How to Spot Fake Influencer Followers in China: The Ultimate Due Diligence Guide for Western Marketers

Chinese social media platforms operate at a massive scale and can drive trends into mass consumer behaviour overnight. KOLs (Key Opinion Leaders) and KOCs (Key Opinion Consumers) have become key drivers of building brand trust and generating Gross Merchandise Value (GMV).

However, due to the sheer size of these social platforms and the need for many marketers to show instant and continuous growth and Return On Investment (ROI), many face a very simple yet very important question: How do I know that my influencer has ‘real’ followers? Many influencers use bots or fake accounts designed to mimic human behaviour, often created through the use of large-scale mobile phone farms or virtual systems.

 

A large industry has developed around fake followers for influencers; behind each successful viral campaign and every record-breaking live stream is a multi-million pound industry centred on fake follower creation.

The pressure to constantly grow and provide fast ROI on campaigns for influencers and MCNs has led to the development of what is known as the ‘ShuiJun’, or Water Army. This term refers to a network of people who create artificial followers, artificial engagement, and even artificial sales performance metrics.

A CMO viewing campaign results may see high impressions and engagement, which would appear to be indicators of success. However, if these numbers were produced from fake followers, the entire investment of both time and money for the campaign would be lost.

In this blog, we will describe the practice of influencer fraud in China, demonstrate how it appears across many of the different social media platforms available in China, and provide tips for performing a thorough background check (due diligence) when considering an influencer partnership.

The Industrialisation of Fraud: Understanding the ‘Shuijun’

To spot influencer fraud in China, you need to understand how advanced it has become. The term Shuijun, or ‘Water Army’, describes large-scale networks that flood platforms with fake traffic.

In the early days of Weibo, fake followers were easy to identify. Fraudsters relied on basic scripts that created ‘zombie’ accounts with random usernames, no profile photos, and no activity. That is no longer the case. Today’s Water Armies operate as organised, tech-driven businesses.

Modern click farms go beyond simple automation. They also use large numbers of smartphones, each with its own SIM card and IP address to avoid detection by platforms. These smartphone devices then simulate normal user activity on an app or social media site—for example, scrolling through content at varying rates, viewing videos, clicking hyperlinks, and posting comments created from AI-powered language models.

The responsibility does not sit with influencers alone. In many cases, Multi-Channel Networks (MCNs) are directly involved. Faced with tight margins and aggressive performance targets, some MCNs treat purchased traffic as a standard operating cost. They combine real and fake engagement to meet KPIs, creating data that looks credible but is difficult to verify.

This is the reality Western marketers are dealing with. You are not just identifying obvious bots; you are analysing patterns within manipulated data.

Platform Nuances: How Fraud Manifests Across the Ecosystem

China’s digital landscape consists of multiple independent ecosystems. Because each platform has its own algorithm to govern the behaviour of its users, as well as how it recognises engagement, fraudulent actions by influencers appear differently depending on the platform.

A complete audit requires an understanding of the type of manipulation that takes place on all of the various platforms.

Xiaohongshu (Little Red Book): The Battle for Authentic ‘Zhongcao’

Xianghongshu (Little Red Book) is one of China’s largest platforms where fashion and beauty brands connect with consumers. Users write in-depth, experience-based content as reviews about products and services they have tried. Because users are searching for products to purchase on the site, the Xiaohongshu algorithm gives priority to saves, favourites, and meaningful comments.

Fraud here focuses on manipulating these engagement signals. Instead of inflating followers, fraudsters boost post-level performance through ‘engagement pods’, where groups of influencers systematically like and save each other’s content to trigger algorithmic reach.

A very obvious red flag that a post is artificially generated is a very low amount of unique user interaction in comparison to other posts in the same niche. Real users tend to give much more detail about their experiences with products, usually leaving detailed feedback including performance and results.

Douyin: The Mirage of Viral Reach and Live Commerce

The Douyin platform (the Chinese sister app to TikTok) is based upon an extremely high-speed, interest-driven algorithm that evaluates how much users watch videos (i.e. their ‘watch-time’), the completion rate for each video and, most importantly, ‘early-engagement’.

Fraudulent activity here relates to artificially increasing view counts and the number of people watching live streams. For example, although a live session may display hundreds of thousands of viewers in the live feed area, many could be automated ‘bots’ designed simply to increase the overall visual presence of the live stream versus actual potential customers.

