Advanced XHS Ad Targeting: Custom Audiences & Interest Layering on Xiaohongshu
Date Published
Table Of Contents
• Why Audience Targeting on XHS Is Different from Western Platforms
• The Four Audience Targeting Pillars on XHS
• Custom Audiences: Leveraging Your First-Party Data
• Interest Layering: Moving Beyond Broad Categories
• Behavioral Targeting: Reading Intent Signals on XHS
• Lookalike Audiences: Scaling What Works
• Building a Full-Funnel Audience Strategy for XHS
• Common Mistakes International Brands Make with XHS Targeting
Most international brands arriving on Xiaohongshu (XHS) — also known as Little Red Book or RedNote — start by running broad demographic campaigns and hoping the platform's algorithm does the heavy lifting. Some see early results. Most quickly hit a ceiling where spend keeps rising but acquisition costs refuse to come down.
The difference between campaigns that scale profitably and those that stall almost always comes down to one thing: audience precision. Xiaohongshu's ad platform has matured significantly, and it now offers a sophisticated suite of targeting tools — custom audiences, interest layering, behavioral segmentation, and lookalike modeling — that allow international brands to move far beyond basic age-and-gender filters. The challenge is that most Western marketers approach these tools using mental models built on Meta or Google, and XHS plays by its own rules.
This guide breaks down every major audience targeting capability available in XHS advertising, explains how they work within the platform's unique ecosystem, and shows you how to layer them strategically to build a full-funnel approach that consistently reaches your highest-value users.
Why Audience Targeting on XHS Is Different from Western Platforms
Before diving into specific targeting tools, it's worth understanding what makes Xiaohongshu's advertising environment fundamentally distinct. Unlike Facebook, which relies heavily on social graph data and pixel-based retargeting, or TikTok, which optimizes primarily around video completion and viral signals, XHS operates as a hybrid engine that blends the aesthetic inspiration of Instagram with the deliberate research intent of a search platform. Users don't scroll XHS passively — they arrive with a purpose, whether that's researching a skincare ingredient, comparing product reviews, or hunting for a trusted recommendation before making a purchase decision.
This behavioral context matters enormously for how you build and layer audiences. On XHS, a user's search history, content saves, and engagement patterns are extraordinarily high-quality intent signals because the platform is where Chinese consumers actively research before they buy. Research indicates that a significant share of younger Chinese consumers now go directly to Xiaohongshu rather than traditional search engines when evaluating products or services — making the platform's behavioral data more commercially predictive than what you'd find on a standard social feed.
XHS's ad platform (Aurora, or 信息流广告 for feed ads, and Chengfeng for search ads) has also developed its own Community Engagement Score (CES) framework, which means the algorithm rewards ads that earn genuine engagement — saves, follows, and comments — rather than just impressions. This creates a reinforcing dynamic: better audience targeting leads to more relevant creative exposure, which generates higher CES scores, which in turn drives down your effective CPM over time. Getting your audience architecture right from the start isn't just a targeting exercise — it directly impacts your algorithmic standing on the platform.
For international brands, there's an additional layer of complexity. Xiaohongshu's targeting parameters are calibrated around its core user base: predominantly female (around 70%), aged 18–35, concentrated in China's tier-one and tier-two cities, and highly attuned to authenticity. Layering in cultural relevance and platform-native behavior patterns is as important as selecting the right interest categories. This is where many Western brands stumble, applying generic interest stacks that might work on Instagram but fail to match the nuanced content discovery behavior of an XHS user.
If you're still building out your foundational knowledge of the platform, exploring industry-specific Xiaohongshu marketing strategies can help you understand how targeting approaches vary across verticals like beauty, fashion, and F&B before you set up your first campaign.
---
The Four Audience Targeting Pillars on XHS
Xiaohongshu's advertising platform organizes audience targeting into four core capabilities, each serving a different strategic purpose. Understanding what each pillar does — and when to use it — is the foundation of any advanced targeting strategy.
Demographic and Geographic Targeting forms the baseline. You can segment audiences by age, gender, location (down to city tier), and language preference. This is necessary but rarely sufficient on its own. Demographic filters narrow the eligible audience before you apply behavioral and interest layers, but treating demographics as your primary targeting method leaves enormous precision on the table.
