90 minutes of unfiltered product advice from Snap and Discord’s product chief | Peter Sellis

在与伦尼·拉奇茨基（Lenny Rachitsky）的一次坦诚对话中，彼得·塞利斯（Peter Sellis）——Snapchat的前首位产品经理及Discord的产品主管——分享了几个挑战传统创业智慧的反直觉经验和信念，这些观点并未受到公司公关团队的束缚。 塞利斯颇具争议地表示，他喜欢将团队设计成一个“恐怖组织”。这并非暗示暴力，而是指两个关键要素：一个清晰的、意识形态化的文化，让每个人都理解“为什么”；以及一个明确的组织结构，界定谁可以被信任来做哪些决策。他认为，协作具有隐藏的协调成本，会减缓高增长公司的速度。通过培养一个共同的“宗教级目标”和明确的责任归属（一个“单一责任人”，即“问责对象”），团队可以避免过度协作并更快地行动，做出的决策仿佛创始人亲自做出。 塞利斯的另一个引人注目的观点是他的人才管理方法：他宁愿“让最优秀的员工竭尽所能”，也不愿努力提升表现平平的员工。他相信，应该给予高绩效者“越来越多的绳索”来挑战自我，将他们推向极限。这源于一个原则：公开维护他的团队，并持续让自信、有能力的个体承担更多责任，直到他们开始显露其局限性，他坚信成功具有自我强化的特性。 塞利斯还提出，“大多数产品经理都很糟糕，并且常常对公司产生负面影响”。他将此归因于一个“数学解释”：产品经理的技能遵循幂次法则分布。最有才华的产品经理会成为创始人或高管，从而离开这个系统。与此同时，该行业对非技术人员而言丰厚的报酬，激励着效率较低的产品经理留下，拉低了产品经理的平均技能水平。他认为，“平庸”的工程师是那些抱怨产品经理的人，而优秀的工程师会利用他们，糟糕的工程师则躲在他们身后。 回顾他在Snapchat的时光，塞利斯分析了为何该公司尽管拥有庞大的用户群，却未能达到预期的业务规模。他为其广告业务指出了三个核心挑战： 1. **年轻用户基础：** Snapchat对Z世代的吸引力意味着针对18/21岁以下用户的广告法规更严格，广告千次展示费用（CPMs）更低，使得投资回报率（ROI）更难实现。 2. **摄像头优先产品：** 与基于信息流的应用不同，Snapchat的核心流程（从相机拍摄到分享）缺乏自然、合乎逻辑的广告位，使得摄像头变现困难。 3. **消息应用核心：** 消息应用是实用工具，而非消费导向，通常无法有效实现广告变现。他强调，尽管存在这些障碍，Snapchat在广告收入方面比推特（Twitter）、红迪（Reddit）和品趣（Pinterest）等竞争对手更快地取得了显著成就。他还思忖，如果能更早地投资于Snapchat Plus等订阅产品（后来证明很成功），或许可以减轻广告业务的压力。 塞利斯强调了产品开发中“品味”的重要性，特别是“说不”的魄力和保持克制。他表示：“如果一个博物馆把所有的藏品都挂在墙上，那策展人就什么也没做。”这暗示一个真正伟大的产品（就像一个精心策划的博物馆）只呈现少量精选的高质量创意，将许多可能不错但“不合时宜”的概念暂时搁置。他指出，Snap的埃文（斯皮格尔）在这方面表现出色，他亲自否决的创意比许多消费级创业公司推出的产品还要好。他认为，随着人工智能生成了大量创意选项，这种“做减法”的技能变得越来越重要，成为产品经理的关键职能。 关于增长，塞利斯认为它“通常来自于业务的核心”，而非总是来自新的扩张。他引用了Snapchat在2018年改版后专注于现有用户的安卓性能提升，以及Discord在2024年转向为游戏玩家（其核心受众）优化的例子。通过加深核心用户群体的参与度和留存率（例如，提高日活跃用户/月活跃用户DAU/MAU比率），可以实现显著增长。这种方法避免了“创新者的窘境”，他认为SpaceX愿意淘汰其成功的猎鹰9号火箭，转而专注于星舰，优先考虑长期、雄心勃勃的任务，就 brilliantly 诠释了这一教训。 他还提出了广告业务中“长期贪婪”的概念——今天优化广告商的投资回报率，确保明天的复合增长，即使牺牲短期收入也在所不惜。他还倡导“核心产品价值”（CPV），这是一个将使命与可衡量的指标相结合的短语（例如，Snap的“与你最好的朋友分享瞬间的最快方式”），并反复传达，以使整个组织保持一致。 最后，他将优秀的产品经理描述为一种“矛盾体”，体现了三对自相矛盾的特质：自信却谦逊，有条理却能适应模糊性，以及既有紧急的行动偏好，又有长期的战略思维。 塞利斯的异见者视角提供了犀利观点：最重要的职业决策是选择你的生活伴侣；品味在于知道何时在正确的趋势前止步；以及某些“不可剥夺的特质”（性别、种族、社会经济背景）可能天生更适合产品经理开发特定的消费产品，尽管他也承认这个观点不可操作且敏感。

In a candid conversation with Lenny Rachitsky, Peter Sellis, former first PM at Snapchat and Head of Product at Discord, shared several counterintuitive lessons and beliefs that challenge conventional startup wisdom, unburdened by corporate communications teams. Sellis controversially stated he likes to design teams like a "terrorist organization." This doesn't imply violence but rather two key elements: a clear, ideological culture where everyone understands the "why," and a distinct organizational structure defining who can be trusted with which decisions. He argues that collaboration has hidden coordination costs, slowing down high-growth companies. By fostering a shared "religious-level goal" and clear ownership (a "single threaded owner," the "throat to choke"), teams can avoid over-collaboration and move faster, making decisions as if the founders themselves had made them. Another striking take is Sellis's approach to talent management: he prefers "riding his best people into the ground" rather than trying to improve average performers. He believes in handing high-performers "more and more rope to hang themselves with," pushing them to their limits. This stems from a principle of publicly defending his team and consistently loading confident, capable individuals with more responsibility until they begin to show limits, believing in the self-reinforcing nature of success. Sellis also posited that "most product managers are bad and are often net negative to the company." He attributes this to a "mathematical explanation": PM skill is a power distribution. The most talented PMs become founders or