"Frontend isn't dying — only the comfort of 'only doing frontend' is dying. Tech stacks age, but problem-solving ability, data-driven hypothesis testing, and rapid iteration are engineer strengths regardless of domain." — JaredL (@jaredliu_bravo)
This is not a theoretical framework. It is a practitioner's playbook.
JaredL spent six years as a frontend engineer at NetEase, Alibaba, and an A-round startup before leaving the corporate track entirely. After a year of indie development and an unfunded team building a news-brief product called Tidyread, he joined YouMind as a product engineer — and gradually evolved into something the industry calls a "growth engineer."
The results speak for themselves: his growth projects now account for more than 50% of YouMind's total platform traffic. This article documents exactly how he got there, what strategies he deployed, and what other frontend engineers should know if they want to make the same transition.
「前端死的不是前端，是『只寫前端』這件事的安全感。技術棧會老化，但解決問題的能力、數據驅動的假設驗證、快速迭代——這些才是工程師不分領域的核心優勢。」 —— JaredL（@jaredliu_bravo）
這不是一個理論框架，這是一份實戰手冊。
JaredL 在網易、阿里巴巴和一家 A 輪新創公司做了六年前端工程師，然後徹底離開了企業軌道。經歷一年的獨立開發、帶著一支沒有融資的團隊做了一款叫 Tidyread 的新聞摘要產品之後，他加入 YouMind 擔任產品工程師——然後逐漸演變成業界所說的「成長工程師」。
成果不言自明：他負責的成長專案現在佔 YouMind 平台總流量的 50% 以上。這篇文章完整記錄了他是怎麼走到這一步的、用了哪些策略、以及其他前端工程師如果想做同樣的轉型，需要知道什麼。
The Career Timeline
2016–2022: The Frontend Years. NetEase, Alibaba, then frontend lead at an A-round startup. Standard big-company trajectory — solid technical chops, growing leadership responsibilities, and an increasingly nagging feeling that writing pixel-perfect components was not the endgame.
2023: The Indie Detour. Left the corporate world. Formed an unfunded team and built Tidyread, a news-brief product. No venture capital, no safety net. This year taught two things that no corporate job could: how to think about distribution from day one, and how painful it is when nobody knows your product exists.
2024: The Pivot. Joined YouMind as a product engineer. Started building features. Gradually noticed that the skills he had picked up during the indie year — understanding user acquisition, thinking about SEO, caring about conversion — were more valuable than his React expertise. Began transitioning toward growth work.
2025: Growth Engineer. Full-time growth at YouMind. Driving 50%+ of platform traffic. The frontend skills did not disappear — they became the execution layer for growth strategies.
The key insight from this timeline: the transition was not a single leap. It was a gradient. Each phase added a layer — from pure technical execution, to product sense, to distribution awareness, to full growth ownership. No single year was wasted, including the "failed" indie year that most career advisors would call a detour.
職涯時間線
2016–2022：前端時期。 網易、阿里巴巴，接著是 A 輪新創公司的前端負責人。標準的大廠路徑——扎實的技術能力、逐漸增加的管理職責，以及一種越來越揮之不去的感覺：寫到像素級完美的元件，不是終點。
2023：獨立開發彎路。 離開企業體制。組了一支沒有融資的團隊，做了 Tidyread，一款新聞摘要產品。沒有創投、沒有安全網。這一年教會了兩件企業工作永遠教不會的事：從第一天就要想 distribution（分發），以及沒人知道你的產品存在時有多痛苦。
2024：轉折。 加入 YouMind 擔任產品工程師。開始做功能開發。逐漸發現獨立開發那一年學到的技能——理解用戶獲取、思考 SEO、關注轉換率——比他的 React 專業更有價值。開始往成長工作的方向靠攏。
2025：成長工程師。 在 YouMind 全職投入成長。驅動平台 50% 以上的流量。前端技能沒有消失——它們變成了成長策略的執行層。
這條時間線的關鍵洞察：轉型不是一次跳躍，而是一個漸變。每個階段加一層——從純技術執行，到產品感覺，到分發意識，到完整的成長負責。沒有哪一年是浪費的，包括那個大多數職涯顧問會稱為「彎路」的獨立開發年。
What Exactly Is a Growth Engineer?
The simplest definition: a growth engineer is an engineer who optimizes for business metrics instead of user experience metrics.
The framework most growth teams use is AARRR — Acquisition, Activation, Retention, Revenue, Referral. In practice, a growth engineer's primary focus tends to be on the first two: getting new users in the door and making sure they experience the product's core value as fast as possible.
This sounds simple. It is not. Because the skill set is fundamentally different from what product engineers are trained to do.
| Dimension | Product Engineer | Growth Engineer |
|---|---|---|
| Primary goal | Deliver great user experience | Move growth metrics |
| Decision approach | Build for completeness and polish | Prioritize by ROI, kill low-performers fast |
| Extra skills required | Deep technical specialization | Track trends, create content, study distribution channels |
The critical difference is not technical — it is mental. A product engineer asks "Is this feature well-built?" A growth engineer asks "Is this feature bringing in users, and at what cost?"
Frontend engineers are actually well-positioned for this transition. You already understand the browser, performance optimization, page structure, and user interaction patterns. Growth engineering simply redirects those skills toward a different objective function.
Why the Timing Is Right
The rise of AI tools has compressed the content creation pipeline dramatically. What used to require a marketing team of five — keyword research, content writing, translation, design, publishing — can now be executed by a single engineer with the right automation stack. This means the barrier to entry for growth engineering has dropped, but the barrier to doing it well remains high. Engineers who can build the automation layer, not just use the tools, have an outsized advantage.
