B站爆款公式与离群榜Bilibili Viral Formulas & Outliers
5 个可复用爆款公式 + 真实离群榜 + 三连/收藏率解读5 reusable viral formulas + real outlier board + interaction-rate reading
一句话定调
B站不是抖音。这里点赞、投币、收藏是三种不同的用户行为,分别对应"看完爽到了""被说服了""想留着用"。看懂这三条曲线,你就不用瞎猜哪条视频是真火还是假火——榜单会自己说话。本章把在题Top全拆开,提炼出5个可复用公式,并明确告诉你:剑桥图灵子的主战场只打其中两个,主动放弃第三个。
真实离群榜(B站在题Top,均为播放量)
数据口径说明:以下为B站站内"道家/道德经/庄子"相关视频的高播放样本,数字为播放量(B站公开口径),互动率为该视频的赞/投币/收藏占播放比。切勿与抖音横向比——抖音榜面是点赞数(非播放)、且为综合推荐排序而非按赞排序,两个平台的数字不在一个维度。下表名人/影视剪辑类已单独标注,不计入"可复制"判断。
| 视频 | UP主 | 播放量 | 关键互动率 | 时长 | 公式归类 |
|---|---|---|---|---|---|
| 庄子与尸骸·放屁清洁工 | (短内容) | 429万 | 赞9.01% | 0:41 | A 诡奇画面+极短 |
| 总有一篇背过的课文变成射向你的子弹 | 啊Sai在纽约 | 366万 | 赞3.71% / 投币0.71% | 6:14 | B 课文打脸反转 |
| 罗韦利爱智慧 | — | 346万 | 赞9.57% | — | (名人红利·不可复制) |
| 詹青云地铁读庄子 | — | 238万 | — | — | (名人剪辑·不可复制) |
| 躲寺庙三天顿悟 | 庄子HD | 176万 | — | 14:10 | (情绪vlog·重投入) |
| 道家的神性高级在哪 | 啊粥粥 | 144万 | — | 28:13 | C/长解说变体 |
| 道德经×GTA洛圣都 | 咪克菌 | 153万 | — | 2:30 | C 古典×现代混搭 |
| 道德经×优良家风 | 咪克菌 | 149万 | — | 2:30 | C 古典×现代混搭(系列化) |
| 曾仕强解密道德经全集 | — | 145万 | 收藏3.86% | ~17h | E 工具收藏型(陷阱) |
| 道德经全本诵读 | — | 141万 | 收藏4.02% | 53:24 | E 工具收藏型(陷阱) |
| 限制贫穷不是想象力 | 苏二郎 | 130万 | 赞5.13% | 0:31 | D 金句短切 |
五个可复用公式(每个带证据)
| 公式 | 内核 | 证据 | 时长甜区 | 优先级 |
|---|---|---|---|---|
| A 诡奇画面+极短 | 一个反常识/猎奇的视觉钩子,极短,看完即转 | 庄子与尸骸·放屁清洁工 429万、赞9.01%、0:41——全场最高赞率之一,时长却最短 | 0:30–0:50 | 主战场 |
| B 课文打脸反转 | 用人人背过的旧课文,推翻当年的标准答案,制造"原来是这意思"的认同 | 啊Sai在纽约 366万、投币0.71%(全场最高)——投币=B站用户掏真金认同你的观点 | 3:00–6:30 | 主战场 |
| C 古典×现代场景混搭 | 道德经套进GTA、家风等现代壳,系列化连发吃推荐 | 咪克菌 GTA 153万 + 家风 149万,月内连发成系列 | 2:00–3:00 | 可选/系列化时用 |
| D 金句短切 | 一句可截图、可转发的暴论式金句 + 极短画面 | 苏二郎"限制贫穷不是想象力"130万、赞5.13%、0:31 | 0:30–0:50 | 主战场补充 |
| E 工具收藏型(陷阱) | 全集/诵读/合辑,用户当工具囤,高收藏低赞 | 曾仕强全集 收藏3.86%、诵读全本 收藏4.02%,但赞率不显——收藏≠认可你,是"先存着" | 数十分钟–十几小时 | 不要碰 |
互动率怎么读:三类视频别混为一谈
B站的播放量会骗人,互动率不会。同样百万播放,底层是三种完全不同的用户关系:
| 类型 | 行为特征 | 代表 | 对UP主意味着 |
|---|---|---|---|
| 情绪钩子型 | 短 + 高赞,赞率冲到9% | 庄子尸骸0:41/赞9.01%、苏二郎0:31/赞5.13% | 看完即爽即转,涨粉快,但记忆浅 |
| 认同说服型 | 中长 + 高投币 | 啊Sai投币0.71% | 用户认你这个人、这套观点,这是建人格IP的核心信号 |
| 工具收藏型 | 超长 + 高收藏低赞 | 曾仕强全集收藏3.86%、诵读全本收藏4.02% | 用户在囤资料,不在认你;粉丝黏性最弱 |
| 名人红利型 | 靠人不靠题 | 罗韦利346万、詹青云238万 | 流量来自名人本身,素人不可复制,别对标 |
关键判断:收藏率高不等于内容好,常常恰恰相反——它说明用户把你当工具书,而不是当一个值得追的人。对"建人格IP"这件事,收藏率是个噪声指标,甚至是反向指标。
剑桥图灵子的主战场设定
你不是要做"道德经全本诵读"的资料库,那条路赞率低、人格不可见,流量再大也沉淀不成你这个人。你的定位是理性研究者讲道家——所以你的武器是认同(投币)和转发(高赞短切),不是囤积(收藏)。
| 维度 | 设定 |
|---|---|
| 打哪两个公式 | A(诡奇画面+极短)× B(课文打脸反转)为主,D(金句短切)为补 |
| 主力时长 | 0:30–1:30 |
| 结构模板 | 一个画面/打脸钩子(前3秒)+ 一个考据反转(中段)+ 一句可截图金句(结尾) |
| 冲的指标 | 赞率 + 投币率(投币=被说服认同,是人格IP的硬通货) |
| 放弃的指标 | 收藏率(那是工具型,不是你要的关系) |
| 不碰的形态 | E 全集/诵读/超长合辑;名人剪辑(你无名可借) |
公式落到一条片子上的样子:开头抛一个反常识画面或一句"你背过的某课文,标准答案是错的"(A/B 钩子),中段用一条考据把当年的解读反转(图灵子的剑桥/Princeton 理性底色在这里发力),结尾甩一句能被人截图发朋友圈的暴论(D)。这就是你赞率和投币率同时拉起来的结构。
一句话收尾
把"火没火"从播放量改成看互动率,你立刻能分清谁在演、谁真被你说服。研究者的优势就是不靠玄、不靠名人、不靠情绪vlog硬熬时长——你靠的是一刀切进去的考据反转。那些深夜刷"躲寺庙三天顿悟"的人,多半不是真要顿悟,是有点撑不住想找个地方喘口气;能接住这种情绪的,内容做不到,产品(比如安心舍)能补一点——但那是另一回事,这一章先把片子打爆。
The One-Line Thesis
Bilibili is not Douyin. Here, likes, coins, and favorites are three distinct user actions—they map to "that hit," "you convinced me," and "I want to keep this." Read those three curves and you never have to guess whether a video is genuinely hot or just inflated—the board tells you itself. This chapter dissects the on-topic Top performers, extracts 5 reusable formulas, and tells you plainly: Cambridge Tulingzi fights on only two of them and deliberately abandons a third.
