The Complete SaaS Conversion Guide

Landing Pages, Pricing Strategy, and CRO

SaaS 轉換率優化完全指南

Landing Page、定價策略與 CRO 實戰

---

"People don't buy products. They buy better versions of themselves." — Samuel Hulick

「人們買的不是產品，而是更好版本的自己。」—— Samuel Hulick

---

Why This Guide Exists

Most SaaS founders build a product, slap up a landing page, pick a price that "feels right," and wonder why conversions are stuck at 2%.

The problem is not your product. It is the stack of decisions between your product and the user's wallet: positioning, messaging, page design, pricing architecture, trial mechanics, and ongoing optimization.

This guide synthesizes the best publicly available resources from the Marketing for Founders collection into a practical playbook. It is organized into four pillars:

1. Landing Pages, Messaging & Positioning — what to say and how to frame it
2. Pricing Strategies — how to charge, how much, and when to change
3. Conversion Rate Optimization (CRO) — systematic experimentation that compounds
4. User Research — the foundation that makes everything above work

The order matters. Start from the bottom (research), then build up.

為什麼寫這篇指南

大多數 SaaS 創辦人做完產品，隨便架個 Landing Page，定個「感覺差不多」的價格，然後疑惑為什麼轉換率卡在 2%。

問題不在產品本身，而在產品和用戶錢包之間的一連串決策：定位、訊息傳達、頁面設計、定價架構、試用機制、持續優化。

本指南整理了 Marketing for Founders 合輯中最實用的公開資源，彙整成一本可操作的 playbook。分為四大支柱：

1. Landing Page、訊息傳達與定位 —— 說什麼、怎麼說
2. 定價策略 —— 怎麼收費、收多少、什麼時候調整
3. 轉換率優化（CRO） —— 可系統化、可複利的實驗方法
4. 使用者研究 —— 讓以上三者成立的根基

順序很重要。從最底層（研究）往上建構。

---

Part 1: Landing Pages, Messaging & Positioning

The Golden Rule: Positioning → Messaging → Design

This is the single most important sequence in SaaS marketing. Get it backwards and you waste months polishing pixels on a page that says the wrong thing.

Step	Question it answers	Output
Positioning	Who is this for, and why should they care?	Category, competitive alternative, key differentiator
Messaging	What specific words do we use?	Headlines, subheads, bullet points, CTAs
Design	How do we present it visually?	Layout, hierarchy, social proof placement

If you redesign your homepage and conversions do not improve, the problem is almost certainly in step 1 or 2, not step 3.

Four Homepage Frameworks

There is no single "correct" homepage structure. Here are four proven approaches, each suited to different contexts:

Framework 1: Problem-Agitation-Solution (PAS)

Best for: Products solving a painful, well-understood problem.

1. Name the problem in the headline
2. Agitate — show the cost of not solving it
3. Present your product as the solution
4. Social proof
5. CTA

Framework 2: Before/After/Bridge

Best for: Products where the transformation is more compelling than the pain.

1. Show the "before" state (the user's current world)
2. Show the "after" state (the better world)
3. Your product is the bridge between them
4. How it works (3-step simplification)
5. Proof + CTA

Framework 3: The Demo-First Page

Best for: Products with strong visual UX (design tools, dashboards, dev tools).

1. One-line headline
2. Immediately show the product in action (interactive demo, video, or animated walkthrough)
3. Feature breakdown below
4. Social proof + CTA

Framework 4: The Competitor Comparison Page

Best for: Entering a crowded market where prospects already use an alternative.

1. Headline: "[Your Product] vs. [Competitor]" or "The [Competitor] alternative for [specific audience]"
2. Side-by-side comparison table
3. Key differentiators in detail
4. Migration story / switching cost reduction
5. CTA

The SaaS Headline Playbook

Your headline has roughly 3 seconds to earn the next 30 seconds of attention. Here are patterns that consistently convert:

Pattern	Example	Why it works
Outcome-first	"Ship 10x faster without breaking prod"	Leads with what the user gets
Negative framing	"Stop wasting 5 hours a week on manual deploys"	Loss aversion is powerful
Specificity	"Reduce page load time by 42% in one line of code"	Numbers build credibility
Category creation	"The first AI-native database"	Positions you as the leader of a new space
For [audience]	"Error tracking built for solo developers"	Instant relevance filter

Headlines to avoid:

• "The all-in-one platform for..." (says nothing specific)
• "Supercharge your workflow" (vague, overused)
• "Welcome to [Product Name]" (wastes the most valuable real estate)
• Anything that requires the subhead to make sense

Positioning Frameworks

The Definitive Guide Approach (April Dunford Style)

April Dunford's positioning framework asks five questions in sequence:

1. Competitive alternatives — What would customers use if you did not exist? (Not just direct competitors — spreadsheets, manual processes, and "do nothing" count.)
2. Unique attributes — What do you have that alternatives lack?
3. Value — What does that attribute enable for the customer?
4. Target customer — Who cares most about that value?
5. Market category — What market context makes your value obvious?

