Aim for "being yanked to a stop in the 0.5 seconds a finger swipes past," butnever sacrifice consistency between the title and the body, and never touch platform-banned words. It's not giving you one title, it's giving you a whole table of dishes—multiple methods, multiple uses, filterable, A/B-testable.
Core principle: A Xiaohongshu title does two things at once—
- Grab people when they swipe past in the feed(emotion + audience + gap);
- Get found in the search box(keyword hits). A title that only does the first thing lives three days; a title that does both eats half a year of long-tail traffic.
Platform fact calibration (remember these first, don't write Xiaohongshu with WeChat Official Account instincts)
Calibrate with these facts when generating. Below, what's marked [Platform rule] is the platform's hard mechanism, and what's marked [Experience] is a general observation on the creator side, not officially published statistics—use it as a rule of thumb, don't cite it as data:
- [Platform rule]Hard cap of 20 characters(emoji also count as character positions).Optimal range 13–18 characters. The body and cover will also expose keywords a second time, so don't be greedy with title length. Anything over 20 characters gets truncated by the platform, and the information at the end of the sentence is simply lost.
- [Platform rule]Xiaohongshu is a search-based community. Besides feed distribution, notes also continuously draw traffic from search and "post-search recommendations." If the title has no words users would type into the search box, you're giving up the long tail.The weight of this point is far higher than on WeChat Official Accounts.
- [Experience]One nailed search term beats three stacked keywords. With a 20-character budget, rather than cramming in category words, it's better to write one main search term completely ("oily-skin foundation" beats "beauty good stuff foundation recommendation"). Keyword stacking also gets judged as marketing behavior.
- [Experience]Strong first person, strong colloquialism. "I" "sisters" "who gets it" "help" "truly thanks"—a bookish tone is a turn-off signal on Xiaohongshu.
- [Experience]emoji are standard equipment, not something to hold back on(completely the opposite of WeChat Official Accounts). 1–2, placed at the start to catch the eye or at the end to set the mood.Don't exceed 3; stuffed with emoji, it gets identified as a marketing account.
- [Experience]Put audience tags up front: "students" "introverts" "petite girls" "new moms" "oily skin" "commuters" "budget-conscious." Readers confirm "that's me" within half a second.
- [Experience]High-save structures are favored: numbered lists + "recommend saving" "bookmark it" "just copy it." Saving is one of Xiaohongshu's strongest positive feedback signals.
- [Experience]Negative-angle topics perform extremely strongly:"Avoid these" "Don't buy" "Talk you out of it" "Wish I'd known sooner" "Never touch these few". Avoidance-type content has long had a higher click-through rate than recommendation-type content.
- [Experience]Symbol conventions：
｜(vertical bar) is most commonly used for sectioning (学生党｜宿舍好物), followed by exclamation marks, question marks, and parenthetical supplements.Colons are less commonly used than on WeChat Official Accounts。
should not be misread: adding "help" and emoji won't automatically make it go viral. Note performance is jointly determined by the cover, account tags, posting time slot, and body copy retention; the title only optimizes the two segments of "getting clicks" and "being searchable".
15 Xiaohongshu title methods (with examples + formulas + red lines)
When generatingcover at least 6 different methods, don't make 10 titles follow one formula. Each title should prioritize one main method, with at most one auxiliary method stacked on.
1. Audience label front-loading
Put the self-referential term of the target audience at the very front, so the right people claim it in half a second, and the platform pushes you to the right tag pool.
- Example:
学生党｜300块搞定一整个宿舍✨/i人友好的搭子型运动💃 - Formula:
[人群词] ｜ [他们的具体处境] + [结果] - Red line: The audience term must be the group the body copy truly serves; you can't force-fit it for traffic (mismatching the audience lowers your account tag precision, which is worse than having no tags).
2. Search term positioning
Use words users would actually type into the search box as the title backbone—usually a combination of "category + scenario + modifier", not pretty literary phrasing.
- Example:
上海周末citywalk路线｜不累版🚶/油皮夏天不脱妆的粉底液 - Formula:
[核心搜索词] + [限定条件] + [差异化后缀] - Red line: Don't force in hot words unrelated to the content. Stuffing irrelevant keywords is an action the platform explicitly cracks down on and will get you throttled.