到底什麼是成長工程師？
最簡單的定義：成長工程師是一個為商業指標而非用戶體驗指標做優化的工程師。
大多數成長團隊使用的框架是 AARRR——Acquisition（獲取）、Activation（啟動）、Retention（留存）、Revenue（營收）、Referral（推薦）。實務上，成長工程師的主要焦點往往在前兩個：把新用戶帶進來，確保他們盡快體驗到產品的核心價值。
聽起來簡單，實際上不是。因為所需的技能組合跟產品工程師被訓練的東西有根本性的不同。
| 維度 | 產品工程師 | 成長工程師 |
|---|---|---|
| 主要目標 | 交付優秀的用戶體驗 | 推動成長指標 |
| 決策方式 | 追求完整性和精緻度 | 按 ROI 排優先級，快速砍掉低效的 |
| 額外需要的技能 | 深度技術專精 | 追蹤趨勢、創作內容、研究分發渠道 |
關鍵的差異不在技術層面——在心態層面。產品工程師問的是「這個功能做得好不好？」成長工程師問的是「這個功能有沒有帶來用戶？成本多少？」
前端工程師其實很適合做這個轉型。你已經懂瀏覽器、效能優化、頁面結構和用戶互動模式。成長工程只是把這些技能重新導向一個不同的目標函數。
為什麼現在是好時機
AI 工具的崛起大幅壓縮了內容創建的 pipeline。以前需要一個五人行銷團隊的工作——關鍵字研究、內容撰寫、翻譯、設計、發佈——現在一個工程師搭配正確的自動化工具就能搞定。這意味著成長工程的入門門檻降低了，但做得好的門檻依然很高。能夠建造自動化層（而不只是使用工具）的工程師，擁有不成比例的優勢。
Strategy 1: Collection Sites (Aggregation Pages)
This was the single biggest traffic driver. The core formula is deceptively simple:
"Free, abundant, high-value resources" → organic sharing by creators and KOLs
Build a site that curates something valuable, give it away for free, and let the people who benefit from it do the distribution for you. The trick is in the execution details.
The 5 Requirements for a Successful Collection Site
1. Speed: Launch within the first week of a trend.
Timing is everything. When a new AI model drops and generates buzz, you have roughly one week before the market is saturated with copycat resource pages. JaredL's team launched a Nano Banana Pro prompts site during the early days of the model's hype cycle. Being first meant capturing the initial wave of creators searching for curated resources.
If you miss the first week, the opportunity cost rises exponentially. By week three, established sites have already claimed the top search positions.
2. Multilingual from day one: 16 languages, 50+ timezones.
This is where most engineers underinvest. A Turkish KOC (Key Opinion Consumer) generated 800,000 impressions from a single piece of localized content. That is traffic you simply do not get if your site only exists in English and Chinese.
The implementation is straightforward — AI translation pipelines handle most of the heavy lifting. The insight is strategic: non-English markets are dramatically underserved for AI resource curation, and the competition is orders of magnitude lower.
3. Continuous updates through automation.
A collection site that stops updating is a dead site. JaredL's team built automated daily scraping of X (Twitter) posts, filtered to quality entries through interaction metrics. The result: 12,000+ accumulated prompts — the largest repository globally — and 10,000+ GitHub stars.
Automation is non-negotiable. Manual curation does not scale, and freshness signals matter for both SEO and user trust.
4. Quality curation is what separates you from a database dump.
Scraping is easy. Curation is hard. The pipeline filters by interaction metrics (likes, reposts, replies), removes low-quality content, and supplements with community submissions. Without this step, you end up with noise, and noise drives users away faster than an empty page.
5. Distinctive UI: avoid the generic component library look.
This matters more than engineers think. JaredL's team adopted a "new brutalism" visual style after observing design trends around Gemini 3's release. The goal is immediate visual differentiation — when a creator screenshots your site to share on social media, it needs to look distinctive enough to generate curiosity.
If your collection site looks like every other Shadcn/Tailwind template, nobody will screenshot it.
Data Infrastructure: Dual-Channel Distribution
The backend runs on Payload CMS with a synchronized GitHub repository. This dual-channel approach serves two purposes:
- CMS: Powers the web frontend, handles content management, enables automated publishing
- GitHub: Provides 12%+ of total traffic. Developers discover the resource through GitHub trending, star it, and many convert to the main product
A critical insight about GitHub: once a repository hits 200+ stars, it qualifies for DR96 (Domain Rating 96) dofollow backlinks. This is enormous for SEO. GitHub's domain authority means a single backlink from a starred repo can move your site's search rankings significantly.
The Wordware Warning
Not all collection sites succeed in converting traffic to product users. Wordware's twitter.wordware.ai is the cautionary tale: it hit 10 million peak monthly visitors, then dropped 95% the following month. Registration conversion was only 3–4%.
The problem was audience-product misalignment. The people visiting a viral Twitter personality analysis tool are not the same people who need an AI workflow builder. Traffic without intent alignment is vanity metrics.
Why 20% Conversion When Wordware Got 3%?
The difference comes down to audience-product fit. YouMind's collection sites attract people who are actively experimenting with AI models and need a tool to organize their workflows. The collection site is a natural top-of-funnel for the product itself. Wordware's viral tool attracted curiosity seekers who had no use for a workflow builder.
The lesson: when designing a collection site, start with the question "who is my product for?" and build the resource for that audience. Do not chase maximum traffic — chase maximum relevant traffic.