The output is a positioning statement that drives every word on your site.

The Quickstart Approach

If you need something faster, answer these three questions:

1. What is the one thing you do better than anyone else?
2. Who specifically needs that one thing?
3. What is the alternative they are using right now?

Your homepage headline should communicate #1 to #2 with reference to #3.

Positioning for Developer Tools

Developer audiences are allergically resistant to marketing speak. Specific rules:

• Show, do not tell. Code snippets > adjectives. A 3-line integration example > a paragraph about "seamless integration."
• Be technically precise. "Sub-50ms p99 latency" > "blazing fast."
• Respect intelligence. Developers will read your docs before your marketing page. Make docs a first-class marketing asset.
• Community signals matter more than logos. GitHub stars, open-source contributions, Discord community size, Stack Overflow answers.
• Pricing transparency is non-negotiable. "Contact sales" is a bounce for most developers.

Social Proof: The Trust Stack

Not all social proof is equal. Build a trust stack from weakest to strongest:

1. Vanity metrics — "10,000+ users" (weak but better than nothing)
2. Logos — Recognizable customer logos (mid-tier)
3. Quotes — Named person + title + company + headshot (strong)
4. Case studies — Named company + specific metrics + story (very strong)
5. Video testimonials — Hardest to fake, highest trust (strongest)

Rules for effective testimonials:

• Always include the person's full name, title, and company
• Specifics beat generics: "Saved us 12 hours per sprint" > "Great product"
• Place social proof near CTAs — that is where doubt peaks
• Match the testimonial to the objection: if price is the concern, feature a quote about ROI

Product Pages Guide

For multi-product SaaS or feature-rich tools, individual product pages convert better than cramming everything onto the homepage.

Structure for a high-converting product page:

1. Hero: Feature name + one-sentence value prop + screenshot/demo
2. Problem context: "If you have ever [painful situation], you know..."
3. How it works: 3-4 steps with visuals
4. Key capabilities: Feature grid or accordion
5. Integration: Where it fits in the user's existing stack
6. Social proof: Testimonials specific to this feature
7. CTA: Trial or demo, specific to this feature

第一部分：Landing Page、訊息傳達與定位

黃金法則：定位 → 訊息 → 設計

這是 SaaS 行銷中最重要的一個順序。搞反了，你會浪費好幾個月在打磨一個「說錯話」的頁面。

步驟	回答的問題	產出
定位	這是給誰用的？他們為什麼要在乎？	品類、競爭替代方案、核心差異化
訊息	我們具體用什麼文字？	標題、副標、要點、CTA
設計	視覺上怎麼呈現？	版面、層級、社會證明的位置

如果你重新設計了首頁但轉換率沒有提升，問題幾乎肯定出在第 1 步或第 2 步，而不是第 3 步。

四種首頁框架

沒有唯一「正確」的首頁結構。以下四種經過驗證的方法，各適合不同情境：

框架一：問題-激化-解方（PAS）

最適合：解決一個痛感明確、使用者都懂的問題。

1. 在標題直接點名問題
2. 激化——展示不解決的代價
3. 把產品當作解方呈現
4. 社會證明
5. CTA

框架二：之前 / 之後 / 橋樑

最適合：轉變本身比痛點更有吸引力的產品。

1. 展示「之前」的狀態（使用者的現狀）
2. 展示「之後」的狀態（更好的世界）
3. 你的產品就是兩者之間的橋樑
4. 運作方式（三步驟簡化說明）
5. 證明 + CTA

框架三：Demo 優先頁面

最適合：視覺 UX 強烈的產品（設計工具、儀表板、開發者工具）。

1. 一行標題
2. 立即展示產品運作（互動 demo、影片或動態導覽）
3. 功能拆解放在下方
4. 社會證明 + CTA

框架四：競品比較頁面

最適合：進入一個擁擠市場，目標用戶已經在用替代方案。

1. 標題：「[你的產品] vs. [競品]」或「專為 [特定族群] 打造的 [競品] 替代方案」
2. 並排比較表
3. 關鍵差異的詳細說明
4. 遷移故事 / 降低轉換成本
5. CTA

SaaS 標題撰寫手冊

你的標題大概有 3 秒鐘的時間來換取接下來 30 秒的注意力。以下是持續有效的轉換模式：

模式	範例	為什麼有效
結果優先	"Ship 10x faster without breaking prod"	直接告訴用戶他能得到什麼
負面框架	"別再每週浪費 5 小時手動部署"	損失趨避效應強大
具體數字	"一行程式碼，頁面載入速度提升 42%"	數字建立可信度
品類創造	"第一個 AI 原生資料庫"	把自己定位為新領域的領導者
指定受眾	"專為獨立開發者打造的錯誤追蹤工具"	瞬間篩選出目標讀者

應避免的標題：

• 「一站式平台，滿足所有需求」（什麼都沒說）
• 「Supercharge your workflow」（模糊、用爛了）
• 「歡迎來到 [產品名稱]」（浪費最值錢的版位）
• 任何需要靠副標才能看懂的標題