3. Numbered list + save instruction
Give a countable quantity + an explicit save-action word to directly activate saving behavior.
- Example:
7个提升效率的神器｜建议码住📌/毕业论文5个救命工具，别谢我 - Formula:
[数字] + [品类/方法] + [建议收藏/码住/直接抄] - Red line: The number must match the number of items in the body copy. Writing 7 when the body only has 4 = baiting saves, and you'll get bad reviews in the comments.
4. Avoidance/talk-you-out-of-it (negative)
Saying "what not to do" gets more clicks than saying "what to do". This is the most reliable category on Xiaohongshu.
- Example:
这3款别买了，真的踩雷😭/装修千万别学我这样铺地板 - Formula:
[品类/行为] + 别买/别踩/千万别 + [代价]or我踩过的[N]个坑 - Red line:Named negative evaluations must be your own real experience, and limited to subjective feelings, you cannot fabricate product defects or insinuate brand fraud (involves legal risk). If you haven't really used it, write about generic category pitfalls without naming names.
5. Regret style / wish I'd known sooner
Use the regretful emotion of "if only I'd known earlier" to imply this is scarce information.
- Example:
后悔没早点知道这个方法😭/这个功能我居然现在才发现 - Formula:
后悔/居然/竟然 + 没早知道 + [具体的事] - Red line: The content must truly have information gain. Common knowledge everyone knows paired with "wish I'd known sooner" = clickbait.
6. Emotional exclamation (help style)
Write the reader's first reaction upon seeing the content directly as the title.
- Example:
救命这也太好用了吧😭/谁懂啊，我终于把它搞定了 - Formula:
救命/天呐/谁懂啊/真的会谢 + [对象] + [强反应] - Red line:Used alone it will be very empty; it must be paired with a specific object or result. Pure emotional words have no search value and need to be stacked with method 2.
7. Personal test comparison (before-and-after contrast)
Use visible change to build credibility; this is the hardest hook for recommendation-type content.
- Example:
坚持3个月，皮肤肉眼可见变好/同一间房，我只换了这2样 - Formula:
[时长/动作] 后，[可观测的变化]or[变量A] vs [变量B] - Red line:If you haven't actually done it, you must not write "personally tested," "stuck with it for X months," or "actually tested". For effect-type descriptions, write your own feelings, not efficacy claims (see compliance red lines).
8. Hand-holding tutorial / beginner-oriented
Promise zero threshold and replicability, directly targeting the "I want to be able to do that too" crowd.
- Examples:
新手也能学会的剪辑｜保姆级🎬/零基础做PPT，3步就够 - Formula:
零基础/新手/小白 + [目标] + 保姆级/N步搞定 - Red line: The body text must truly be step-by-step, and the number of steps must match the title. If you can't provide actionable steps, don't use "hand-holding."
9. Counterintuitive correction
Overturn a default practice and create the unease of "Have I been doing this wrong all along?"
- Examples:
原来这一步一直都做反了😳/别再这样洗了，越洗越糟 - Formula:
原来/其实 + [常见做法] + 是错的or别再 [常见动作] 了 - Red line: What is reversed must truly be a wrong practice, and there must be support in the body text. Don't reverse just for the sake of reversing.
10. Suspense gap
Only say "something happened" or "there's something really good," without saying what it is.
- Examples:
打工人必备的这个App，我藏不住了/我妈的这招，全家都服了 - Formula:
[人群] + [强评价] + 但不说明对象 - Red line:Use pure suspense cautiously on Xiaohongshu—hiding all keywords will also cost you search traffic. A suspense title must retain at least 1 searchable word.
11. Scene concretization
Write a slice of life into the title, and readers will automatically insert themselves.
- Examples:
加班到10点还能吃口热的🍜/出租屋厨房，也能做出这一桌 - Formula:
[具体时间/地点/困境] + [依然做到了什么] - Red line: The scene is real life; do not fabricate persona experiences.
12. Affordable alternative / money-saving / cost-effectiveness
A clear monetary benefit point, one of the most frequent consumer motivations on Xiaohongshu.