The Real Outlier Board (Bilibili on-topic Top, all figures are views)
Data caveat: the rows below are high-view samples of Bilibili "Daoism / Tao Te Ching / Zhuangzi" videos. Figures are view counts (Bilibili's public metric); interaction rates are likes/coins/favorites as a share of views for that video. Do not compare these to Douyin—Douyin's surface number is likes (not views) and its board is recommendation-ranked, not like-ranked; the two platforms' numbers are not the same dimension. Celebrity / film-edit rows are tagged separately and excluded from any "reproducible" judgment.
| Video | Creator | Views | Key interaction rate | Length | Formula |
|---|---|---|---|---|---|
| Zhuangzi & the corpse · farting janitor | (short) | 4.29M | Likes 9.01% | 0:41 | A Bizarre visual + ultra-short |
| A memorized school text becomes a bullet aimed at you | A-Sai in NY | 3.66M | Likes 3.71% / Coins 0.71% | 6:14 | B Schooltext reversal |
| Rovelli loves wisdom | — | 3.46M | Likes 9.57% | — | (celebrity halo · not reproducible) |
| Zhan Qingyun reads Zhuangzi on the subway | — | 2.38M | — | — | (celebrity edit · not reproducible) |
| Hid in a temple 3 days, enlightened | Zhuangzi HD | 1.76M | — | 14:10 | (emotional vlog · high effort) |
| Where the Daoist divine is sophisticated | A-Zhouzhou | 1.44M | — | 28:13 | C/long-explainer variant |
| Tao Te Ching × GTA Los Santos | Mike-jun | 1.53M | — | 2:30 | C Classic × modern mashup |
| Tao Te Ching × good family ethos | Mike-jun | 1.49M | — | 2:30 | C Classic × modern (serialized) |
| Zeng Shiqiang decodes the Tao Te Ching, full set | — | 1.45M | Favorites 3.86% | ~17h | E Tool-hoarding (trap) |
| Full recital of the Tao Te Ching | — | 1.41M | Favorites 4.02% | 53:24 | E Tool-hoarding (trap) |
| What limits you isn't imagination, it's poverty | Su Erlang | 1.30M | Likes 5.13% | 0:31 | D Punchline short-cut |
Five Reusable Formulas (each with evidence)
| Formula | Core | Evidence | Length sweet spot | Priority |
|---|---|---|---|---|
| A Bizarre visual + ultra-short | One counterintuitive/freakish visual hook, ultra-short, share-on-finish | Zhuangzi & corpse · farting janitor 4.29M, Likes 9.01%, 0:41—one of the highest like rates on the board, yet the shortest | 0:30–0:50 | Main front |
| B Schooltext reversal | Take a text everyone memorized in school, overturn the official answer, manufacture "oh THAT'S what it meant" agreement | A-Sai in NY 3.66M, Coins 0.71% (highest on board)—coins = Bilibili users spending real currency to endorse your take | 3:00–6:30 | Main front |
| C Classic × modern mashup | Wrap the Tao Te Ching in GTA, family ethos, etc.; serialize to ride recommendation | Mike-jun GTA 1.53M + ethos 1.49M, dropped as a series within a month | 2:00–3:00 | Optional / when serializing |
| D Punchline short-cut | One screenshot-able, share-able provocative line + an ultra-short visual | Su Erlang "what limits you isn't imagination, it's poverty" 1.30M, Likes 5.13%, 0:31 | 0:30–0:50 | Main-front supplement |
| E Tool-hoarding (trap) | Full sets / recitals / compilations users stash like a tool—high favorites, low likes | Zeng Shiqiang full set Favorites 3.86%, full recital Favorites 4.02%, but like rate unremarkable—favorites ≠ approval of YOU, just "I'll keep this" | tens of minutes–10+ hours | Don't touch |
How to Read Interaction Rate: Don't Conflate the Three Types
