interest
• Low saves → content is visually appealing but not persuasive

Measuring Share of Search Voice (SoSV)

The content posted to Xiaohongshu typically does not go away fast. Traffic and conversion from good content can be generated long after it has been posted.

That is why long-term ROI is measured through Share of Search Voice (SoSV).

Instead of focusing only on post engagement, track:

• Brand name searches
• Product-specific searches
• Category keywords

A successful campaign should increase search demand over time.

This will also help to increase your presence on search results. Consistently appearing near the top of the list as people search for keywords related to your brand is creating an asset that will be around in the future, and not just a short-lived campaign.

The Comment Sentiment and Keyword Density

Users have a preference for detail on Xiaohongshu, especially when it comes to items within the categories of beauty and luxury/high-end. Comments that users leave on the content indicate whether the users found the content helpful or not.

Strong indicators:

• Questions about price or availability
• Mentions of product features
• Comparison with competitors

Weak indicators:

• Comments unrelated to the product

If most discussions are not about your product, the content fails to communicate its value.

Deep Dive: Measuring Douyin (TikTok China) ROI

Douyin was made for velocity and exposure with instant transactional capabilities. While it’s great at generating high levels of reach quickly, it is also the platform that most brands misinterpret their performance on.

High views do not guarantee real impact. The only way to accurately measure your Douyin ROI is by measuring attention, user behaviour, and actual revenue.

Beyond the View: 3-Second Retention and Completion Rates

A view on Douyin means very little on its own.

The platform pushes content quickly, and users scroll fast. Many views happen without real engagement.

To understand if content is working, focus on:

• 3-second retention rate → Did users stay past the opening?
• Completion rate → Did they watch until the end?

The first few seconds determine whether the content captures attention. A high drop-off early on signals weak hooks, even if total views look strong.

Completion rate is a stronger indicator of engagement. If users stay until the end, the content likely held their attention and delivered the message effectively.

The Illusion of Live GMV vs. Settled GMV

Douyin live streaming can generate tremendous brand awareness and massive GMV (Gross Merchandise Value) figures quickly.

Brand partnerships with top-tier influencers can create an appearance of success based on views and large GMV figures generated during the session.

However, Live GMV may be deceiving.

China’s e-commerce system is heavily regulated by laws protecting consumers. For example, “There is a 7-day no questions asked return policy.” Douyin’s live commerce is designed to drive impulse purchases using many different methods of creating pressure. Some examples include limited-time offers, countdown timers, and exclusive discounts.

Returns and cancellations can be extremely high because of these factors. It is not uncommon to have 50-70% of all orders returned to a merchant after a significant live event.

Additionally, some merchants experience what is called “brushing”, which is when someone makes a purchase fraudulently to increase the GMV figure to achieve performance objectives. Others will simply cancel their orders due to uncertainty and regret after making an impulsive decision.

Therefore, Live GMV can significantly skew performance results. After accounting for returns, shipping costs, and other platform fees, successful-appearing campaigns can ultimately result in losses.

Measuring the ‘Spillover’ Effect

Douyin’s influence rarely exists exclusively within the confines of the Douyin app.

Many Chinese consumers will first come across products via Douyin (e.g., influencer videos) and subsequently make their final purchase on an e-commerce platform such as Tmall or JD.com, where the consumers trust product authenticity and customer service.

This creates a major attribution challenge. A user may discover your product through a Douyin video, then leave the app and open Taobao or JD to complete the purchase. When that happens, the sale is no longer directly tied to the KOL’s tracking link.

To measure the “spillover” effect created by KOL campaigns, brands need to use time-series analysis. Time-series analysis allows you to compare when your Douyin content went live versus when traffic to your branded keywords on e-commerce platforms like Tmall increased.

For instance, if your KOL campaign launches at 8:00 PM and immediately after (between 8:15 PM to 10:00 PM) sees an increase in branded keyword searches and conversions on Tmall, that increase should be attributed as part of your KOL campaign’s halo effect.

In short, some of the greatest value in ROI from Douyin occurs after the user leaves Douyin and converts somewhere else.

Deep Dive: Measuring WeChat and Bilibili ROI

Xiaohongshu and Douyin are currently leading the charge at the top and middle of the funnel. However, WeChat and Bilibili each have their own distinct role within the overall path consumers take through China’s digital landscape. Therefore, they require unique methods to measure performance.