Results
- ~20% registration conversion rate (5x higher than Wordware's)
- 50%+ of total site traffic attributed to collection sites
- 10+ point improvement in Domain Rating
- GitHub companion repos contributed 12%+ of total referral traffic
- The largest prompt repository in the world at 12,000+ entries
策略一：收錄站（聚合頁）
這是最大的單一流量來源。核心公式看起來騙人般的簡單：
「免費、大量、高價值的資源」→ 創作者和 KOL 自發分享
做一個網站去策展某些有價值的東西，免費提供，然後讓受益的人替你做分發。訣竅在執行的細節。
成功收錄站的 5 大要件
1. 速度：在趨勢出現的第一週上線。
時機就是一切。當一個新的 AI 模型發佈並引發討論，你大概只有一週的時間窗口，之後市場就會被跟風的資源頁淹沒。JaredL 的團隊在 Nano Banana Pro 模型剛爆發熱度時就上線了 prompts 收錄站。搶先意味著抓住第一波搜索策展資源的創作者。
如果你錯過了第一週，機會成本會指數級上升。到了第三週，已經有站穩搜索頂部位置的網站了。
2. 第一天就做多語言：16 種語言、50+ 個時區。
這是大多數工程師投入不足的地方。一個土耳其的 KOC（關鍵意見消費者）光靠一篇本地化內容就產生了 80 萬次曝光。如果你的網站只有英文和中文，這些流量你根本拿不到。
實作其實不難——AI 翻譯 pipeline 承擔了大部分工作。關鍵洞察是策略層面的：非英語市場在 AI 資源策展上被嚴重低估，競爭程度低了好幾個數量級。
3. 透過自動化持續更新。
停止更新的收錄站就是一個死站。JaredL 的團隊建了自動化的每日 X（Twitter）貼文抓取，通過互動指標過濾到高品質內容。成果：累積 12,000+ 個 prompts——全球最大的資料庫——以及 10,000+ GitHub stars。
自動化不是可選項。手動策展無法規模化，而新鮮度信號對 SEO 和用戶信任都很重要。
4. 品質策展是你跟資料庫 dump 的分界線。
抓取很容易，策展很難。pipeline 會按互動指標（按讚、轉發、回覆）過濾、移除低品質內容，並補充社群投稿。沒有這一步，你最後得到的就是雜訊，而雜訊趕走用戶的速度比空白頁面還快。
5. 有辨識度的 UI：不要用一看就是元件庫的外觀。
這比工程師想的更重要。JaredL 的團隊在觀察到 Gemini 3 發佈前後的設計趨勢後，採用了「新粗獷主義」（new brutalism）的視覺風格。目標是即時的視覺差異化——當創作者截圖你的網站分享到社群媒體時，它必須看起來夠獨特，才能引發好奇心。
如果你的收錄站看起來跟其他每一個 Shadcn/Tailwind 模板一樣，沒有人會截圖。
資料基礎設施：雙渠道分發
後端跑在 Payload CMS 上，同步到一個 GitHub 倉庫。這個雙渠道策略有兩個目的：
- CMS：驅動 web 前端、處理內容管理、實現自動化發佈
- GitHub：提供 12% 以上的總流量。開發者通過 GitHub trending 發現資源、star 它，然後很多人轉化到主產品
一個關於 GitHub 的關鍵洞察：當一個 repo 達到 200+ stars，就有資格獲得 DR96（Domain Rating 96）的 dofollow 反向連結。這對 SEO 的影響是巨大的。GitHub 的域名權威意味著一條來自高星 repo 的反向連結就能顯著推動你網站的搜索排名。
Wordware 的警示
不是所有收錄站都能成功把流量轉化為產品用戶。Wordware 的 twitter.wordware.ai 是一個典型的反面教材：它的月訪問量峰值達到 1,000 萬，然後下個月暴跌 95%。註冊轉化率只有 3–4%。
問題出在受眾和產品的錯配。去玩 Twitter 人格分析工具的人，跟需要 AI 工作流構建器的人，不是同一群人。沒有意圖匹配的流量就是虛榮指標。
為什麼轉化率 20% 而 Wordware 只有 3%？
差別在受眾和產品的契合度。YouMind 的收錄站吸引的是正在積極嘗試 AI 模型、需要工具來組織工作流的人。收錄站是產品本身的天然漏斗頂端。Wordware 的病毒式工具吸引的是好奇心驅動的人，他們根本不需要工作流構建器。
教訓：設計收錄站時，先問「我的產品是給誰用的？」然後為那群人建資源。不要追最大流量——追最大的相關流量。
成果
- 約 20% 的註冊轉化率（是 Wordware 的 5 倍）
- 50% 以上的全站流量歸因於收錄站
- Domain Rating 提升 10+ 分
- GitHub 配套 repo 貢獻了 12% 以上的推薦流量
- 全球最大的 prompt 資料庫，累積 12,000+ 條
Strategy 2: Chrome Extension Optimization
Think of Chrome Web Store optimization as SEO for extensions. The ranking factors are similar in spirit: title keywords, conversion rate (search-to-install ratio), ratings, description copy, and keyword frequency.
JaredL's team drove YouMind's Chrome extension to 40,000 weekly active users through three specific tactics.
Tactic 1: Brand Term Strategy (Competitor Keywords)
This is the highest-leverage play. Competitor brand names carry high search volume because users are already looking for that category of tool.
What worked:
- Embedded "Gemini Assistant" in the extension title → ranked on page 1 across multiple languages
- Embedded "Rytr" in the description to capture writing-tool traffic
What went wrong:
- Embedding "MyMind" as a brand term triggered a complaint and the extension was temporarily removed from the store
Risk mitigation: Distribute competitor brand terms across different language versions of your listing. If a complaint targets the English listing, your French, Japanese, and Spanish versions continue ranking. Do not concentrate all brand terms in one language.
Tactic 2: Informational Term Strategy (Feature Keywords)
These are non-branded, feature-based keywords that capture users searching for capabilities rather than specific products.
- English: "Web Clipper", "YouTube Notes"
- Chinese: "剪藏" (web clipping)
The placement hierarchy matters: primary keywords go in the title, secondary keywords go in multilingual descriptions. The title carries disproportionate ranking weight, so reserve it for your highest-volume terms.
Tactic 3: Get the Featured Badge Early
This was a counterintuitive insight. JaredL's team applied for Chrome Web Store's "Featured" badge when they had only 1,000 users and basic reviews. They expected rejection. Instead, they were approved in roughly one week.
The immediate effect was a traffic boost from the badge's visibility in search results.