Bilibili view counts lie; interaction rates don't. Behind the same million views sit three completely different user relationships:
| Type | Behavior signature | Exemplar | What it means for the creator |
|---|---|---|---|
| Emotional-hook | Short + high likes, like rate spiking to 9% | Zhuangzi corpse 0:41/Likes 9.01%, Su Erlang 0:31/Likes 5.13% | Hit-and-share, fast follower growth, but shallow memory |
| Persuasion-endorsement | Mid-length + high coins | A-Sai Coins 0.71% | Users endorse YOU and your thesis—this is the core signal for building a persona IP |
| Tool-hoarding | Ultra-long + high favorites, low likes | Zeng Shiqiang full set Favorites 3.86%, full recital Favorites 4.02% | Users are stockpiling material, not endorsing you; weakest loyalty |
| Celebrity halo | Powered by the person, not the topic | Rovelli 3.46M, Zhan Qingyun 2.38M | Traffic comes from the celebrity—unknowns can't reproduce it, don't benchmark against it |
Key call: a high favorite rate does NOT mean good content—often the opposite. It means users treat you as a reference book, not a person worth following. For building a persona IP, favorite rate is a noise metric, even an inverse one.
Cambridge Tulingzi's Battlefield Setting
You are not building a "full Tao Te Ching recital" archive—that path has low like rates, an invisible persona, and however much traffic it draws it never compounds into YOU. Your positioning is a rational researcher explaining Daoism—so your weapons are endorsement (coins) and forwarding (high-like short cuts), not hoarding (favorites).
| Dimension | Setting |
|---|---|
| Which two formulas | A (bizarre visual + ultra-short) × B (schooltext reversal) as the spine, D (punchline short-cut) as supplement |
| Primary length | 0:30–1:30 |
| Structural template | One visual/reversal hook (first 3s) + one textual-research reversal (middle) + one screenshot-able punchline (end) |
| Metrics to push | Like rate + coin rate (coins = persuaded endorsement, the hard currency of a persona IP) |
| Metric to abandon | Favorite rate (that's the tool relationship, not the one you want) |
| Forms to avoid | E full sets/recitals/ultra-long compilations; celebrity edits (you have no fame to borrow) |
What the formula looks like on a single clip: open with a counterintuitive visual or a line like "that school text you memorized—the official answer was wrong" (A/B hook), use one piece of textual research to overturn the old reading in the middle (where Tulingzi's Cambridge/Princeton rational baseline does its work), and close with a provocative line people will screenshot and repost (D). That is the structure that lifts your like rate and coin rate at the same time.
One-Line Close
Swap "is it hot" from view count to interaction rate and you instantly see who's performing versus who's actually been convinced by you. The researcher's edge is precisely this: no mysticism, no celebrity, no grinding out length with an emotional vlog—you win on a single clean cut of textual reversal. The people scrolling "hid in a temple 3 days, enlightened" at 2am mostly aren't seeking enlightenment; they're a little overwhelmed and looking for somewhere to breathe. Content can't fully catch that emotion—a product (like AnXinShe) can patch a bit of it—but that's another matter. This chapter is about making the clip land first.