定位框架

完整定位法（April Dunford 風格）

April Dunford 的定位框架，依序問五個問題：

1. 競爭替代方案 —— 如果你不存在，客戶會用什麼？（不只是直接競品——Excel、手動流程、「什麼都不做」都算。）
2. 獨特屬性 —— 你有什麼是替代方案沒有的？
3. 價值 —— 那個屬性為客戶帶來什麼？
4. 目標客戶 —— 誰最在乎那個價值？
5. 市場品類 —— 什麼市場脈絡能讓你的價值一目了然？

產出是一個定位聲明，驅動網站上的每一個字。

快速定位法

如果你需要更快的方法，回答三個問題：

1. 你做得比誰都好的那一件事是什麼？
2. 誰最需要那一件事？
3. 他們現在用的替代方案是什麼？

首頁標題就是用 #3 的語境，對 #2 傳達 #1。

開發者工具的定位

開發者對行銷話術有天生的免疫力。特殊規則：

• 展示，不要說教。 程式碼片段 > 形容詞。三行整合範例 > 一段「無縫整合」的廢話。
• 技術上要精準。 「Sub-50ms p99 latency」 > 「超級快」。
• 尊重智商。 開發者會先看你的文件，才看你的行銷頁面。把文件當成第一級行銷資產。
• 社群訊號比 logo 牆重要。 GitHub stars、開源貢獻、Discord 社群規模、Stack Overflow 回答數。
• 定價透明是底線。 「聯絡業務」會讓大多數開發者直接跳出。

社會證明：信任階梯

並非所有社會證明都一樣有效。由弱到強建構你的信任階梯：

1. 虛榮指標 —— 「10,000+ 用戶」（弱，但比沒有好）
2. Logo 牆 —— 知名客戶 logo（中等）
3. 引言 —— 具名人物 + 職稱 + 公司 + 大頭照（強）
4. 案例研究 —— 具名公司 + 具體數據 + 故事（很強）
5. 影片見證 —— 最難造假，信任度最高（最強）

有效見證的規則：

• 一定要附上全名、職稱、公司
• 具體勝過空泛：「每個 sprint 省下 12 小時」 > 「很棒的產品」
• 把社會證明放在 CTA 附近——那是疑慮最高峰的地方
• 見證要對應異議：如果顧慮是價格，就放 ROI 相關的引言

產品頁面指南

對多產品 SaaS 或功能豐富的工具來說，獨立的產品頁面比把所有東西塞進首頁的轉換率更好。

高轉換率產品頁面的結構：

1. Hero： 功能名稱 + 一句話價值主張 + 截圖或 demo
2. 問題脈絡： 「如果你曾經 [痛苦場景]，你一定懂……」
3. 運作方式： 3-4 個步驟搭配視覺化
4. 核心能力： 功能格狀表或手風琴式展開
5. 整合： 它如何融入用戶現有的技術棧
6. 社會證明： 針對此功能的見證
7. CTA： 針對此功能的試用或 demo

---

Part 2: Pricing Strategies

The WTF-to-WTP Framework

"WTF" is exactly what most founders say when they realize they have been guessing their price. WTP stands for Willingness to Pay — the maximum a customer would pay before choosing an alternative.

The framework:

1. Discover WTP through direct customer conversations (see Part 4)
2. Segment WTP — different customer segments will have vastly different WTP. An enterprise team might pay $500/month for what a solo founder values at $29/month.
3. Design tiers around WTP boundaries — your pricing tiers should map to natural WTP clusters, not to feature bundles you invented in a meeting room.
4. Capture WTP — set prices at 70-80% of discovered WTP to leave room for perceived value while maximizing revenue.

B2B SaaS Pricing Guide for Early Stage

When you have fewer than 100 customers, pricing is not a math problem. It is a conversation.

Early-stage pricing principles:

• Price higher than you think. Most founders underprice by 2-5x. If nobody pushes back on your price, it is too low.
• Charge from day one. Free users give you vanity metrics. Paying users give you signal.
• Three tiers maximum. Starter, Pro, Enterprise. That is it.
• Annual discount = 2 months free. This is industry standard and customers expect it.
• Grandfather early customers. Lock their price. They took a risk on you.

The pricing page checklist:

• [ ] Highlight the recommended tier visually
• [ ] Use per-seat or usage-based pricing if your value scales with team size or consumption
• [ ] Show annual savings explicitly ("Save 20%")
• [ ] Include a FAQ section — price objections disguised as questions are conversion gold
• [ ] Add a "Book a call" option for Enterprise — this segment expects it

How to A/B Test Pricing

You cannot A/B test pricing the way you test button colors. Showing different prices to different visitors simultaneously creates trust disasters if discovered.

Safe approaches:

1. Time-based testing. Run Price A for two weeks, then Price B for two weeks. Compare conversion rates and revenue per visitor.
2. Segment-based testing. Test different prices in different geographies or acquisition channels.
3. Feature packaging tests. Keep the price the same but change what is included in each tier. This is functionally a price test with plausible deniability.
4. New visitor only. Only test on net-new visitors who have no existing price anchor.