The strategic insight is more important: get the Featured badge BEFORE you start aggressive keyword optimization. Larger, more established products that are already keyword-stuffing their listings may face higher scrutiny during the badge review process. Apply when you are small and clean, then optimize aggressively afterward.
The Ranking Formula in Practice
Chrome Web Store ranking is simpler than Google Search, which makes it more hackable. The core factors, roughly in order of weight:
- Title match: Does the search term appear in your extension title?
- Install conversion rate: Of people who see your listing in search results, what percentage clicks "Add to Chrome"?
- Ratings and reviews: Both quantity and average score matter
- Description keyword density: Secondary to title, but still influential
- Weekly active users: A proxy for quality that the algorithm uses for ranking stability
The implication: optimizing for Chrome Web Store is a two-step game. First, get your listing in front of searchers (keyword strategy). Second, make them click install (conversion optimization — icon design, screenshots, description copy).
Results
40,000 weekly active users on the Chrome extension — a meaningful acquisition channel that compounds over time as ratings and install counts improve organic ranking.
策略二：Chrome 擴充功能優化
把 Chrome Web Store 優化想成是擴充功能的 SEO。排名因素在精神上是類似的：標題關鍵字、轉化率（搜索到安裝的比率）、評分、描述文案和關鍵字密度。
JaredL 的團隊通過三個具體戰術，把 YouMind 的 Chrome 擴充功能推到了每週 40,000 活躍用戶。
戰術一：品牌詞策略（競品關鍵字）
這是槓桿最高的打法。競品品牌名稱自帶高搜索量，因為用戶已經在搜索那個品類的工具了。
有效的做法：
- 在擴充功能標題嵌入「Gemini Assistant」→ 在多個語言版本中排到第一頁
- 在描述中嵌入「Rytr」來截獲寫作工具的流量
翻車的經歷：
- 嵌入「MyMind」作為品牌詞，結果觸發投訴，擴充功能被暫時下架
風險控制： 把競品品牌詞分散到不同語言版本的 listing 中。如果投訴針對的是英文 listing，你的法語、日語和西班牙語版本會繼續排名。不要把所有品牌詞集中在一種語言。
戰術二：資訊詞策略（功能關鍵字）
這些是非品牌的、基於功能的關鍵字，捕獲的是搜索能力而非特定產品的用戶。
- 英文："Web Clipper"、"YouTube Notes"
- 中文：「剪藏」
放置的層級很重要：主要關鍵字放標題，次要關鍵字放多語言描述。標題的排名權重不成比例地高，所以留給你搜索量最大的詞。
戰術三：盡早拿到 Featured 徽章
這是一個反直覺的洞察。JaredL 的團隊在只有 1,000 個用戶和基本評分的時候就申請了 Chrome Web Store 的「Featured」徽章。他們預期被拒絕。結果大約一週就通過了。
直接效果是搜索結果中徽章的可見度帶來的流量提升。
但策略洞察更重要：在你開始激進的關鍵字優化之前，先拿到 Featured 徽章。 那些已經在 listing 裡塞滿關鍵字的大型成熟產品，在徽章審核時可能面臨更高的審查標準。在你還小而乾淨的時候申請，通過之後再激進優化。
排名公式的實際運作
Chrome Web Store 的排名比 Google 搜索簡單，這意味著更容易被破解。核心因素大致按權重排列：
- 標題匹配：搜索詞是否出現在你的擴充功能標題中？
- 安裝轉化率：在搜索結果中看到你 listing 的人，有多少比例點了「加到 Chrome」？
- 評分和評論：數量和平均分都重要
- 描述關鍵字密度：次於標題，但仍有影響
- 每週活躍用戶：演算法用來維持排名穩定的品質代理指標
啟示：Chrome Web Store 的優化是兩步棋。第一步，讓你的 listing 出現在搜索者面前（關鍵字策略）。第二步，讓他們點安裝（轉化優化——圖標設計、截圖、描述文案）。
成果
Chrome 擴充功能達到每週 40,000 活躍用戶——這是一個會隨時間複利增長的獲客渠道，因為評分和安裝數會持續改善自然排名。
Strategy 3: SEO and Blog Content
For a frontend engineer, technical SEO is home turf. Page structure, internal linking architecture, per-page performance optimization, conversion tracking implementation — these are all skills you already have. The gap is on the content side.
Technical SEO (The Easy Part for Frontend Engineers)
Everything here is frontend work wearing a different hat:
- Page structure: Semantic HTML, proper heading hierarchy, schema markup
- Internal linking: Programmatic link architectures that distribute page authority
- Performance: Core Web Vitals optimization — you have been doing this for years
- Conversion tracking: Event instrumentation, funnel analysis setup
If you are a frontend engineer reading this, you can handle all of the above in your sleep. The hard part is knowing what pages to build and what content to put on them.
AI-Assisted Content Pipeline
This is where the workflow gets interesting. JaredL's team built a full pipeline:
- OpenClaw scrapes trending topics and identifies content opportunities
- YouMind's SEO Blog Skill generates article drafts based on keyword research
- Articles are published through the CMS
- Automated translation extends each article to all supported languages
The key insight: AI does not replace content strategy. It replaces content production labor. You still need to identify the right keywords, validate search intent, and ensure topical authority. AI just makes it economically viable to produce content at the scale needed to compete.
Benchmark Sites Worth Studying
- Skywork.ai: 30% of traffic from SEO — proof that AI products can build organic search moats
- NoteGPT.io: Strong programmatic SEO execution with well-structured tool pages
- QuillBot.com: One of the best examples of content-led growth in the writing tool space
Study how these sites structure their pages, what keywords they target, and how they convert organic traffic into product users. Reverse-engineering successful SEO is one of the fastest ways to learn.
The Content Volume Game
One uncomfortable truth about SEO: it is partially a volume game. A single well-written article might rank, but topical authority — which Google increasingly rewards — requires covering a topic cluster comprehensively. That means dozens or hundreds of related articles.