- Examples:
平价平替｜30块用出300块的效果/预算500，我全买齐了 - Formula:
[预算数字] + 搞定/买齐 + [目标]or[贵价对象] 的平替 - Red line:Prices must be real and note the time/channel basis; "affordable alternative" is a subjective similarity judgment, and you cannot claim it has the "same effect" or "same ingredients" as a big brand (easily touches false advertising).
13. Time promise
Use a specific time scale to make the result seem achievable.
- Examples:
7天，我把卧室彻底改造了/每天10分钟，一个月的变化 - Formula:
[时长] + [投入] + [结果] - Red line: The duration and result must come from real experience, and do not promise others will reproduce the same effect.
14. Asking for help / question interaction
Write the title as a sincere question, directly trading for comments.
- Examples:
这样穿显高吗？在线等👀/姐妹们，这两个选哪个？ - Formula:
[具体选择困境] + 在线等/求推荐/怎么选 - Red line: The body text must really be able to sustain discussion; don't ask and then answer it yourself. This type of titlehas high interaction but low search value, so it is not suitable as the account's mainstay.
15. Tie-in with occasions / seasons / trending topics
Attach it to a time occasion everyone is searching for right now.
- Examples:
换季必备清单｜这5件先囤/国庆3天，穷游也能玩爽 - Formula:
[节点词] + [人群/预算] + [具体产出] - Variant (social proof):
第一批 [做了某事] 的人，已经 [结果了]—use "others have already started" to create herd pressure; especially useful for occasion-based content. - Red line: Trending topics must be genuinely related to the content; never piggyback on social events, disasters, or public safety topics. "The first batch" must be a real existing group phenomenon, not an invented crowd.
Additional: Use-with-caution type (strong tension, but also most likely to violate rules or overdraft trust)
- Absolute elevation：
最好用的XX——Xiaohongshu reviews absolute wording more strictly than WeChat public accounts, and "most," "No. 1," "top-tier," and "sales champion" very easily trigger throttling, so do not use them unless necessary. - Anxiety manufacturing：
不看这条你会后悔一辈子—excessive fearmongering will be judged as clickbait. - Data bragging：
月入5万的副业— Income-related promises are key review targets and need to be carefully phrased as personal experience.
10 Xiaohongshu enhancers
After generating candidates, "spice up" them one by one; these can be stacked:
- Front-load audience words: insert "students", "introverts", "petite girls", "new moms" at the start of the sentence.
- Vertical bar partitioning：
场景｜结果, two pieces of information at a glance, the most typical layout feel on Xiaohongshu. - Emoji sets the tone: 1–2, placed at the end of the sentence to set the emotion (😭 breakdown / ✨ good find / 📌 save / 🔥 trending / 👀 curiosity).Don't exceed 3。
- Colloquial word swaps: "improve efficiency" → "no more overtime"; "cost-effective" → "so cheap it's absurd".
- Search term fill-in: check whether the title has a noun users would actually search for; if not, replace an adjective with one.
- Add a save hook at the end: add "recommend saving", "save this", "just copy it" at the end of the sentence (only when the content is truly reusable).
- Replace vague words with specific numbers: "many" → "7"; "cheap" → "30 yuan".
- First-person-ize: turn declarative sentences into "my" experience—"doing this is better" → "after I did this…".
- Put original quotes in quotation marks: move the complaint/exclamation users would blurt out verbatim into the title and put it in quotation marks—
"这也太难删了吧！". Quotation marks carry the on-the-spot feel of "someone is speaking," which grabs attention more than paraphrasing.The original quote must be something actually heard or that appeared in the body text; do not fabricate user testimonials. - Identity contrast: let "the person who shouldn't say this" say it—
我妈看完直接下单了/理发师朋友劝我别买. The contrasting identity must genuinely exist in the body text; do not invent a persona.
Enhancers do not stack facts: spicing up changes the expression, not adding things to the title that aren't in the body text.
Workflow
Step 0: Determine the input type
- A. Create a title for a topic(the user provides a topic/draft/product) → go to Step 1.
- B. Optimize an existing title(the user pastes an existing title) → skip to "Rewrite an existing title" at the end of this document, and after completing the diagnosis return to Steps 2–4.
Step 1: Build a content brief
After reading the content, first extract these (mark any missing ones,do not fabricate）：
- Target audience: What do they call themselves? (This determines the audience label words.)
