故事 · 川喜田二郎与他的便签
Origin Story · Kawakita Jiro and His Sticky Notes
上世纪 60 年代,日本人类学家川喜田二郎常在喜马拉雅做田野调查,带回海量零散见闻,却无从下手分析。
他发明了一套方法:把每条观察写到一张小卡片上,全部摊开,不预设任何分类,
凭「这两张感觉是一类」的直觉,把亲近的卡片自下而上慢慢归拢、再给每组命名。
神奇的是,结构会自己浮现 —— 而且往往是你事先根本想不到的结构。
因为以他姓名缩写命名,这套方法被称为「KJ 法」,也就是质量工具里的亲和图(Affinity Diagram),
至今仍是整理用户之声、头脑风暴产出的标准手法。
In the 1960s, Japanese anthropologist Kawakita Jiro kept returning from Himalayan fieldwork with mountains of scattered observations and no way in.
He invented a method: write every observation on a small card, lay them all out, and — without imposing any prior categories —
cluster kindred cards bottom-up on the gut feeling "these two belong together," then name each cluster.
The remarkable thing is that structure surfaces on its own — usually one you could never have guessed up front.
Named after his initials, the method became known as the KJ method, the affinity diagram in the quality toolbox —
still the default move for digesting voice-of-customer data and brainstorming output today.
1 便签亲和:自下而上长出主题 Sticky-Note Affinity: Themes Grown Bottom-Up
拖动便签 / 自动亲和2 自下而上「涌现」 vs 自上而下「硬套」 Bottom-Up "Emergence" vs. Top-Down "Framing"
KJ 法:自下而上涌现
KJ method: bottom-up emergence
先有便签、后有分类。凭亲和度把卡片聚拢,让结构从数据里自己浮现 —— 常发现意料之外的真实模式。
Notes first, categories later. Cards cluster by affinity, structure surfaces from the data — and the patterns you find are often ones you never would have predicted.
反面:自上而下硬套
Anti-pattern: top-down framing
先定好「就分这 5 类」再往里塞卡片。框架挤压数据,塞不进去的真实信号被丢弃或扭曲。
Pick five buckets in advance, then jam the cards in. The frame crushes the data — real signals that don't fit get dropped or distorted.
3 现实里的 KJ 法 The KJ Method in the Real World
亲和图:六西格玛/精益的经典工具,把头脑风暴或客户之声的海量条目归拢成可处理的主题群。
Affinity diagram: a classic Six Sigma / Lean tool that condenses brainstorming output or voice-of-customer feeds into a manageable set of themes.
KJ 法:源自人类学田野研究,强调悬置预判、凭直觉亲和,让分类从材料中生长。
KJ method: born in anthropological fieldwork — suspend judgment, cluster by intuition, let the categories grow out of the material.
定性数据归纳:访谈、客诉、评论这类非结构化文本,KJ 法是把它变成洞察的主流路径。
Qualitative synthesis: for interviews, complaints, reviews and other unstructured text, the KJ method is the mainstream route from raw VOC to insight.
自下而上:不被既有框架束缚,最适合探索阶段 —— 你还不知道答案长什么样的时候。
Bottom-up: free of pre-built frameworks — ideal for the exploration phase, when you still don't know what the answer should look like.
一句话In One Line
面对未知,人本能地想先定框架再填内容 —— 但框架是过去的认知,会把数据里的新信号挤掉。
KJ 法反其道而行:悬置分类,先让卡片自由亲和,结构最后才浮现。
关键在那个「不预设」:正因为没先入为主,你才会聚出连自己都没想到的主题,
才能从一堆杂音里读出客户真正在意的东西。7 步只是脚手架,灵魂是那一句:让结构从数据里长出来,而不是把数据塞进结构。
Facing the unknown, the instinct is to pick the frame first and fill in the content later — but that frame is yesterday's understanding, and it squeezes out tomorrow's signal.
The KJ method runs the other way: suspend the categories, let the cards cluster freely, and let structure surface last.
The whole point is that no prior: precisely because nothing is preconceived, you discover themes you never would have invented yourself —
and you finally hear what the customer actually cares about under all the noise. The seven steps are just scaffolding; the soul is one sentence: let structure grow out of the data, don't cram the data into a structure.
常见误用Common Mistakes
先定好分类,再把便签往里塞(自上而下硬套)。悬置分类,让卡片凭亲和度自下而上聚拢。
Picking the buckets first, then stuffing notes into them (top-down framing). Suspend the categories — let the cards cluster bottom-up by affinity.
由领导一人主导分组,其他人只是旁观。全员一起动手亲和,分歧本身就是有价值的信号。
The boss drives the clustering alone while everyone else watches. Everyone clusters together — the disagreements are themselves valuable signal.
分完组就结束,跳过命名/图解/叙述。命名与叙述才把分组升华成可传播的洞察。
Stopping right after clustering, skipping naming / diagramming / narrating. Naming and narration are what turn clusters into insights you can actually share.