19年专注于一款产品：Xmind教会我的事
我在2006年使用iBook G4、Eclipse和Java输入了Xmind的第一行代码。大约十年后，我们用Electron + Vue（JS/SVG）重建了前端，以统一平台并淘汰旧的Java用户界面。如今，该产品每月服务约400万用户和数千个付费团队。没有风险投资，没有上市——从第一天起就保持盈利。
以下是六个在现实中经受住考验的原则。
1) 保持私密小型；内心的平静会不断积累。
我们在前十年内团队人数保持在25人以下，并且保持盈利。实际上，保持盈利意味着：现金流纪律、小规模招聘、异步决策文档、按发布周期而非日期驱动的紧急工作，以及不进行季度末的销售冲刺以“达成目标”。权衡是品牌扩展较慢；回报是对路线图的控制以及减少关于幻灯片的会议。
2) 选择一个20年的问题并获得痴迷的权利。
大多数想法像牛奶一样会过期；而思维导图则像葡萄酒一样会随着时间而变得更好。我们用简单的启发式方法来测试“长寿”：低教学成本、高转换成本以及与人类工作流程（规划、学习、研究）的深厚联系。在一个持久的领域中，你可以修复小问题、优化默认设置，并在不担心该领域在下个季度消失的情况下投资文档。
3) 如果用户在晚餐时不向你推荐，你已经在衰退了。
你可以租用注意力；但你无法租用支持。我们最健康的用户群来自直接/推荐流量；我们努力将其保持为主要的注册来源。影响者的热度在一两天内就会消退；而推荐用户的留存率和扩展性更好。优化“推荐朋友”的时刻：可分享的模板、便捷的导出/嵌入，以及显而易见的初步成功。跟踪来源组合、7/30天的留存率以及推荐率。
4) 在你确定之前就收费；市场的反馈比模型更快。
我们早期开始收费（后来转向订阅）。定价是产品的表面：它选择客户并框定期望。我们的循环很简单：发布一个可辩护的价格 → 观察转化率/留存率/退款 → 细分计划 → 根据明确的公开规则进行调整。对早期购买者进行保护。保持免费层的边界清晰。
5) 设计影响潜意识——发布它，而不仅仅是测量它。
人们会使用那些在使用前就感觉可用的工具。我们不再追踪模糊的设计KPI，而是发布直接减少摩擦的功能。一个例子是：我们构建了一个“智能色彩主题”系统，可以自动将你的思维导图样式化为干净的对比和层次结构——这对那些对视觉设计不自信的用户尤其有用。它简单、在第一天就能看到，并且随着时间的推移保持吸引力。
6) 不要根据竞争对手的清单来构建。
功能平衡是模仿工程。我们保持一份世界观文档（我们解决什么问题以及如何解决）和一份负面路线图（我们不会做什么，即使竞争对手在做）。说“不”可以防止杂物抽屉效应，并保持界面的可读性。不同的东西比更大的东西更重要。如果你必须复制任何东西，那就复制其他人忽视的限制。
这些都不是普遍法则；它们使我们在不同周期中保持小型、冷静和活力。如果你想要更少的头痛：选择一个在2045年仍然重要的问题，赢得餐桌上的支持，让定价成为与用户的对话，把设计视为认知，并在功能积累上保持一致性。欢迎在评论中分享更多细节——定价、重写或自筹资金的权衡。
查看原文
I typed the first line of Xmind in 2006 on an iBook G4 with Eclipse and Java. About a decade later we rebuilt the frontend with Electron + Vue (JS/SVG) to unify platforms and retire legacy Java UI. Today the product serves ~4M monthly users and thousands of paying teams. No VC, no IPO—default‑alive since day one.<p>Here are six things that survived contact with reality.<p>1) Stay privately tiny; peace of mind compounds.<p>We kept the team under 25 for the first decade and stayed profitable. Default‑alive in practice means: cash‑flow discipline, small hiring batches, asynchronous decision docs, release trains instead of date‑driven crunches, and no quarter‑end sales blitzes to “make the number.” The trade‑off is slower brand expansion; the return is control of the roadmap and fewer meetings about slides.<p>2) Pick a 20‑year problem and buy the right to obsess.<p>Most ideas age like milk; mind mapping ages like wine. We test “long‑lived” with simple heuristics: low teaching cost, high switching cost, and deep ties to human workflows (planning, learning, research). In a durable category you can fix paper‑cuts, sharpen defaults, and invest in docs without worrying the category evaporates next quarter.<p>3) If users don’t pitch you at dinner, you’re already dying.<p>You can rent attention; you can’t rent advocacy. Our healthiest cohorts come from direct/referral traffic; we try to keep them as the leading sign‑up source. Influencer spikes fade in a day or two; referral users retain and expand better. Optimize for the tell‑a‑friend moment: shareable templates, easy export/embedding, and obvious first wins. Track source mix, 7/30‑day retention by source, and referral rate.<p>4) Charge before you’re sure; the market teaches faster than models.<p>We charged early (then moved to subscriptions later). Pricing is a product surface: it selects customers and frames expectations. Our loop is simple: ship a defensible price → watch conversion/retention/refunds → segment plans → adjust with clear, public rules. Grandfather early buyers. Keep free‑tier boundaries crisp.<p>5) Design hacks the subconscious—ship it, don’t just measure it.<p>People adopt tools that feel usable before they are usable. Instead of tracking vague design KPIs, we ship features that directly reduce friction. One example: we built a “smart color theme” system that auto‑styles your mind map into clean contrast and hierarchy—especially useful for users who aren’t confident with visual design. It’s simple, visible on day one, and sticky over time.<p>6) Don’t build from competitor checklists.<p>Feature parity is cargo‑cult engineering. We keep a Worldview Doc (what problem we solve and how) and a Negative Roadmap (what we won’t do, even if competitors do). Saying “no” prevents the junk‑drawer effect and keeps the interface legible. Different beats bigger. If you must copy anything, copy the constraints others ignore.<p>None of this is universal law; it’s what kept us small, calm, and alive across cycles. If you want fewer headaches: pick a problem that still matters in 2045, earn dinner‑table advocacy, let pricing be a conversation with users, treat design as cognition, and guard coherence over feature accretion. Happy to share details in the comments—pricing, rewrite, or bootstrapping trade‑offs.