- Search intent: What would this audience type into the search box? List 3–5 candidate search terms.
- Core subject: What exactly is the product/place/method/tool?
- Deliverable result: What can readers gain/avoid, and to what extent can the body text prove it?
- Evidence assets: real numbers, prices, durations, before-and-after comparisons, personal experience.
- Emotional tone: breakdown / surprise / heartache / catharsis / healing / curiosity.
- Reason to save: Why is this worth saving? (If it isn't worth saving, don't use save-oriented methods.)
- Timeliness level: real-time hotspots / seasonal nodes / evergreen long-tail.Evergreen content should max out search term weighting(it relies on half a year's worth of search traffic); only seasonal content is worth betting on seasonal keywords, and it becomes ineffective once the season passes.
- Factual boundaries: body textdoes notproven conclusions, effects, prices, rankings — not allowed in the title.
- Compliance risk points: whether the content involves medical claims, efficacy, investment returns, or absolute statements.
If the content includesreal prices, real durations, verbatim user complaints, before-and-after comparison photo descriptions, definitely pull them out first — they are the best raw material for Xiaohongshu titles, more valuable than any adjective.
If information is insufficient, put out a conservative version first and explain what is missing that limits the title's strength —do not rely on fabrication to make up for strength。
Step 2: Route methods by content type
| Content type | Priority methods | Not preferred |
|---|---|---|
| Product recommendations, reviews | Personal testing comparisons, affordable alternatives to save money, audience tags, warning people off | Pure suspense, asking for help |
| Practical tutorials, tools | Numbered lists + save-worthy, foolproof guide, search term positioning | Emotional exclamations alone |
| Warning people off and deterring purchases, debunking rumors | Warning people off, counterintuitive corrections, regret narratives | Absolute exaggeration |
| Store visits, travel, routes | Search term positioning (place names must be included), seasonal nodes, concrete scenarios | Suspense gaps |
| Outfit styling, beauty | Audience tags (body type/skin type), personal testing comparisons, asking for interaction | Data showing off |
| Personal growth, Vlogs | Time commitments, concrete scenarios, regret narratives, first-person contrast | Numbered lists |
| Emotional resonance, complaints | Emotional exclamations, concrete scenarios, asking for interaction | Foolproof guide |
| Knowledge popularization, answering questions | Counterintuitive corrections, search term positioning, numbered lists + save-worthy | Emotional exclamations alone, suspense |
| Home, maternal and baby, renovation | Warning people off, personal testing comparisons, affordable alternatives to save money, concrete scenarios | Absolute exaggeration |
| Commercial posts, brand collaborations | Audience tags, concrete scenarios, search term positioning (brand/category terms must be included) | Warning people off, absolutes, efficacy assertions, data showing off |
| Weak material, no hard content | Search term positioning, concrete scenarios, asking for interaction | Personal testing, numbers, affordable alternatives |
Step 3: Batch generate with multiple methods
Default 12 candidates, at least 6 methods; if the user says "more," expand to 20. Try to cover: search-oriented, list-oriented, warning-oriented, experience-oriented, emotion-oriented, and scenario-oriented, 1–3 each.
Only use the corresponding method when supported by the body text: if you have not actually used it, do not write "personal testing"; if there are not N items, do not write numbers; if there is no price evidence, do not write a specific amount.
Self-check character count item by item, cut anything over 20 characters down to within 18 characters before scoring.
Step 4: Scoring, deduplication, compliance check
Score each dimension according to the scoring table below, then: delete the ones that exceed the character count, delete the ones the body cannot deliver on, delete homogeneous titles that only change one or two words, delete the ones that touch prohibited words, and flag the medium- and high-risk ones.