Before AI, this required a content team. Now, a single growth engineer with the right pipeline can produce 5–10 articles per day across multiple languages. The quality bar is "useful and accurate," not "Pulitzer-worthy." Most SEO content competes with other AI-generated content, so the differentiator becomes speed, topical coverage, and the technical SEO fundamentals that most content farms ignore.
This is why the AI-assisted pipeline matters: it is not about replacing human judgment, it is about making human judgment scalable.
策略三：SEO 與部落格內容
對前端工程師來說，技術 SEO 是主場。頁面結構、內部連結架構、逐頁效能優化、轉化追蹤——這些全是你已有的技能。差距在內容端。
技術 SEO（前端工程師的容易部分）
這裡所有的事情都是穿了不同帽子的前端工作：
- 頁面結構：語意化 HTML、正確的標題層級、schema 標記
- 內部連結：程式化的連結架構，分配頁面權重
- 效能：Core Web Vitals 優化——你已經做了好幾年了
- 轉化追蹤：事件埋點、漏斗分析設置
如果你是正在讀這篇文章的前端工程師，以上這些你閉著眼睛都能搞定。難的部分是知道該做什麼頁面、上面放什麼內容。
AI 輔助的內容 Pipeline
這裡的工作流程開始有趣了。JaredL 的團隊建了一條完整的 pipeline：
- OpenClaw 抓取熱門話題，識別內容機會
- YouMind 的 SEO Blog Skill 基於關鍵字研究生成文章草稿
- 文章通過 CMS 發佈
- 自動翻譯把每篇文章擴展到所有支援的語言
關鍵洞察：AI 取代的不是內容策略，而是內容生產的勞動力。你仍然需要識別正確的關鍵字、驗證搜索意圖、確保主題權威性。AI 只是讓在所需的規模上生產內容變得經濟可行。
值得研究的標竿網站
- Skywork.ai：30% 的流量來自 SEO——證明 AI 產品可以建立自然搜索護城河
- NoteGPT.io：出色的程式化 SEO 執行，工具頁面結構清晰
- QuillBot.com：寫作工具領域內容驅動成長的最佳案例之一
研究這些網站如何組織頁面結構、瞄準哪些關鍵字、如何把自然流量轉化為產品用戶。逆向工程成功的 SEO 是最快的學習方式之一。
內容量的遊戲
一個不太舒服的 SEO 真相：它部分是一個量的遊戲。一篇寫得好的文章也許能排上去，但主題權威性——Google 越來越看重的東西——需要全面覆蓋一個主題集群。這意味著幾十甚至幾百篇相關文章。
在 AI 之前，這需要一個內容團隊。現在，一個成長工程師搭配正確的 pipeline 就能每天產出 5–10 篇多語言文章。品質標準是「有用且準確」，不是「普立茲獎級別」。大多數 SEO 內容的競爭對手也是 AI 生成的內容，所以差異化因素變成了速度、主題覆蓋範圍，以及大多數內容農場忽略的技術 SEO 基本功。
這就是 AI 輔助 pipeline 重要的原因：它不是取代人類判斷，而是讓人類判斷可以規模化。
Strategy 4: Product Hunt
Product Hunt's direct traffic contribution is minimal. The real value lies elsewhere: external link badges embedded across your site, press mentions, and the social proof that comes with top rankings.
JaredL has two launches to his name: Tidyread won Day #1 in 2024, and YouMind achieved Day, Week, and Month rankings in 2025.
Vote Preparation: Start 3 Months Out
Product Hunt success is not about launch day. It is about the three months before it.
Team account cultivation: Every team member needs to build their Product Hunt account to at least bronze or silver hunter status. Fresh accounts casting votes look suspicious and may be discounted by the algorithm.
Automated outreach tooling: JaredL's team built a plugin that auto-scrapes the social links of teams launching on the same day, sorted by member count. This enables targeted outreach for mutual vote support — you help their launch, they help yours.
LinkedIn PH Groups: Direct connection requests within Product Hunt community groups on LinkedIn yield better results than public posts asking for votes. The conversion rate on personal, one-to-one requests is significantly higher than broadcast asks.
The Hunter Advantage
Top hunters can dramatically improve your launch visibility:
- Chris Messina (Top 1 hunter): "100% featured if he hunts your product." Getting hunted by Messina essentially guarantees the featured badge.
- Ben Lang and Ryan Hoover are also high-impact hunters worth pursuing.
Building relationships with top hunters should start months before your launch. Cold outreach on launch day rarely works.
The Vote Exchange Economy
Product Hunt launches operate in a surprisingly well-organized mutual support ecosystem. Here is how it works in practice:
- Identify launches on your target day: Use the upcoming launches page or your custom scraping tool
- Reach out 1–2 weeks before: Offer genuine feedback on their product plus a vote exchange
- Day-of coordination: Share your launch link in a shared group; upvote theirs in return
- Post-launch follow-up: Leave thoughtful comments on each other's launches — comments carry weight in the ranking algorithm
The teams that win on Product Hunt are not necessarily the ones with the best products. They are the ones with the best-organized launch playbooks. Engineering this process — building the scraper, automating outreach, tracking vote exchanges — is exactly the kind of systems thinking that growth engineers excel at.
The Featured Badge Is Everything
Without the featured badge, you cannot achieve meaningful rankings or earn external links. The decision happens within the first 4 hours of your launch.
JaredL's advice: If you are not featured within 4 hours, seriously consider abandoning the launch and relaunching later with better preparation. An unfeatured launch yields almost zero value and burns your product's launch opportunity.