WeChat: Deep Reading and Private Traffic Transition

Long-form storytelling, CRM, and high-value customer relationships continue to be the heart of all marketing activities on WeChat. Brands create extensive, magazine-style content using official accounts, which are used to tell stories about heritage, craftsmanship, and detail-rich products.

Total reads on WeChat are not as important a statistic.

Because the platform operates as a closed network, content is mostly seen by existing followers unless it is actively shared. That makes the Read-to-Share ratio (also known as the “Wow” or “In Top Stories” metric) a more meaningful indicator of performance.

Measuring the effectiveness of an in-depth article from a KOL is best achieved by analysing how many people read the article and then choose to share the article with other readers.

Therefore, the better the KOL performs in terms of getting the majority of those who read the article and also decide to share it, whether through their Moments feed or through direct messaging (private messages), the more successful the KOL is at conveying value to readers.

WeChat also plays a key role in building private traffic (Si Yu Liu Liang).

Private traffic refers to audiences that the brand owns and can reach directly without relying on paid ads. This is usually managed through Official Accounts, Mini Programs, and direct connections with sales representatives.

Because of this, the most important ROI metric is conversion into private traffic.

Track:

• How many users scanned a QR code to follow your account?
• How many added your sales contact or brand assistant?
• How many entered your Mini Program ecosystem?

A strong private traffic pool reduces future acquisition costs and creates long-term value beyond a single campaign.

Bilibili: Engagement Depth and the Danmu Metric

Bilibili has been described as one of the top long-form video platforms in China and is home to many of China’s strongest Gen Z and niche communities. As such, it lends itself to content that has a certain level of depth, for example, tech reviews, tutorials, and product demonstrations.

Bilibili stands out because viewers can use the Danmu (bullet chat) feature to send real-time comments that are superimposed on the video at specified points in time.

This creates a shared viewing experience and provides detailed feedback.

When evaluating a Bilibili KOL (or “UP Host”), focus on:

• Danmu density → How active the audience is
• Sentiment → What viewers are saying
• Timing → When reactions happen during the video

Because comments are tied to exact moments, you can identify which parts of the content resonated most. For example, which parts of the content generated the most positive or negative responses.

Another aspect of how Bilibili encourages engagement is through a system where viewers have the ability to pay “Coins” to support the creators.

The coin-to-view ratio used by each creator is an excellent indicator of both the quality of their content and the degree to which their audience trusts them. A high ratio suggests that the creator has built a loyal community that values their content and supports their recommendations.

Building Your KOL Measurement Dashboard: A Step-by-Step Guide

To move from passive reporting to real optimisation, brands need a structured, multi-platform measurement dashboard.

Post-campaign reports from KOL agencies are not enough. They are often curated to highlight wins and downplay weak performances. Brands require access to uncurated and third-party validated data to properly measure true ROI.

Step 1: Establish Pre-Campaign Benchmarks

Create a single reference point of all pre-campaign benchmark KPIs measuring influencer performance, as well as KPIs measuring your brand performance before initiating campaigns with your chosen influencers.

You will use Feigua (Douyin), Chanmama (e-commerce), and Miaozhen (multi-platform analytics) to compare past performance.

For the KOL:

• What is their real engagement rate?
• How do their past campaigns perform?
• What are their live stream return rates?

For your brand:

• What percentage of the total search voice does your brand currently have on Xiaohongshu?
• What is your brand’s average number of organic searches per day on Tmall?

You will be unable to measure the degree of improvement generated by your campaign without establishing benchmark measurements.

Step 2: Implement Rigorous Tracking Mechanisms

Attribution in China is complex due to platform restrictions. For example, Douyin often blocks direct links to Taobao.

To track performance accurately, you need alternative methods:

• Exclusive Promo Codes: Assign a unique code or phrase to each KOL. When used at checkout on Tmall or WeChat Mini Programs, it directly links sales to that influencer.
• Customer Service Keywords: Ask users to message your customer service channel (such as WangWang) with a specific phrase to unlock an offer. This helps track even if users do not purchase immediately.
• Platform-Specific Storefronts: Use native integrations like Douyin stores or Xiaohongshu shops whenever possible. This keeps users within the platform and allows clearer, real-time attribution.

Step 3: Post-Campaign Time-Series Analysis

After the campaign concludes, do not rely only on total sales numbers. Instead, map content activity against performance across platforms.