Step 5: Give results according to the output format
Scoring rules (100-point scale, score each dimension, don't report a total score based on feeling)
| Dimension | Points | Judgment |
|---|---|---|
| Content deliverability | 20 | Can the body deliver on the title's promise? |
| Search term coverage | 15 | Does it contain at least 1 noun users would actually search? Is it a mainstream search phrasing rather than an obscure expression? |
| Audience targeting | 15 | Can the target audience confirm within half a second that "this is about me"? |
| Emotion and hook | 12 | Is there an emotional point or gap that makes people stop scrolling? |
| Save motivation | 10 | Do readers have a reason to save it (checklist/tutorial/pitfall avoidance/reusable)? |
| Concreteness | 10 | Does it have a clear object/number/scenario rather than vague description? |
| Character count and rhythm | 8 | ≤20 characters, key point up front, understood in one pass, no filler words? |
| Colloquial feel and platform feel | 5 | Does it sound like something a Xiaohongshu user would say, not a WeChat official account tone? |
| Uniqueness | 5 | Is it different from the phrasing that fills the screen on the same topic? |
Mandatory deductions / direct elimination：
- Over 20 characters → Direct elimination(or first cut to compliant and then score)
- The title's core promise cannot be delivered by the body → Direct elimination
- Fabricated numbers, prices, personally tested experiences, item counts, rankings → Direct elimination
- Touches any item in the "platform violation red lines" below → Direct elimination
- Presenting speculation/isolated cases as普遍 facts that have already happened → −15
- Not a single searchable noun in the title → −15
- Exaggerating causation or expanding "it worked well for me" into "suitable for everyone" ("everyone should buy it," "buy blind with no pitfalls") → −12
- The audience term does not match the actual target served by the body (forced labeling) → −12
- Presenting subjective feelings as objective effects → −12
- Unsupported "most/first/only/permanent" → −12
- Purely emotional stacking, no specific object → −10
- Numbers inconsistent with the number of items in the body → −10
- Excessive suspense, hiding the object and keywords entirely (no one can search it, and no one knows what it is saying) → −8
- More than 3 emoji or emoji unrelated to the meaning → −5
- Written tone, strong WeChat official account flavor → −5
Risk level：
- Low: facts are all in the body, no efficacy/absolutist expressions, audience terms accurate.
- Medium: uses strong emotional words, names a specific brand, contains price or duration promises — the words to verify must be marked.
- High: Involves efficacy/medical/benefit promises, absolute wording, unverified negative accusations, or claims that only hold up through misunderstanding. High-risk items must not be made the overall top pick, and the output must clearly state "why it is high-risk + a safe revision."
Platform violation red lines (Xiaohongshu is much stricter than WeChat Official Accounts; be sure to check them one by one)
The following are on Xiaohongshua major disaster area for reduced reach/violations, with consequences more serious than "clickbait":
- Medical and efficacy: Prohibited: efficacy/benefit assertions such as "treatment," "acne removal," "spot removal," "whitening," "slimming," "detox," "anti-inflammatory," "barrier repair," and "hair growth." → Safe revision: Write subjective feelings and observable phenomena—"how it felt after I used it," "no breakouts after applying."
- Absolute wording: Prohibited: "best," "No. 1," "top-tier," "best-selling champion," "100 percent," "permanent," and "national-level." → Safe revision: Limit it to a personal perspective—"the handiest one I have used" (still requires caution) or directly replace it with a specific point of difference.
- Income and investment promises: Prohibited: promise-style statements such as "guaranteed profit," "guaranteed pass," and "earn X ten thousand per month." → Safe revision: Write it as a personal experience and do not promise it can be replicated.
- False and exaggerated: Do not fabricate before-and-after comparisons, invent personal testing, misstate prices, or falsify item counts.
- Naming and disparaging: Do not make unverified negative accusations against specific brands (such as claiming counterfeiting, toxicity, or causing disease). Subjective experiences must be clearly labeled as personal feelings.
- Inducement and traffic diversion: Titles must not contain traffic-diversion instructions such as "add on WeChat," "DM to receive," "join the group," or "follow to get materials."
- Sensitive and borderline content: Do not do any mildly sexual borderline content, incitement of gender antagonism, regional attacks, or piggybacking on social disasters.
- Minors-related: Do not make titles involving evaluations of minors' appearance or body shape.
Judgment mantra: If this sentence were screenshotted and sent to the platform's reporting entry, could it hold up? If not, revise it.
Output format
① Content title brief
46 lines: target audience and how they refer to themselves, candidate search terms (35), core subject, biggest viral point, available evidence, factual and compliance boundaries.
② Candidate title matrix
| # | Candidate title | Character count | Method | Hook/search term | Score | Risk |
|---|
- Use the standard names of the 15 methods above for method tags; write combinations as "primary method + secondary method."
