The Five-Second Test
The first screen's message is decided in five seconds. Get someone who's never touched the product, let them look for five seconds, then ask three things: what is this? what do I get? what am I supposed to do? If they can't answer, the first screen hasn't said it.
What you'll run into
- 用户第一眼不知道这是干什么的，要靠小字慢慢猜
- 首屏堆满了功能入口，没有一句在说「这是什么」
- 自己看久了觉得哪都懂，换个陌生人就懵
What the five seconds are testing
The five-second test isn't measuring whether it looks good — it's measuring message delivery: in those five seconds, does the user build the mental model of "what this product is, and why it matters to me"?
- Only the first screen. Five seconds only covers the first screen. Whatever the later screens have to say, a user who never got the first one may never see them.
- It's testing "stay or leave". What it verifies is whether the user is willing to keep going — not whether they'll pay, not whether they'll register.
- The usual cause of "can't answer" is too much information. A first screen trying to say ten things has said none. Stacking features isn't delivering value. See One job per screen.
How to run it
The flow is simple; the hard part is picking the right people and asking the right questions:
对象没接触过产品的人同事、用户都不行
时长只看五秒看久了就不是第一印象
提问这是什么 / 我能得到什么 / 我该做什么三问定生死
样本三到五个人够暴露问题，别等「量」
- 找没接触过的人。同事、已注册用户都不行——他们已经「知道这是什么」了，测不出信息传达。样本脏了，结论就脏了。
- 严格五秒。看一眼就收走。看久了就变成「仔细阅读测试」，测的是阅读能力，不是第一印象。
- 问的是开放问题。别给选项，别提示。他说出「这是个做笔记的软件」，说明讲清了；他说「呃，一个网站」，说明没有。
- 三五个人的答案就够判断方向。五秒测试是暴露问题的定性检查，不是给结论的定量依据——别拿它当统计调查。定量的事交给真实数据。
If they can't answer
Each unanswerable question has a different fix:
- 答不出「这是什么」→ 首屏缺一句能概括产品的表达。加一句白话说清产品的一句话定位，而不是摆更多功能入口。参见一句话产品。
- 答得出「是什么」、答不出「我能得到什么」→ 价值被功能列表盖住了。把「对我有什么用」从「有什么功能」里挑出来，放到看得见的位置。
- 前两个都对、答不出「我该做什么」→ 缺一个明确的下一步。首屏要有且只有一个主入口，别让用户自己找。参见首次体验设计。
- 修完再测一遍。五秒测试的好处就是便宜：改一版，再找三个人，十分钟出结果。循环到三问都能答上为止。