executives, exiting the system. Meanwhile, the profession's lucrative nature for non-technical individuals incentivizes less effective PMs to stay, skewing the median skill level downwards. He believes the "mid-wit" engineers are the ones who complain about PMs, whereas great engineers utilize them, and poor engineers hide behind them. Reflecting on his time at Snapchat, Sellis analyzed why it, despite its massive user base, didn't achieve the expected business scale. He identified three core challenges for its ads business: 1. **Young User Base:** Snapchat's appeal to Gen Z meant stricter advertising regulations and lower ad CPMs for users under 18/21, making ROI harder to drive. 2. **Camera-First Product:** Unlike feed-based apps, Snapchat's core flow (camera to sharing) lacked a natural, logical placement for ads, making camera monetization difficult. 3. **Messaging App Core:** Messaging apps are utilities, not consumption-oriented, and don't typically monetize ads effectively. He highlighted that despite these hurdles, Snapchat achieved significant ad revenue faster than competitors like Twitter, Reddit, and Pinterest. He also mused that earlier investment in subscription products like Snapchat Plus, which proved successful, might have eased the pressure on the ad business. Sellis emphasized the importance of "taste" in product development, particularly the muscle of "saying no" and exercising restraint. He stated, "If a museum has all of their collection on the walls, then the curator hasn't done anything," suggesting that a truly great product (like a well-curated museum) presents only a select, high-quality slice of ideas, leaving many potentially good but not "right-for-now" concepts on the shelf. He noted Evan (Spiegel) at Snap was exceptional at this, personally rejecting ideas better than many consumer startups launch. This skill, he believes, is increasingly vital as AI generates abundant creative options, making "removing things" a critical PM function. Regarding growth, Sellis argued that it "typically comes from the core of the business," not always from new expansions. He cited Snapchat's post-2018 redesign focus on Android performance for existing users and Discord's 2024 pivot to optimize for gamers, their core audience, as examples. By deepening engagement and retention within the core (e.g., increasing DAU/MAU ratios), substantial growth can be achieved. This approach avoids the "innovator's dilemma," a lesson he sees brilliantly exemplified by SpaceX's willingness to sunset its successful Falcon 9 rocket to focus on Starship, prioritizing a long-term, ambitious mission. He introduced the concept of "long-term greedy" for ads – optimizing for advertiser ROI today ensures compounding growth tomorrow, even at the cost of short-term revenue. He also championed "Core Product Value" (CPV), a phrase (e.g., "fastest way to share a moment with your best friends" for Snap) that combines mission with measurable metrics, repeatedly communicated to align the entire organization. Finally, he described a great PM as an "oxymoron," embodying three paradoxical pairs: confident yet humble, organized yet comfortable with ambiguity, and possessing urgent bias for action alongside long-term strategic thinking. Sellis's contrarian corner offered spicy takes: the most important career decision is choosing your life partner; taste is knowing when to stop at the right trend; and certain "inalienable traits" (gender, race, socioeconomic background) may inherently suit PMs better for developing specific consumer products, though acknowledging the non-actionable and sensitive nature of this point.