Track:

• Timing of content releases
• Search volume changes (e.g., Baidu, in-platform search)
• Traffic and conversion spikes on Tmall

As an example, if you release multiple Douyin video content pieces on Friday evening and then observe an increase in search traffic Saturday morning, that uplift is most likely associated with your campaign.

In addition to conversions, reviews have the opportunity to influence future purchases; they also have the ability to produce a long-term impact (over days/weeks).

By connecting these benchmarks, you obtain a better understanding of how each consumer navigates their customer journey and allocate the correct level of value to that customer at every stage of their journey (from first impression to final purchase).

The Necessity of Localised Expertise

Executing this level of multi-platform analysis from outside China presents real challenges. The platforms operate in Chinese, backend systems are complex, APIs change frequently, and interpreting user behaviour requires cultural context that automated tools cannot fully capture.

Without local expertise, measuring true ROI becomes difficult. Having data is one thing. Interpreting it correctly is another.

For example, there are many luxury, fashion, and beauty companies using partner agencies specialising in the Chinese market.

The approach at Red Ant Asia does not rely solely on surface-level metrics when evaluating campaign success. All aspects of customer interaction, including first impression exposure on Douyin, product research on Xiaohongshu, verification of purchase through Tmall, and even interaction through private WeChat communities, are used to measure campaign success.

We want to determine the difference between actual consumer influence and artificially high levels of engagement and make sure our clients’ marketing dollars are working towards long-term brand building and tangible returns on their investments.

Competition in China’s digital market is intense. The ability to achieve success in this type of environment will require us to understand and interpret the information contained in our data accurately and continually assess whether or not we are meeting our goals.

If most, if not all, of the current reporting relies on using data points (such as views and likes), it may be time for you to review and potentially revise your reporting process.

FAQ: Which Influencer Metrics to Track on Chinese Social Media Platforms?

1. Why shouldn’t I use “likes” to measure KOL success in China? Using likes as an influencer metric in China is inadvisable because they are considered a vanity metric. Likes are easily artificially inflated through click farms (also known as water armies) and typically don’t reflect the level of user intent required for a view or other engagement metrics. As such, “likes” are an unreliable metric to determine if brands will achieve brand recall, search behaviour and purchase intent through Social Media Platforms (e.g., Douyin) in China.
2. What is the O-5A model and why is it important? The O-5A model provides insight into how users move from Awareness (A1) to Advocacy (A5). Instead of just tracking impressions, the O-5A model tracks whether a KOL drives users towards Active Interest (A3), Conversion (A4), etc., which are more credible indicators of return on investment (ROI).
3. What is the most important metric to track on Xiaohongshu (RED)? The save-to-view ratio. When a user chooses to save something, they’ve determined the content has enough value to them, so they will return to it at some point before purchasing. A high number of saves will indicate strong consumer Intent, as such, a strong brand or product position.
4. How do I measure the long-term ROI of a Xiaohongshu campaign? You must track your brand’s Share of Search Voice (SoSV). If a campaign was successful, then there should be continued growth in search volume and increased visibility within search results.
5. What is the difference between Live GMV and Settled GMV on Douyin? Live GMV represents all orders generated through a live stream, while Settled GMV represents revenue after returns. However, in Live GMV, after consumers cancel their orders or receive refunds for returned items, that money no longer belongs to you. Because of this, returns can greatly affect ROI; therefore, whenever possible, it is best to utilise Settled GMV as opposed to Live GMV.
6. How can I track sales if Douyin blocks links to my Tmall store? Spillover attribution. Spillover attribution uses various methods, including promotional codes, customer service keywords, and temporal analysis, to establish direct correlations between campaigns and subsequent spikes in searches leading to purchases.
7. What is “Private Traffic” and how do I measure its growth? Private traffic consists of the people you can contact directly via channels like WeChat. Private Traffic growth is tracked by monitoring the number of users who follow your account, join your ecosystem, and connect with your brand after a campaign.
8. How do you measure sentiment and engagement on Bilibili? Danmu (bullet chat) and coin-to-view ratio. Danmu provides a snapshot of real-time audience reaction to content, and the coin-to-view ratio provides insight into whether or not the content provides significant value to the community, thereby establishing trust.
9. Why is video completion rate more important than views on Douyin? Views may occur quickly as users scroll past content. Video completion rate, on the other hand, demonstrates that the viewer stays engaged with the content for an extended period and therefore provides a more accurate measurement of content retention and messaging retention.