What to measure:

• Revenue per visitor (not just conversion rate — a 50% price increase that drops conversion 10% is still a huge win)
• Trial-to-paid conversion rate at each price point
• Churn at 30/60/90 days — low price attracts low-quality customers who churn faster

Revenue You Are Leaving on the Table

Common revenue leaks in SaaS:

Leak	Fix
No annual plan option	Add it. Instant 15-25% cash flow improvement.
Flat pricing (no tiers)	Segment by usage/seats/features. Capture more WTP from power users.
Free tier is too generous	Tighten limits. If free users never hit the paywall, they never convert.
No expansion revenue	Add usage-based pricing or per-seat pricing that grows with the customer.
Discounting on every deal	Set a discount policy. Max 20%, only for annual, only for startups.
No price increase in 2+ years	Raise prices. Existing customers stay; new customers pay more.

Having Pricing Conversations with Customers

The Van Westendorp Price Sensitivity Meter uses four questions:

1. At what price would this be so cheap you would question its quality?
2. At what price is this a bargain — a great deal for the money?
3. At what price is this getting expensive — you would still buy but would hesitate?
4. At what price is this too expensive — you would not buy regardless?

Plot the answers on a chart. The intersection points define your acceptable price range.

Rules for pricing conversations:

• Never ask "What would you pay?" in isolation — people default to the lowest number
• Always anchor to a competitor or alternative first
• Ask about value before asking about price
• Talk to people who are already paying for something similar

Reverse Trials: The Freemium Alternative

A reverse trial gives users full access to your premium tier for a limited time (typically 14 days), then downgrades them to a free tier — the opposite of traditional freemium where users start limited.

Why reverse trials work:

• Users experience the full value before being asked to pay
• Loss aversion kicks in — losing features feels worse than never having them
• You collect usage data on which premium features drive retention
• Users self-segment: those who need premium features convert; those who do not stay on free (and may convert later)

Implementation checklist:

1. Give full access for 14 days (no credit card required)
2. Send usage-based emails during the trial: "You used [premium feature] 12 times this week"
3. At day 12, show what they will lose
4. Downgrade gracefully — do not delete data or break workflows
5. Offer a discounted "win-back" at day 30

Free Trial Fundamentals for PLG

Product-Led Growth relies on the product itself as the primary acquisition and conversion engine.

Key decisions:

Decision	Option A	Option B
Credit card	Require upfront	No card required
Duration	7 days	14 days
Scope	Full access	Limited features
Onboarding	Self-serve	Guided

The data says:

• No credit card required = 2-5x more trial starts, but lower trial-to-paid rates
• 14-day trials outperform 7-day trials for complex products
• For simple tools, 7-day trials create urgency without frustration
• Activation milestones matter more than trial length — if a user hits your "aha moment" on day 2, they will convert whether the trial is 7 or 30 days

The activation framework:

1. Define your activation metric (e.g., "created first project," "invited a teammate," "ran first query")
2. Track time-to-activation for converting vs. non-converting users
3. Optimize onboarding to reduce time-to-activation
4. Trigger conversion prompts after activation, not after a calendar deadline

14 Tactics for Selling Annual Plans

Annual plans improve cash flow, reduce churn, and increase LTV. Here are 14 ways to push adoption:

1. Default the pricing toggle to "Annual"
2. Show the annual savings as a dollar amount, not just a percentage
3. Offer 2 months free (the standard discount)
4. Lock in current pricing for annual customers before a price increase
5. Add a bonus feature or perk exclusive to annual plans
6. Send a mid-trial email with the annual option and its savings
7. At the monthly-to-annual upgrade moment, show how much they have "overpaid" so far
8. Offer quarterly as a middle option (makes annual look even better)
9. Give annual customers priority support
10. Run an annual promotion during renewal season (Q4)
11. Show a comparison: "Monthly = $588/year vs. Annual = $468/year"
12. Add social proof: "80% of our customers choose annual"
13. Provide an annual-only onboarding session or workshop
14. For enterprise: offer net-30 payment terms on annual invoices

第二部分：定價策略

從 WTF 到 WTP 框架

「WTF」正是大多數創辦人發現自己一直在猜價格時的反應。WTP 指的是 Willingness to Pay（支付意願） —— 客戶在轉向替代方案之前願意支付的最高金額。

框架流程：

1. 透過客戶對話發掘 WTP（見第四部分）
2. 分群 WTP —— 不同客戶群的 WTP 差異極大。企業團隊可能願意為某功能付 $500/月，但獨立創辦人只覺得值 $29/月。
3. 根據 WTP 邊界設計方案 —— 定價方案應該對應自然的 WTP 聚落，而非你們在會議室裡拍腦袋的功能組合。
4. 擷取 WTP —— 把價格定在發掘到的 WTP 的 70-80%，留下感知價值的空間，同時最大化營收。