The Honest Truth About Product Hunt
PH's pure direct traffic is negligible for most products. The real value is threefold:
- External link badges: The "Product of the Day" badge on your site provides a high-quality backlink and improves DR
- Press and newsletter mentions: Being a top-ranked launch often triggers coverage in tech newsletters and blog roundups
- Social proof: The ranking badge on your landing page improves conversion rates for all other traffic sources — paid ads, organic search, referrals all convert better when visitors see "#1 Product of the Day"
策略四：Product Hunt
Product Hunt 的直接流量貢獻微乎其微。真正的價值在其他地方：嵌入你網站各處的外部連結徽章、媒體提及，以及排名帶來的社會證明。
JaredL 有兩次上線紀錄：Tidyread 在 2024 年拿到 Day #1，YouMind 在 2025 年拿到了 Day、Week 和 Month 排名。
投票準備：提前 3 個月開始
Product Hunt 的成功不在於上線當天，而在於之前的三個月。
團隊帳號養成： 每個團隊成員都需要把自己的 Product Hunt 帳號養到至少 bronze 或 silver 等級。新帳號投票看起來可疑，可能會被演算法打折扣。
自動化外聯工具： JaredL 的團隊做了一個插件，自動抓取同一天上線的團隊的社群連結，按成員數排序。這讓你可以精準地做互助投票的外聯——你幫他們的上線，他們幫你的。
LinkedIn PH 社群： 在 LinkedIn 的 Product Hunt 社群中直接發送連結請求，效果比公開發文求投票好得多。一對一私訊的轉化率顯著高於群發廣播。
獵人的優勢
頂級獵人可以大幅提升你的上線曝光：
- Chris Messina（Top 1 獵人）：「只要他來 hunt 你的產品，100% 會被 featured。」被 Messina hunt 基本上等於保證拿到 featured 徽章。
- Ben Lang 和 Ryan Hoover 也是值得爭取的高影響力獵人。
與頂級獵人建立關係應該在上線前幾個月就開始。上線當天才冷郵件聯繫很少有用。
投票交換的經濟學
Product Hunt 的上線運作在一個組織得出乎意料地好的互助生態中。實際操作是這樣的：
- 識別你目標日期的上線項目：用即將上線頁面或你自己的抓取工具
- 提前 1–2 週聯繫：對他們的產品給出真誠反饋，同時提議投票互助
- 上線當天協調：在共享群組中分享你的上線連結；回投他們的
- 上線後跟進：在對方的上線頁面留下有深度的評論——評論在排名演算法中有權重
在 Product Hunt 上贏的團隊不一定是產品最好的。而是上線 playbook 組織得最好的。把這個過程工程化——建抓取器、自動化外聯、追蹤投票交換——正是成長工程師擅長的那種系統性思維。
Featured 徽章就是一切
沒有 featured 徽章，你不可能獲得有意義的排名或賺到外部連結。這個決定在上線後的頭 4 小時內就會做出。
JaredL 的建議： 如果你在 4 小時內沒有被 featured，認真考慮放棄這次上線，之後準備更充分再重新來。一次沒有被 featured 的上線幾乎零價值，而且浪費了你產品的上線機會。
關於 Product Hunt 的真相
PH 的純直接流量對大多數產品來說可以忽略不計。真正的價值有三個：
- 外部連結徽章：你網站上的「Product of the Day」徽章提供一條高品質反向連結，改善 DR
- 媒體和電子報提及：排名靠前的上線經常觸發科技電子報和部落格合集的報導
- 社會證明：落地頁上的排名徽章提升所有其他流量來源的轉化率——付費廣告、自然搜索、推薦流量，當訪客看到「#1 Product of the Day」時，轉化率都會更高
Strategy 5: AI Automation Workflows
This is where the growth engineer's mindset and the frontend engineer's execution ability converge most powerfully.
Case Study: Nano Banana 2 Release
When Nano Banana 2 was about to launch, JaredL's team did not wait for the announcement and scramble to create content. They built an automation pipeline that executed the entire content creation and publication workflow before most competitors had even opened their laptops.
The sequence:
- Monitor Vertex AI every 30 minutes for new model release signals
- Detect new model ID → automatically test generation capability with predefined prompts
- Batch-generate images for 100 pre-prepared prompts the moment the model is confirmed live
- Auto-publish via CMS — content goes live within minutes of model release
- Auto-generate SEO blog by scraping X for early reactions and pulling Google documentation
- Auto-translate content to all supported languages
- Pre-drafted influencer marketing copy ready for immediate outreach
The entire pipeline — from model detection to published content in 16 languages — executes with minimal human intervention. A human reviews the output, but does not produce it.
This is what it looks like when engineering skills meet growth objectives: you are not just reacting faster than competitors, you are making human reaction speed irrelevant.
Why Engineers Have an Unfair Advantage Here
A marketer could conceptually design this pipeline. But building it — connecting APIs, handling error states, managing rate limits, structuring data flows, and deploying reliably — requires engineering skills. This is the moat that growth engineers have over traditional growth marketers: the ability to build the machine, not just operate it.
Most growth teams at startups cannot afford dedicated engineering resources for automation. A growth engineer who can both design the strategy and implement the automation is doing two jobs simultaneously, which is why the role is increasingly valued at early-stage companies.
Routine Automations
Beyond event-driven pipelines, the team runs daily automations that form the backbone of their content operation:
- Daily hot topic scraping: Monitor X, Reddit, and industry forums for trending AI topics
- Blog generation: Feed trending topics into the content pipeline for draft production
- Translation pipeline: Every new piece of content automatically spawns translations in all supported languages
- Social media monitoring: Track mentions, competitor launches, and emerging trends that could become the next collection site opportunity
The philosophy is simple: anything a growth engineer does more than three times should be automated. Your competitive advantage is not in doing the work — it is in designing the system that does the work.