Some networks go further by simulating interaction. Bots flood the chat with messages like ‘Bought it!’ or ‘Restock please!’ to create urgency and push real users to purchase. If you are evaluating a Douyin KOL, do not rely on viewership alone. Always check conversion rates; high traffic with low sales is a strong sign of manipulated engagement.

WeChat: Unpacking the ‘100k+ Read’ Myth

WeChat Official Accounts remain essential for brand storytelling and CRM. A ‘100k+ Reads’ badge is often seen as a major success indicator. However, because WeChat is a closed ecosystem, reaching that level organically requires a highly engaged audience. Fraud here typically involves purchasing read bots.

The key metric to watch is the read-to-like ratio. Genuine articles with tens of thousands of reads usually generate hundreds of likes and active discussion. If a post consistently hits 100k+ reads but shows minimal likes and no comments, the traffic is likely purchased.

Weibo: The Legacy of Zombie Fans

Weibo has the longest history of follower inflation. It is heavily driven by celebrity culture and large-scale campaigns, where follower counts can reach tens of millions.

A significant portion of these audiences can be inactive or fake. Fraud often appears through ‘forwarding armies’, where large numbers of low-quality accounts repost content without real engagement. Follower count on Weibo is not a reliable metric; the real signal lies in the quality of discussion within the comments.

The Illusion of Success: Unpacking Fake GMV and ‘Brushing’

One of the most damaging forms of influencer fraud in China goes beyond fake engagement and into direct financial manipulation. This is known as fake Gross Merchandise Value (GMV), often driven by a tactic called ‘Shadan’, or brushing.

Western brands increasingly prefer performance-based deals, where influencers are paid based on sales rather than flat fees. Whilst this sounds effective in theory, it has opened a new path for fraud.

A typical scenario looks like this: a mid-tier beauty KOL guarantees £100,000 in sales during a live stream, in exchange for a large upfront fee and exclusive pricing. During the stream, sales surge. Inventory sells out. The campaign appears successful, with GMV exceeding expectations. Then the problem begins: within one to two weeks, most of those orders are cancelled or refunded—often 70% to 80%.

This is how brushing works. The KOL’s agency or a network of buyers uses their own funds to place bulk orders during the live stream. This inflates GMV, secures the influencer’s fee, boosts their ranking in platform algorithms, and creates strong case studies for future deals.

Because Chinese e-commerce sites allow consumers to cancel or return items purchased online during a specific timeframe, many of these purchases are cancelled before payment is finalised. This leaves the brand covering shipping and operational costs whilst the influencer keeps the fee and reputation. Brands must think differently about measuring success. Live GMV is not reliable; the focus should be on Settled GMV—revenue that remains after the return window—and the actual return rate compared to your baseline.

Quantitative Red Flags: Auditing the Metrics

Protecting your brand requires moving from passive to active auditing. Screenshots from MCNs are not enough. You need third-party tools and a structured way to evaluate performance data.

1. The Sawtooth Follower Growth Curve: Organic growth tends to trend upward over time with small spikes from viral content. Repeated large jumps (tens of thousands of new followers in one day) followed by a slow decline suggest paid followers. The spikes represent mass purchases and the declines represent the platform’s removal of fake accounts.
2. Disproportionate Engagement-to-Follower Ratios: There are normal ranges for each platform. For a mid-tier Xiaohongshu KOL, the engagement rate typically falls between 3%–8%. If an account with 500k followers produces 80,000 likes per post, the numbers are likely artificially inflated.
3. The ‘3 AM Boom’ (Anomalous Temporal Engagement): Engagement patterns should mimic natural human behaviour—peaking in the late afternoon and dropping off in the early morning. Sudden increases in likes and comments between 2 am and 5 am Beijing Time indicate probable bot traffic.
4. Zero Variance in Engagement Metrics: Real content varies in performance. Consistent engagement numbers across all types of content suggest that artificial engagement packages have been purchased to create a ‘perfect’ image.