Interest and Behavioral Targeting allows you to reach users based on what they engage with on the platform — content categories, search behavior, product page interactions, and purchase history within the XHS ecosystem. The platform's algorithm analyzes these signals to build rich user profiles that go well beyond what users explicitly declare.
Custom Audiences let you upload your own first-party customer data — email addresses or phone numbers — to match against XHS's registered user base. This is where your CRM data becomes a direct targeting asset, allowing you to retarget known customers, suppress existing buyers from prospecting campaigns, or segment your customer base by value tier.
Lookalike Audiences use XHS's machine learning to identify users who share characteristic patterns with your best-performing custom audience seeds. This is the primary mechanism for scaling your campaigns beyond your existing customer base while maintaining targeting quality.
The power of advanced XHS targeting comes from combining these pillars in ways that match your funnel stage, budget, and campaign objective. Running each in isolation is a common beginner mistake; the real leverage comes from layering.
---
Custom Audiences: Leveraging Your First-Party Data
For international brands that already have a customer database, custom audiences represent the fastest path to high-efficiency XHS campaigns. By uploading your existing customer data — typically hashed email addresses or phone numbers — you can match against XHS's registered users and build audiences that reflect known purchase behavior rather than inferred interest signals.
The practical application goes well beyond simple retargeting. Sophisticated brands segment their uploaded lists before building custom audiences, separating high-LTV customers from one-time buyers, recent purchasers from lapsed customers, and VIP members from trial-period users. Each segment warrants a different campaign approach. Your highest-LTV customers, for example, are the right seed audience for lookalike modeling (more on that below), while lapsed customers might be the right target for a reactivation campaign with a time-sensitive offer or new product announcement.
Within XHS's ecosystem, custom audiences are also valuable for exclusion logic — a targeting discipline that many brands overlook entirely. When running prospecting campaigns designed to acquire new users, uploading your existing customer list as an exclusion prevents you from spending budget reaching people who have already purchased. This single adjustment can recover a meaningful percentage of wasted prospecting spend that would otherwise go toward converted customers. The same principle applies at the campaign level: if you're running a conversion-focused bottom-funnel campaign, exclude users who have already completed that conversion action to avoid redundant exposure.
One important nuance for international brands: your first-party data quality directly determines your custom audience match rate on XHS. Lists built on international email formats may see lower match rates than domestically acquired contacts, since XHS's user base is overwhelmingly registered with Chinese phone numbers. If you have any China-market CRM data — from WeChat mini-programs, Tmall stores, or previous XHS shop interactions — prioritize that data for your custom audience uploads. It will match at a higher rate and produce more accurate downstream lookalikes.
---
Interest Layering: Moving Beyond Broad Categories
Interest targeting on XHS allows you to reach users based on their demonstrated content preferences — categories like skincare, fitness, travel, food, fashion, luxury goods, mother and baby, and dozens of sub-categories beneath each. The mistake most brands make is selecting one or two broad interest categories and calling it a day. That approach produces large audiences with poor precision, high impression volume, and frustrating conversion rates.
Interest layering is the practice of combining multiple interest signals to define an audience intersection rather than a broad union. Instead of targeting all users interested in beauty, you might target users interested in both Korean skincare routines and clean ingredient formulations — a much narrower intersection that describes someone far more likely to respond to a specific product positioning. This concept, sometimes described as AND-logic targeting versus OR-logic targeting, is what separates campaigns that find genuinely qualified users from those that spray impressions across a loosely defined category.
Effective interest layering on XHS typically works in three stages. You start with a primary interest category that broadly describes your product or service — skincare, outdoor apparel, gourmet food, and so on. You then add a qualifying interest layer that narrows the audience to a more specific sub-set within that category, such as users interested in anti-aging specifically, or trail running rather than general fitness. Finally, you can apply a behavioral qualifier — for example, users who have recently searched related terms or engaged with content in your category — to further concentrate your audience toward higher-intent users.
The critical balance in interest layering is avoiding over-narrowing. If you stack too many AND conditions, your audience shrinks to a size that prevents the algorithm from optimizing delivery effectively. A practical rule of thumb is to monitor your estimated audience size as you build layers, and remove the most restrictive qualifier if the audience drops below a threshold where meaningful data accumulation becomes difficult. Starting broader and refining over time as you gather performance data is almost always more efficient than building an ultra-narrow audience from day one.