早期 B2B SaaS 定價指南

客戶不到 100 個的時候，定價不是數學題，而是一場對話。

早期定價原則：

• 定價要比你以為的高。 大多數創辦人定價偏低 2-5 倍。如果沒有人對你的價格有意見，那就太低了。
• 第一天就收費。 免費用戶給你虛榮指標，付費用戶給你真實訊號。
• 最多三個方案。 Starter、Pro、Enterprise，就這樣。
• 年繳折扣 = 免費兩個月。 這是業界標準，客戶也預期如此。
• 保護早期客戶的價格。 鎖定他們的費率。他們是冒著風險選了你。

定價頁面檢查清單：

• [ ] 視覺上突顯推薦方案
• [ ] 如果價值隨團隊規模或用量增長，使用 per-seat 或用量計費
• [ ] 明確顯示年繳省下多少（「省 20%」）
• [ ] 加上 FAQ 區塊——偽裝成問題的價格異議是轉換的黃金地段
• [ ] Enterprise 要有「預約通話」選項——這個族群預期這樣

如何 A/B 測試定價

你不能用測試按鈕顏色的方式來 A/B 測試定價。同時對不同訪客顯示不同價格，一旦被發現就是信任災難。

安全的做法：

1. 時間區間測試。 先跑 Price A 兩週，再跑 Price B 兩週。比較轉換率和每位訪客營收。
2. 分群測試。 在不同地區或不同獲客管道測試不同價格。
3. 功能組合測試。 價格不變，但改變每個方案包含的功能。這實質上是價格測試，但有合理的解釋空間。
4. 僅限新訪客。 只對沒有既有價格錨點的全新訪客測試。

衡量什麼：

• 每位訪客營收（不只是轉換率——漲價 50% 但轉換率只掉 10%，仍然是巨大的勝利）
• 試用轉付費率（在每個價格點）
• 30/60/90 天流失率 —— 低價吸引低品質客戶，流失更快

你正在漏掉的營收

SaaS 常見的營收漏洞：

漏洞	修復方式
沒有年繳選項	加上去。現金流立即改善 15-25%。
單一定價（無分級）	按用量/人數/功能分級。從重度用戶身上擷取更多 WTP。
免費方案太大方	收緊限制。免費用戶永遠碰不到付費牆，就永遠不會轉換。
沒有擴展營收	加入用量計費或 per-seat 定價，讓營收隨客戶成長。
每筆交易都打折	設定折扣政策。上限 20%，僅限年繳，僅限新創。
超過兩年沒有漲價	漲價。既有客戶留下來，新客戶付更多。

與客戶的定價對話

Van Westendorp 價格敏感度模型用四個問題：

1. 在什麼價格下，你會覺得太便宜而懷疑品質？
2. 在什麼價格下，你會覺得是個划算的交易？
3. 在什麼價格下，你會覺得開始有點貴——還是會買，但會猶豫？
4. 在什麼價格下，你會覺得太貴了——不管怎樣都不會買？

把答案畫在圖上。交叉點定義了你的可接受價格區間。

定價對話的規則：

• 永遠不要單獨問「你願意付多少？」——人們會預設說出最低的數字
• 先錨定一個競品或替代方案的價格
• 先問價值，再問價格
• 去問已經在為類似方案付費的人

逆向試用：Freemium 的替代方案

逆向試用讓用戶在有限期間（通常 14 天）內使用完整的付費功能，然後降級到免費方案——跟傳統 freemium 的「先限制再升級」正好相反。

為什麼逆向試用有效：

• 用戶在被要求付費之前就體驗了完整價值
• 損失趨避效應啟動——失去功能的痛苦比從未擁有更強烈
• 你能收集到哪些付費功能驅動留存的使用數據
• 用戶自我分群：需要付費功能的人轉換，不需要的人留在免費（之後可能再轉換）

實施清單：

1. 提供 14 天完整存取（不需信用卡）
2. 試用期間發送基於使用量的 email：「你這週用了 [付費功能] 12 次」
3. 第 12 天展示他們即將失去什麼
4. 優雅降級——不要刪除數據或打斷工作流
5. 第 30 天提供折扣「回歸」方案

PLG 免費試用基本功

Product-Led Growth 依賴產品本身作為主要的獲客和轉換引擎。

關鍵決策：

決策	選項 A	選項 B
信用卡	先付卡號	免卡號
時間	7 天	14 天
範圍	完整功能	有限功能
引導	自助式	有人帶

數據怎麼說：

• 免卡號 = 試用開始數多 2-5 倍，但試用轉付費率較低
• 複雜產品用 14 天試用表現優於 7 天
• 簡單工具用 7 天試用能製造緊迫感又不至於挫折
• 啟動里程碑比試用長度更重要——如果用戶在第 2 天就到達「Aha 時刻」，不管試用是 7 天還是 30 天都會轉換

啟動框架：

1. 定義你的啟動指標（例：「建立第一個專案」「邀請第一個隊友」「執行第一次查詢」）
2. 追蹤有轉換 vs. 未轉換用戶的「到達啟動時間」
3. 優化引導流程以縮短到達啟動時間
4. 在啟動之後觸發轉換提示，而非在日曆到期時