Qualitative Red Flags: The Human Element

1. The Anatomy of a Bot Comment: Look for hyper-generic praise (‘It is beautiful’, ‘Nice’), irrelevant tangents that don’t reference the content, or repetitive formatting where multiple users use the exact same sentence structure.
2. The Absence of Conversational Depth: Genuine influence creates conversation. If a comment section has lots of praise but lacks any questions, debate, or response from the creator, the audience is likely not real.
3. Profile Audits of the Commenters: Check who is responding. If commenters have no original posts, poor-quality content, or seemingly unrelated usernames, they are unlikely to be genuine customers.
4. Assessing the ‘Commercial to Organic’ Ratio: If virtually all of an influencer’s recent posts are sponsored, the audience may have become desensitised. Even true followers begin to tune out if every post is an advertisement.

Contractual Shields and Legal Protections

Identifying fraudulent activity via vetting is important, but insufficient. Brands require strong contractual protections. Standard Western influencer contracts are not suited for the Chinese market; agreements should be localised, ideally bilingual, and aligned with Chinese law.

Your contracts should include:

• Anti-Fraud Clauses: Prohibiting the use of click farms or bots.
• The Right to Audit: The right to use third-party software (Miaozhen, Feigua, or Chanmama) to confirm performance data.
• Performance Penalties: Financial penalties deducted from compensation if red flags (like the 3 am engagement boom) cannot be reasonably explained.
• Settled GMV Benchmarks: Tying performance bonuses to Post-Return Revenue, not Live Sales figures.

The Red Ant Asia Due Diligence Framework

Managing China’s influencer landscape from overseas puts brands at a disadvantage. Without local insight and access to platform-level data, it is difficult to spot manipulated metrics.

Red Ant Asia goes beyond the numbers presented in MCN pitch decks. Our framework helps Western luxury, fashion, and beauty brands identify real influence and avoid wasted spend:

• Proprietary Data Tracking: Analysing follower growth and engagement timing using advanced Chinese analytic tools.
• Deep Semantic Sentiment Analysis: Auditing comment sections to separate real purchase intent from bot-generated engagement.
• E-commerce Reality Checks: Evaluating historical return rates and brushing patterns for live commerce.
• Ironclad Local Contracting: Ensuring strict anti-fraud clauses are in place to protect your budget.

Conclusion: The Pivot to Authentic Influence

The era of blind spending in China’s digital ecosystem is ending. Identifying fake followers is necessary to protect your marketing investment. Brands that go beyond vanity metrics and focus on reliable results will see real returns. Authenticity and trust are what Chinese consumers expect; working with creators who offer true influence is the only way to establish a long-term strategy.

FAQ: How to Spot Fake Influencer Followers in China

1. What exactly is a ‘Water Army’ (Shuijun) in Chinese social media?
   A ‘Shuijun’ is essentially a network of bots and click farms used to make engagement look real by artificially increasing numbers.
2. Are all Chinese influencers using fake followers?
   No. Many have real, active bases. However, some influencers supported by MCNs are pushed to artificially increase their base to remain competitive.
3. Why do MCNs allow or encourage fake engagement?
   In an extremely competitive space, MCNs use purchased traffic to create KPIs that make it appear a creator is successful to secure advertising spend.
4. How does fraud on Xiaohongshu (RED) differ from fraud on Douyin?
   RED fraud typically involves manipulating comments and ‘saves’ to trigger the algorithm. Douyin fraud usually involves inflating video views or live stream viewer counts.
5. What is ‘Brushing’ or ‘Shadan’ in Chinese e-commerce?
   It is when an agency purchases a massive amount of products during a live stream to inflate sales numbers, only to return them later.
6. How can a brand tell if a KOL’s followers are mostly bots?
   Look for unusual patterns like sudden spikes in growth followed by steady declines, or high follower counts with very low unique engagement.
7. What is the ‘3 AM Boom’ when auditing an influencer?
   It refers to unusually high engagement between 2 am and 5 am Beijing time, which is typically driven by automated bot activity.
8. Can I rely on the screenshots provided by the influencer’s agency?
   No. These can be easily manipulated. Always verify data through third-party analytics tools like Feigua or Miaozhen.
9. What legal clauses should a Western brand include?
   Include anti-fraud provisions, the right to audit data, and tie bonuses to settled GMV rather than live sales.
10. Why is qualitative analysis of comments important?
    Numbers alone do not represent influence. Qualitative analysis determines if the audience is legitimately engaged; real users ask questions, whilst bots post generic, repetitive messages.