For brands operating across multiple product lines or categories — which is common for the beauty, fashion, and lifestyle brands that dominate XHS — consider building separate interest-layered ad groups for each product family rather than a single campaign trying to address all audiences simultaneously. This gives the algorithm cleaner signals and gives you cleaner attribution data to understand which interest intersections are driving the best results for each product.
You can explore how interest targeting aligns with platform-specific content strategies across different verticals in AllXHS's free resources library, which includes data-driven reports across 20+ industry categories.
---
Behavioral Targeting: Reading Intent Signals on XHS
Behavioral targeting on XHS goes a step beyond interest categories by incorporating what users have actually done on the platform, not just what content categories they consume. XHS tracks an unusually rich set of behavioral signals given that users are actively searching for products, reading reviews, saving notes for future reference, visiting brand pages, and completing purchases — all within a single closed ecosystem. This makes XHS behavioral data particularly predictive of purchase intent compared to platforms where commerce is more peripheral to the user experience.
The most commercially valuable behavioral signals on XHS fall into a few categories. Search-based signals capture users who have recently queried specific product categories or keywords — a strong indicator that active research is underway. Engagement-based signals reflect users who have saved notes, commented, or followed accounts related to your category, indicating both interest and investment in the topic. Purchase-intent signals include users who have visited product pages, added items to a shopping cart, or clicked through from content to a store page without completing a purchase. This last group represents the warmest prospecting audience available outside of your own retargeting pools.
Behavioral targeting becomes especially powerful when combined with interest layers. Targeting users who are interested in luxury skincare AND have recently searched for anti-aging products AND have engaged with skincare review content creates an audience with a remarkably concentrated likelihood of converting. This multi-signal approach is what XHS's platform is designed to support, and it's where the gap between brands running basic demographic campaigns and brands running sophisticated audience strategies becomes most visible.
It's also worth noting how XHS influencer campaigns interact with behavioral targeting. When your brand runs KOL or KOC collaborations through the platform's official Pugongying system, every user interaction with those influencer posts — likes, saves, profile visits — becomes a trackable signal. You can retarget users who engaged with a specific influencer's content about your brand, creating a remarkably precise warm audience that has already received a trusted peer endorsement. This feedback loop between influencer content and paid targeting is a capability that few international brands outside China have access to or actively exploit.
---
Lookalike Audiences: Scaling What Works
Once you've identified which custom audiences deliver your best campaign results — whether measured by conversion rate, cost-per-acquisition, or customer lifetime value — lookalike modeling lets you scale those successes by finding new users who share the same behavioral and demographic patterns. XHS's machine learning analyzes the characteristics of your seed audience and identifies users within its 300 million+ active user base who exhibit similar profiles.
The single most important factor in lookalike audience performance is seed quality. A smaller, tightly defined seed audience of your highest-value customers will consistently outperform a larger but more diffuse seed that includes everyone who has ever purchased from you. If you have enough data, build separate lookalike seeds for your top-spending customers, your most loyal repeat purchasers, and your highest-engagement followers — then test which seed produces the most efficient prospecting campaigns. The seed that generates your best converting customers is the one worth scaling.
As your lookalike campaigns generate conversions, implement a dynamic refinement loop. Users who engage with your lookalike-targeted ads but do not convert can be moved into a retargeting pool for sequential follow-up messaging. Users who do convert should be added back into your custom audience seed to continuously improve the quality of signal you're feeding into the lookalike model. This creates a compounding improvement cycle where each successful campaign makes your next prospecting campaign slightly more precise.
One practical consideration for international brands: lookalike audiences on XHS are most effective once you have a sufficiently robust seed audience — typically at least several hundred matched users at a minimum, with larger seeds producing more stable model performance. If your custom audience match rate is low because of data format issues, prioritize building your XHS-native first-party data through on-platform conversions, store interactions, and lead form submissions before scaling lookalike campaigns aggressively.
---
Building a Full-Funnel Audience Strategy for XHS
The most effective XHS advertising approaches treat audience targeting as a layered system aligned to the customer journey rather than a single campaign configuration. A full-funnel audience strategy on XHS typically looks like this:
Top of Funnel (Awareness): Use broad interest targeting layered with relevant behavioral qualifiers to reach new users who fit your brand's category profile but haven't encountered your brand yet. At this stage, creative should be educational and lifestyle-oriented — content that introduces a problem your product solves or a desire it fulfills, without hard-selling. The goal is to move users into your behavioral retargeting pool by generating meaningful engagement: saves, profile visits, and content shares.