- The "Hook/search term" column must indicate which search term this title is targeting.
- Medium and high risk must state the reason.
③ Top 5 recommendations (assigned roles by use, not mechanically ranked by score)
- Overall top pick: Best balance of clicks, search, and credibility
- Search version: Most complete keywords, captures long-tail traffic, suitable for evergreen notes
- Viral version: Strongest emotion/contrast/pitfall-avoidance, suitable for pushing first-release traffic
- Save-worthy version: Strongest list feel, save-rate oriented
- Steady version: Zero compliance risk, suitable for brand collaborations, sponsored jobs, and professional accounts
Explain in one sentence why each is suitable, and point out the main scoring points and actual deductions.If 5 come from the same method, replace them with candidates that score slightly lower but use different methods.
④ A/B testing suggestions
Give 2~3 groups, each groupchange only one variable, and write the hypothesis clearly:
- with emoji vs without emoji
- audience word first vs result first
- numbered list vs single-point story
- positive recommendation vs negative warning
- statement vs help-seeking question
Each group must clearly state which metric to look at, otherwise even after testing you cannot read out a conclusion:
- Emotion/emoji/contrast type variables → look atclick-through rate(first-day impressions to clicks)
- Search term/audience term type variables → look atsearch entry share and long-tail impressions after 7 days, first-day data does not count
- List/save hook type variables → look atsave rate For two notes on the same account, the posting time slot and cover style should be as consistent as possible, otherwise there is more than one variable.
⑤ Supporting tips (optional, mention in one line)
Are the core search terms in the title also included in the first sentence of the body and the topic tags? Only when all three places are consistent can search weight build up.
Rewrite existing titles
When the user pastes the original title, proceed in this order:
- Diagnosis(check/cross item by item): whether the character count exceeds the limit / whether there is a searchable noun / whether there is audience targeting / whether there is emotion or a gap / whether there is a reason to save / whether it has an official-account tone / whether it touches prohibited words.
- Preserve assets: clearly point out the parts of the original title that are already effective (a good search term, a real number, an accurate audience word), and when rewritingdo not change them。
- Produce at least 12 rewritten versions, covering more than 4 methods, among which:
- 3 steady optimization(only fix character count, colloquialness, and search terms; style unchanged)
- 6 rewrite with a different method(switch to a different title method)
- 3 aggressive tests(emotion/contrast maxed out, risks self-labeled)
- Clearly labelwhich category each one belongs to, and explain "what variable was changed."
Do not treat "more exaggerated" as the only optimization method. Many title problems on Xiaohongshu are not that they are insufficiently stimulating, but that not a single word can be searched.
Red lines (general principles)
- Do not fabricate numbers, prices, item counts, personally tested experiences, rankings, or exclusivity.You can leave a gap, but you cannot lie.
- Do not write efficacy assertions, absolute terms, or income promises—on Xiaohongshu this is a throttling-level issue, not just a credibility issue.
- Do not force audience labels or pile on irrelevant keywords; this pollutes the account labels and is more harmful long term than throttling a single note.
- Clickbait that does not match what arrives will lose followers, lose saves, and lower completion rate: no matter how big the gap, once clicked it must deliver.
- "Viral hit" is not a promised result; this skill only optimizes the two controllable parts: "being clicked" and "being found in search."
Relay (where to go after the title is finalized)
The title is not the endpoint. After finalizing, proactively prompt the user for the next step and hand over the deliverables together:
| Next step | Hand off to whom | What to hand off |
|---|---|---|
| Write the body | space-xhs-writer |
Finalized title + 3–5 search terms you hit (the first sentence of the body and the tags must echo them) |
| Make the cover | xhs-html |
Finalized titleThen compress to ≤12 characterscover copy + emotional tone |
| No topic, don't know what to write | space-xhs-positioning(no positioning) or space-xhs-hotspot(positioning but no topic) |
niche keyword |
| Posted but nobody clicks | space-xhs-note-analytics |
Impressions + click-through rate—low CTR is the title/cover's fault, low impressions mean the problem is upstream, don't randomly change the title |
Hard consistency requirement: the core search term must appear in the title, the first sentence of the body, and the topic tags all three places at the same time. If it appears only in the title, search weight will be diluted.