14 個提升年繳率的戰術

年繳方案改善現金流、降低流失率、提升 LTV。以下 14 種方法推動採用率：

1. 定價開關預設為「年繳」
2. 用金額而非百分比顯示年繳省下多少
3. 提供免費兩個月（標準折扣）
4. 漲價前鎖定年繳客戶的現有價格
5. 年繳方案加入獨享功能或福利
6. 試用期中段寄一封 email 介紹年繳選項和節省金額
7. 月繳轉年繳的升級時刻，顯示他們到目前為止「多付了多少」
8. 提供季繳作為中間選項（讓年繳看起來更划算）
9. 年繳客戶享有優先支援
10. 在續約季（Q4）推年繳促銷
11. 做比較：「月繳 = $588/年 vs. 年繳 = $468/年」
12. 加入社會證明：「80% 的客戶選擇年繳」
13. 提供年繳專屬的 onboarding 課程或工作坊
14. 對企業客戶：年繳發票提供 net-30 付款條件

---

Part 3: Conversion Rate Optimization (CRO)

Common Website Mistakes That Kill Conversion

After analyzing thousands of SaaS websites, these are the most frequent killers:

1. No clear CTA above the fold. If a visitor cannot figure out what to do within 5 seconds, you lose them.
2. Too many CTAs competing. "Start free trial" + "Book a demo" + "Watch video" + "Read docs" = decision paralysis.
3. Jargon-heavy copy. You know what "AI-powered automated orchestration layer" means. Your customer does not.
4. No loading speed optimization. Every additional second of load time drops conversion by ~7%.
5. Mobile is an afterthought. 50%+ of traffic is mobile. If your pricing page is unreadable on a phone, half your visitors bounce.
6. Navigation has too many options. Every nav link is an exit. Limit to 5-7 items maximum.
7. Social proof is buried. If testimonials are only on a dedicated "Testimonials" page that nobody visits, they are not helping.
8. The "wall of text" hero section. If your above-the-fold area has more than 30 words, you are losing people.

Lessons from 20,000 A/B Experiments

The most counterintuitive findings from large-scale experimentation:

What usually works:

• Removing fields from forms (each field removed = ~5-10% more completions)
• Making the CTA button larger and higher contrast
• Adding urgency (real urgency, not fake countdown timers)
• Reducing choices on the page
• Adding inline validation to forms
• Placing social proof directly adjacent to the CTA

What usually does not work (or backfires):

• Changing button colors without changing anything else
• Adding chatbots to landing pages (increases engagement, but often decreases conversion)
• Popup modals on first visit (they work on return visits, but annoy first-timers)
• Clever or cute CTA copy ("Let's do this!" vs. "Start free trial" — the boring version usually wins)
• Redesigning the entire page at once (you cannot learn what worked)

The 80/20 of CRO:

• 80% of conversion impact comes from copy and offer (messaging, pricing, positioning)
• 15% comes from layout and UX (page structure, form design, mobile optimization)
• 5% comes from visual design (colors, fonts, imagery)

Most teams spend 80% of their time on the 5% category.

CTA Optimization: Why "Book a Demo" Underperforms

"Book a Demo" is the default CTA for B2B SaaS. It is also one of the worst-performing.

Why it fails:

• It requires a commitment (scheduling time) before showing value
• "Demo" implies a sales pitch, which triggers resistance
• It creates friction: calendar scheduling, timezone coordination, waiting

Higher-converting alternatives:

Instead of...	Try...	Why
"Book a Demo"	"See it in action" (with an interactive product tour)	No commitment required
"Book a Demo"	"Try it free"	Lower friction, immediate value
"Contact Sales"	"Get a custom quote"	Implies value, not a sales call
"Sign Up"	"Start building" or "Create your first [outcome]"	Outcome-oriented
"Learn More"	"See how [Company] saved 40%"	Specific and curiosity-driven

When "Book a Demo" IS appropriate:

• Enterprise sales ($10k+ ACV) where buyers expect a guided process
• Complex products that genuinely require explanation
• When paired with a secondary, lower-friction CTA ("Or try it free")

The Paywall Redesign: 23% More Signups

A single change to a paywall screen increased signups by 23%. The change:

Before: The paywall showed the plan name, price, and a list of features.

After: The paywall showed what the user had already accomplished in the free tier, then framed the paid tier as "Continue your progress."

The psychological mechanism: endowed progress effect. When people feel they have already made progress toward a goal, they are significantly more likely to continue. Showing users what they built (projects created, data imported, integrations set up) makes the paid tier feel like the natural next step, not a new commitment.

How to apply this:

1. Track user actions during free usage or trial
2. At the conversion point, summarize their activity: "You have created 3 projects, invited 2 teammates, and processed 1,847 records"
3. Frame the upgrade as continuation: "Keep going with Pro" instead of "Upgrade to Pro"
4. Show what they would lose (loss aversion) alongside what they would gain

The Blurry Background Effect: 94% More Conversions

One of the most surprising CRO findings: using a blurred background image behind a signup form increased conversions by 94% compared to a sharp image or solid color.