策略五：AI 自動化工作流
這是成長工程師的思維方式和前端工程師的執行能力最強烈交匯的地方。
案例分析：Nano Banana 2 發佈
當 Nano Banana 2 即將上線時，JaredL 的團隊不是等公告出來再手忙腳亂地做內容。他們建了一條自動化 pipeline，在大多數競爭對手還沒打開電腦的時候，就已經完成了整個內容創建和發佈的工作流。
流程如下：
- 每 30 分鐘監控 Vertex AI 的新模型發佈信號
- 偵測到新模型 ID → 自動用預設的 prompts 測試生成能力
- 批次生成圖片——模型確認上線的瞬間，為 100 個預先準備的 prompts 生成圖片
- 通過 CMS 自動發佈——內容在模型發佈後幾分鐘內就上線
- 自動生成 SEO 部落格——抓取 X 上的早期反應和 Google 文件
- 自動翻譯內容到所有支援的語言
- 預先擬好的網紅行銷文案，可以立即發送
整條 pipeline——從模型偵測到 16 種語言的內容發佈——幾乎不需要人工介入。人負責審核產出，但不負責生產。
這就是工程技能遇上成長目標的樣子：你不只是比競爭對手反應更快，你是讓人類反應速度變得無關緊要。
為什麼工程師在這裡有不公平的優勢
一個行銷人員可以在概念上設計出這條 pipeline。但真正去建造它——連接 API、處理錯誤狀態、管理速率限制、組織數據流、可靠地部署——需要工程技能。這就是成長工程師相對於傳統成長行銷人員的護城河：能夠建造機器，而不只是操作它。
大多數新創公司的成長團隊負擔不起專門的工程資源來做自動化。一個既能設計策略又能實現自動化的成長工程師，等於同時在做兩個人的工作，這就是為什麼這個角色在早期公司越來越被重視。
日常自動化
除了事件驅動的 pipeline，團隊還跑著日常自動化任務，構成內容運營的骨幹：
- 每日熱門話題抓取：監控 X、Reddit 和行業論壇上的 AI 熱門話題
- 部落格生成：把熱門話題餵進內容 pipeline 生成草稿
- 翻譯 pipeline：每篇新內容自動生成所有支援語言的翻譯
- 社群媒體監控：追蹤提及、競品上線和可能成為下一個收錄站機會的新興趨勢
哲學很簡單：任何成長工程師做超過三次的事情都應該被自動化。你的競爭優勢不在於做這些工作——在於設計做這些工作的系統。
4 Recommendations for Engineers Wanting to Transition
JaredL's advice for frontend engineers considering the growth engineering path is specific and sequential. Each step builds on the previous one.
1. Build 1,000 Social Media Followers
This is not about vanity metrics. A thousand followers gives you two things: a distribution channel for product launches, and a feedback loop for validating ideas before you build them.
You do not need to become an influencer. You need enough of an audience that when you ship something, at least a few hundred people see it. That is the difference between launching into silence and launching with signal.
Platform recommendation: X (Twitter) for the tech and AI audience, or the platform where your target users actually spend time.
2. Launch a Monetizable Product and Get Your First Payment
This is the single most transformative experience for an engineer. Not because the revenue matters (it probably will not be much), but because the process teaches you:
- Demand validation: Is anyone willing to pay for this?
- Promotion: How do you get the word out?
- Conversion: What makes someone go from "interested" to "paying"?
These three skills are the foundation of growth engineering. You cannot learn them from tutorials. You learn them by watching real money (or the absence of it) respond to your decisions.
3. Master SEO and Target 10,000+ Monthly Clicks
SEO forces you to think about macro demand. Keyword research is essentially market research — you learn what people actually want by studying what they search for.
More practically, SEO provides a sustained traffic floor. Paid acquisition stops when the budget runs out. Social media posts decay within 48 hours. SEO compounds. A well-ranked page generates traffic for months or years.
For a frontend engineer, the technical execution of SEO is straightforward. The strategic layer — keyword selection, content architecture, topical authority building — is where the real learning happens.
4. Learn from Practitioners, Not Theorists
The growth engineering space is full of people selling courses about growth who have never actually grown anything. Seek out practitioners with verifiable results:
- Gefei (@gefei55): Deep, specific SEO case studies with real numbers
- Zimu (@CoderJeffLee): Multi-channel growth strategies — shows how to combine SEO, social, and product-led growth
- Daqi (@daqi029): Conversion funnel optimization — the art of turning traffic into paying users
- Lenny Rachitsky: The gold standard for growth interviews and frameworks (newsletter + podcast). Essential reading for understanding how top companies think about growth
- Elena Verna (Lovable growth lead): Enterprise and product-led growth expertise. Her writing on the differences between sales-led and product-led growth is particularly valuable
The common thread: these are people who share what actually worked, not what theoretically should work.
The Sequence Matters
These four steps are not arbitrary. They are ordered by the feedback loops they create:
- 1,000 followers gives you a distribution channel → which you need for step 2
- First payment teaches you conversion → which gives you the mental model for step 3
- 10,000 SEO clicks proves you understand demand → which makes you credible for step 4
- Learning from practitioners accelerates everything → but only after you have enough context to understand what they are teaching
Skipping ahead rarely works. An engineer who tries to learn SEO without ever having launched a product will optimize for vanity keywords. An engineer who launches a product without any audience will hear nothing but silence. The sequence is the strategy.