Middle of Funnel (Consideration): Target users who have already interacted with your content or brand page, supplemented by interest-layered prospecting for users showing active category research behavior. This is where you introduce social proof — user reviews, ingredient explanations, comparison content — and begin building the product-specific desire that will eventually drive conversion. Behavioral signals like note saves and search activity are strong mid-funnel audience indicators.
Bottom of Funnel (Conversion): Focus on custom audiences built from high-intent behaviors — users who have visited your XHS shop, engaged with shopping-linked content, or searched your brand or product terms directly. This is also where retargeting users who engaged with your KOL campaign content can drive exceptional conversion rates, since they carry both peer endorsement exposure and demonstrated interest. Bids at this stage should be your most aggressive, reflecting the higher conversion probability of this audience.
Across all funnel stages, use exclusion audiences actively. Exclude recent purchasers from conversion campaigns, exclude highly engaged followers from broad prospecting, and exclude users who have already completed a specific goal action from campaigns optimized toward that same goal. This discipline prevents ad fatigue, reduces wasted spend, and keeps your algorithm training on the signals that matter.
For brands looking to build and execute this kind of multi-layered XHS strategy with expert guidance, AllXHS's marketing services provide hands-on campaign architecture support tailored to your specific industry and growth stage.
---
Common Mistakes International Brands Make with XHS Targeting
Even well-resourced brands make predictable targeting errors when they first scale up XHS advertising. Being aware of these pitfalls can save significant budget and time.
Relying on demographics alone. Age and gender filters are a starting point, not a strategy. XHS's behavioral and interest data is far more predictive of purchase intent than basic demographic profiles, and campaigns that skip past these layers tend to generate high impression counts with poor conversion efficiency.
Applying Western interest stacks directly. Interest category behavior on XHS reflects the platform's distinctive content culture. Users who engage with skincare content on XHS are often researching at a depth and specificity that differs significantly from Instagram skincare browsing. Generic interest selections miss the nuanced sub-categories and behavioral combinations that define high-intent XHS audiences.
Neglecting exclusion logic. Brands that don't actively build exclusion audiences routinely waste 20–40% of their campaign budget delivering ads to users who have already converted or who are definitively outside the target profile. Building exclusion lists is not optional housekeeping — it's a structural component of efficient XHS campaign management.
Using poor-quality seed audiences for lookalike modeling. Uploading your full customer list as a lookalike seed produces a diluted model that tries to mirror both your best and your worst customers simultaneously. Segment first; seed from your highest-value cohort.
Treating XHS targeting in isolation from content quality. Unlike platforms where algorithmic delivery can compensate for mediocre creative, XHS's CES framework means that even perfect audience targeting will underdeliver if your ad content doesn't earn genuine engagement. Targeting precision and content authenticity must be developed in parallel.
Putting It All Together
Advanced XHS ad targeting is not about finding the single perfect audience setting — it's about building a coherent system where custom audiences, interest layers, behavioral signals, and lookalike modeling work together across your funnel. Each component reinforces the others: better custom audience quality produces better lookalike seeds; more precise interest layering generates higher-quality behavioral signals; stronger mid-funnel retargeting pools feed your bottom-funnel conversion campaigns.
For international brands, the additional challenge is doing all of this within a platform whose cultural dynamics, content norms, and user behavior patterns differ substantially from any Western equivalent. The targeting tools are powerful, but they require platform-specific judgment to deploy well — judgment that is built through experience, data, and genuine familiarity with how Xiaohongshu's community thinks and shops.
The brands that are building durable competitive advantages on XHS right now are the ones investing in that platform-specific expertise while the window of relatively low competition and underpriced ad inventory remains open. Getting your audience architecture right today positions you to scale more profitably as the platform's advertising ecosystem continues to mature.
---
Ready to build a more precise XHS audience strategy for your brand?
AllXHS is the #1 English-language resource hub for international brands marketing on Xiaohongshu, with 378+ data-driven industry reports, a 21-module training academy, and 25+ ready-to-use tools and templates covering 20+ verticals. Whether you're refining your targeting approach independently or want expert hands-on support, we can help.