Why it works:

• A blurred background suggests depth and context without competing for attention
• It reduces visual noise around the form
• It creates a "spotlight" effect on the form itself
• The user's eye has nowhere to wander

Application principles:

• Use blur behind any high-priority form or CTA area
• The blur should be 15-25px Gaussian — enough to be obviously blurred, not just slightly out of focus
• The underlying image should suggest the product or its context
• Combine with a semi-transparent dark or light overlay for contrast

The Small Changes Principle

CRO is not about moonshot redesigns. It is about compounding small improvements.

A 5% improvement per month compounds to:

• After 6 months: 34% total improvement
• After 12 months: 80% total improvement
• After 24 months: 3.2x total improvement

The process:

1. Measure — install analytics, heatmaps, and session recordings (Hotjar, PostHog, or similar)
2. Hypothesize — based on data, not opinions: "We believe that [change] will [outcome] because [evidence]"
3. Test — one variable at a time, with sufficient sample size
4. Ship or kill — if the variant wins with 95% confidence, ship it. If not, kill it and move on.
5. Document — every test, win or lose, goes into a learning repository
6. Repeat — weekly test cadence minimum

第三部分：轉換率優化（CRO）

扼殺轉換率的常見網站錯誤

分析了上千個 SaaS 網站後，以下是最常見的致命錯誤：

1. 首屏沒有明確的 CTA。 如果訪客在 5 秒內搞不清楚要做什麼，你就輸了。
2. 太多 CTA 互相競爭。 「免費試用」+「預約 demo」+「看影片」+「讀文件」= 決策癱瘓。
3. 充滿術語的文案。 你知道「AI 驅動的自動化編排層」是什麼意思。你的客戶不知道。
4. 沒有優化載入速度。 每多一秒的載入時間，轉換率大約掉 7%。
5. 行動版是事後想到的。 超過 50% 的流量來自手機。如果你的定價頁在手機上看不清楚，一半的訪客直接離開。
6. 導覽列太多選項。 每個連結都是一個出口。最多限制在 5-7 個。
7. 社會證明被埋在深處。 如果見證只放在一個沒人會點的「客戶評價」頁面，它們就沒有用。
8. 首屏是一面文字牆。 如果你的首屏區域超過 30 個字，你正在流失用戶。

20,000 場 A/B 實驗的教訓

大規模實驗中最反直覺的發現：

通常有效的做法：

• 減少表單欄位（每移除一個欄位 ≈ 完成率提升 5-10%）
• 讓 CTA 按鈕更大、對比更強
• 加入緊迫感（真正的緊迫感，不是假的倒計時）
• 減少頁面上的選項
• 表單加入即時驗證
• 社會證明直接放在 CTA 旁邊

通常無效（或適得其反）的做法：

• 只改按鈕顏色，其他都不動
• 在 Landing Page 加聊天機器人（互動率上升，但轉換率常下降）
• 首次造訪就跳彈窗（回訪有效，但惹惱第一次來的人）
• 聰明或可愛的 CTA 文案（「開始吧！」vs.「免費試用」——無聊的版本通常贏）
• 一次重新設計整個頁面（你無法知道哪個改動有效）

CRO 的 80/20 法則：

• 80% 的轉換影響來自文案和方案（訊息、定價、定位）
• 15% 來自版面和 UX（頁面結構、表單設計、手機優化）
• 5% 來自視覺設計（顏色、字體、圖片）

大多數團隊把 80% 的時間花在那 5% 的類別上。

CTA 優化：為什麼「Book a Demo」表現不佳

「Book a Demo」是 B2B SaaS 的預設 CTA，也是表現最差的之一。

為什麼失敗：

• 它要求在展示價值之前先做出承諾（安排時間）
• 「Demo」暗示一場推銷，會觸發抗拒心理
• 它製造摩擦：日曆安排、時區協調、等待

更高轉換率的替代方案：

取代…	試試…	原因
"Book a Demo"	「看看實際操作」（搭配互動產品導覽）	不需承諾
"Book a Demo"	「免費試用」	摩擦更低，立即獲得價值
"Contact Sales"	「取得客製報價」	暗示價值，而非推銷電話
"Sign Up"	「開始建造」或「建立你的第一個 [成果]」	以成果為導向
"Learn More"	「看 [公司] 如何省下 40%」	具體且引發好奇

什麼時候「Book a Demo」是合適的：

• 企業銷售（ACV $10k 以上），買家預期有引導流程
• 確實需要解釋的複雜產品
• 搭配次要、低摩擦的 CTA 使用（「或免費試用」）

付費牆重新設計：註冊率提升 23%

一個付費牆畫面的單一改動，讓註冊率提升了 23%。改動內容：

之前： 付費牆顯示方案名稱、價格和一列功能清單。

之後： 付費牆顯示用戶在免費方案中已經完成了什麼，然後把付費方案包裝成「繼續你的進度」。

心理機制：賦予進度效應（endowed progress effect）。 當人們覺得自己已經朝目標前進了一段距離，他們就會顯著更有意願繼續。向用戶展示他們已經建立的東西（建立的專案、匯入的數據、設定的整合），讓付費方案感覺像是自然的下一步，而非一個新的承諾。