給想轉型的工程師的 4 個建議
JaredL 對考慮走成長工程路線的前端工程師的建議是具體且有順序的。每一步都建立在上一步的基礎上。
1. 累積 1,000 個社群媒體粉絲
這不是為了虛榮指標。一千個粉絲給你兩樣東西：一個產品上線的分發渠道，以及一個在動手開發之前驗證想法的回饋迴路。
你不需要變成網紅。你需要的是足夠的受眾，讓你每次出貨時至少有幾百個人看到。這就是「上線後一片寂靜」和「上線後有信號回饋」的差別。
平台建議：X（Twitter）面向技術和 AI 受眾，或者你的目標用戶實際上待的平台。
2. 上線一個可變現的產品，拿到第一筆付款
這是工程師最具變革性的一次經歷。不是因為收入有多重要（可能不會太多），而是因為這個過程教會你：
- 需求驗證：有沒有人願意為此付費？
- 推廣：怎麼讓消息傳出去？
- 轉化：是什麼讓人從「有興趣」變成「付錢」？
這三個技能是成長工程的基礎。你從教程裡學不會。你得通過看著真金白銀（或者它的缺席）回應你的決策來學。
3. 掌握 SEO，目標月點擊量 10,000+
SEO 逼你思考宏觀需求。關鍵字研究本質上就是市場研究——你通過研究人們搜索什麼來了解他們真正想要什麼。
更實際地說，SEO 提供了一個持續的流量底線。付費獲客在預算花完時就停了。社群媒體貼文在 48 小時內衰減。SEO 是複利。一個排名好的頁面能產生數月甚至數年的流量。
對前端工程師來說，SEO 的技術執行很直接。策略層——關鍵字選擇、內容架構、主題權威建設——才是真正的學習發生的地方。
4. 向實戰者學習，不要向理論家學習
成長工程的圈子裡到處都是賣成長課程但自己從來沒做出過成長的人。找有可驗證成果的實戰者：
- Gefei（@gefei55）：深入、具體的 SEO 案例研究，有真實數據
- Zimu（@CoderJeffLee）：多渠道成長策略——展示如何結合 SEO、社群和產品驅動成長
- Daqi（@daqi029）：轉化漏斗優化——把流量變成付費用戶的藝術
- Lenny Rachitsky：成長訪談和框架的黃金標準（電子報 + 播客）。理解頂級公司如何思考成長的必讀材料
- Elena Verna（Lovable 成長負責人）：企業級和產品驅動成長的專業知識。她關於銷售驅動和產品驅動成長之間差異的文章特別有價值
共同的線索：這些人分享的是實際有效的東西，不是理論上應該有效的東西。
順序很重要
這四步不是隨便排的。它們按照創建的回饋迴路來排序：
- 1,000 粉絲給你一個分發渠道 → 你需要這個來做第 2 步
- 第一筆付款教會你轉化 → 這給你做第 3 步的心智模型
- 10,000 SEO 點擊證明你懂需求 → 這讓你在做第 4 步時有足夠的語境
- 向實戰者學習加速一切 → 但只有在你有足夠的背景知識來理解他們教的東西之後
跳步很少有用。一個從沒上線過產品就試圖學 SEO 的工程師，會為虛榮關鍵字做優化。一個沒有任何受眾就上線產品的工程師，只會聽到一片沉默。順序就是策略。
The Core Philosophy
The frontend engineer career path is not dying. What is dying is the safety of treating "frontend" as a complete identity.
Every technology stack eventually ages out. The frameworks you mastered three years ago are already being replaced. The ones you are learning now will be obsolete within five years. This is not a crisis — it is the normal rhythm of the industry.
What does not age out is the underlying engineering mindset: the ability to decompose complex problems, form hypotheses, test them against data, iterate rapidly, and ship. These skills transfer across every domain — product engineering, growth engineering, data engineering, or whatever the next category turns out to be.
JaredL's trajectory illustrates this perfectly. He did not abandon his frontend skills. He added a new objective function on top of them. The React code, the CSS precision, the browser performance knowledge — all of it is still there, still useful, still deployed daily. It is just in service of driving traffic instead of shipping features.
The transition from frontend to growth is not a career change. It is a career expansion. And for engineers willing to make it, the upside is significant: you become one of the rare people who can both build the product and fill it with users.
The market is signaling this clearly. At early-stage startups, the most valuable engineer is not the one who writes the cleanest code — it is the one who writes code that brings in users. At later-stage companies, growth engineering teams are among the highest-impact (and highest-compensated) engineering functions. The demand for engineers who understand both systems and funnels will only increase as AI makes content production cheaper and distribution becomes the primary bottleneck.
JaredL's path from NetEase to YouMind is one example. But the underlying pattern is universal: take the engineering mindset you already have, point it at the metrics that determine whether your product lives or dies, and build systems that compound over time.
"Frontend isn't dying — only the comfort of 'only doing frontend' is dying. Tech stacks age, but problem-solving ability, data-driven hypothesis testing, and rapid iteration are engineer strengths regardless of domain."
核心理念
前端工程師的職涯路徑沒有在死亡。正在死亡的是把「前端」當作完整身份的那份安全感。
每個技術棧最終都會過時。你三年前精通的框架已經在被取代。你現在正在學的那些，五年內也會過時。這不是危機——這是行業的正常節奏。
不會過時的是底層的工程師思維：分解複雜問題的能力、形成假設、用數據驗證、快速迭代、交付。這些技能在每個領域都能遷移——產品工程、成長工程、數據工程，或者未來會出現的任何新類別。
JaredL 的軌跡完美說明了這一點。他沒有拋棄前端技能。他在上面加了一個新的目標函數。React 程式碼、CSS 的精確度、瀏覽器效能的知識——全部都還在，全部仍然有用，每天都在用。只是從服務於交付功能，變成了服務於驅動流量。
從前端到成長的轉型不是轉行，是職涯的擴展。而對願意做出這個轉變的工程師來說，回報是顯著的：你會變成少數既能建造產品、又能讓產品裝滿用戶的人。
市場的信號已經很清楚了。在早期新創公司，最有價值的工程師不是寫最乾淨程式碼的人——是寫能帶來用戶的程式碼的人。在後期公司，成長工程團隊是影響力最大（薪酬也最高）的工程職能之一。隨著 AI 讓內容生產越來越便宜、分發成為主要瓶頸，市場對同時理解系統和漏斗的工程師的需求只會增加。
JaredL 從網易到 YouMind 的路徑是一個例子。但底層的模式是普適的：拿你已經有的工程師思維，對準決定你的產品生死的指標，建造隨時間複利增長的系統。
「前端死的不是前端，是『只寫前端』這件事的安全感。技術棧會老化，但解決問題的能力、數據驅動的假設驗證、快速迭代——這些才是工程師不分領域的核心優勢。」