Aug 28, 2026 · Building products
The mistakes that quietly kill new products悄悄殺死新產品的那些錯誤
From Paul Graham and YC: the subtle early stage mistakes that slowly kill new products, plus a short checklist to keep you honest.從 Paul Graham 和 YC 的建議，看那些在早期默默累積，最後把新產品拖垮的錯誤，外加一份簡短自我檢查清單。
Most products do not die because of one big blow up. They fade out quietly, eaten away by early mistakes that feel harmless at the time. The patterns below come from founders who have seen that slow failure from up close and wrote down how to avoid it.
Paul Graham: the slow ways a startup drifts off course
In "Startup Mistakes" Paul Graham points out that founders rarely get killed by a single obvious bad decision, they get worn down by working on the wrong thing in the wrong way. Drifting away from users, hiring too fast, or letting morale slide all look minor in any single week, but compound until there is no real company left. Staying close to users and default alive in cash buys you time, which is usually the real scarce resource.
Paul Graham: when growth hides that you are in denial
In "Do Things that Do not Scale" Graham warns that teams often hide behind dashboards and automation to avoid the awkward manual work that would actually teach them what to fix. Early growth can mask the fact that no one on the team has spoken to a passionate user this month. Doing unscalable things is not a stunt, it is the antidote to wishful thinking about what users really care about.
YC: advice that sounds basic, but saves products
YC's essential startup advice repeats the same themes because founders keep ignoring them. Talk to users every week, launch earlier than feels comfortable, and keep your burn low enough that you have years of runway, not months. The mistakes that kill products are usually the opposite habits, long stretches without user contact, shipping in big rare batches, and spending as if future fundraising is guaranteed.
A short checklist before your product quietly drifts
- Do at least one real user conversation every week and write down one thing that surprised you
- Ship something visible at least every two weeks so you can watch how real people react
- Keep a single number that tracks whether users would be very disappointed if the product vanished
- Look at burn and runway every month and adjust hiring before the runway gets short
- Make it explicit which user problem you are solving this month and say no to work that does not help
The takeaway
- New products usually die from many small avoidable habits, not one big disaster
- The cure is boring and repetitive, talk to users, ship often, watch cash, repeat
- If you make those habits default, your product gets time to find something people truly care about
多數新產品不是因為一次大爆炸才失敗，而是被一連串看起來不嚴重的小決定慢慢吃掉。下面整理幾種常見的「安靜錯誤」，還有創業者前輩提出的避免方式。
Paul Graham：公司是慢慢偏掉的，不是一夕崩毀
在〈Startup Mistakes〉裡，Paul Graham 提到，新創通常不是被某個超明顯的大錯打趴，而是長期做著「有點不對」的事。離使用者越來越遠、太早大舉招人、讓士氣一路下滑，每一週看起來都還撐得住，累積起來公司就空了。緊貼使用者、維持現金能活下去，買到的其實是時間，而時間往往才是真正稀缺的資源。
Paul Graham：成長有時只是你還不想面對現實
在〈Do Things that Do not Scale〉裡，他提醒很多團隊會躲在儀表板和自動化背後，只為了逃避那些麻煩又尷尬，但真的能學到東西的手工工作。漂亮的早期成長數字，常常遮住一個事實：團隊好幾個禮拜沒有跟任何一個「超級喜歡產品」的使用者說話。那些一開始看起來不有效率的事，其實是對抗自我催眠的解藥。
YC：聽起來老生常談的建議，剛好救了產品
YC 的「必備新創建議」一直在重複幾件事，只是很多團隊聽懂了卻沒有真的去做。每週跟使用者講話、比你覺得安全的時間點更早上線、把燒錢壓在可以撐幾年而不是幾個月的水位。會悄悄殺死產品的，通常就是反過來的習慣：長期沒有接觸使用者、一次憋很久才大改版、用未來一定融得到資金的樂觀心情在花錢。
出發前的自我檢查清單
- 每週至少安排一次真實使用者訪談，事後寫下一件讓你驚訝的觀察
- 每兩週推出一次可見的更新，盯著實際行為變化而不是腦中的想像
- 設一個指標，追蹤「如果產品消失，他會有多失望」的使用者比例
- 每個月檢查一次燒錢和可活多久，先調整招募再去擔心融資
- 清楚寫下這個月主要要解決的單一使用者問題，對不相關的需求練習說不
總結
- 新產品通常是被許多可避免的小習慣拖死，而不是一件大事
- 解法很無聊也很重複：常常跟使用者聊天、頻繁上線、隨時看現金
- 把這些變成團隊預設模式，你才有時間慢慢找到「真的有人在乎」的價值