如何應用：

1. 追蹤用戶在免費使用或試用期間的操作
2. 在轉換節點，總結他們的活動：「你已經建立了 3 個專案、邀請了 2 位隊友、處理了 1,847 筆記錄」
3. 把升級包裝成延續：「繼續用 Pro」而非「升級到 Pro」
4. 同時展示他們會失去什麼（損失趨避）和會得到什麼

模糊背景效應：轉換率提升 94%

CRO 領域最令人驚訝的發現之一：在註冊表單後方使用模糊背景圖片，比清晰圖片或純色背景多出 94% 的轉換率。

為什麼有效：

• 模糊背景暗示深度和脈絡，但不會搶奪注意力
• 減少表單周圍的視覺噪音
• 對表單本身產生「聚光燈」效果
• 用戶的目光無處可逛

應用原則：

• 在任何高優先級的表單或 CTA 區域背後使用模糊
• 模糊程度應為 15-25px 高斯模糊——明顯的模糊，不是微微失焦
• 底層圖片應暗示產品或使用情境
• 搭配半透明的深色或淺色遮罩以增加對比

小改動原則

CRO 不是一次性的大翻修，而是小改善的持續複利。

每月 5% 的改善會複利成：

• 6 個月後： 總改善 34%
• 12 個月後： 總改善 80%
• 24 個月後： 總改善 3.2 倍

流程：

1. 衡量 —— 安裝分析工具、熱力圖和 session 錄製（Hotjar、PostHog 或類似工具）
2. 假設 —— 基於數據，不是意見：「我們相信 [改動] 會 [結果]，因為 [證據]」
3. 測試 —— 一次一個變數，樣本數要夠
4. 上線或淘汰 —— 如果變體在 95% 信心水準下勝出，上線。否則淘汰，繼續下一個。
5. 記錄 —— 每次測試，不管成敗，都寫入學習資料庫
6. 重複 —— 至少每週一次測試節奏

---

Part 4: User Research — The Foundation of Everything

If you skip this section and go straight to optimizing your landing page, you are optimizing in the dark. Every principle above — positioning, messaging, pricing, CRO — depends on understanding your users deeply enough to make good decisions.

The Engineer's Guide to Talking to Users

Most technical founders dread user interviews. Here is the reframe: a user interview is debugging, but for your business model.

You already know how to debug: reproduce the problem, isolate variables, form hypotheses, test them. User interviews are the same process applied to "why do people buy (or not buy) my product?"

Rules for technical founders:

1. Shut up and listen. Your goal is to understand their world, not to explain your product.
2. Ask about the past, not the future. "Tell me about the last time you..." > "Would you use a feature that..."
3. Follow the emotion. When they say "that was really frustrating," that is your cue to dig deeper: "Tell me more about that."
4. Do not pitch. The moment you start selling, you stop learning.
5. Record everything. With permission, record the call. You will miss things in real-time.

Minimum viable interview process:

• 5 interviews per round (you will hear patterns by interview 3-4)
• 30 minutes each
• One interviewer, one note-taker
• Synthesize within 24 hours while memory is fresh

Automated Insights Collection

You cannot interview everyone. Supplement with automated methods:

• In-app surveys (1-2 questions, triggered after specific actions): "What almost stopped you from signing up?"
• NPS with follow-up: The score is useless without the "why." Always ask "What is the primary reason for your score?"
• Exit surveys on cancellation: "What could we have done differently?"
• Post-onboarding survey (day 3-5): "On a scale of 1-5, how easy was it to get started?"
• Feature usage analytics: What do people actually do vs. what you think they do?

High-Speed Research for Busy Founders

If you have 30 minutes, not 30 hours:

1. The 5-minute test. Show your homepage to 5 people for 5 seconds each. Ask: "What does this company do?" If they cannot answer correctly, your messaging is broken.
2. Support ticket mining. Your support inbox is a research goldmine. Categorize the last 100 tickets. The top 3 categories tell you where your product and messaging are misaligned.
3. Review mining. Read your competitors' G2/Capterra reviews. The negative reviews tell you what to build. The positive reviews tell you what to say.
4. Sales call recordings. If you have a sales team, listen to 5 discovery calls. Write down every objection. Your landing page should preemptively address the top 3.

How to Write Survey Questions

Bad survey questions produce bad data, and you will not realize it until you have made decisions based on them.

Rules:

Do	Do not
Ask one thing per question	Ask double-barreled questions ("How satisfied are you with our price and features?")
Use specific language	Use vague scales ("How good is our product?")
Randomize option order	Always put your preferred answer first
Include "Other" with a text field	Force people into your categories
Keep it under 5 minutes	Send a 30-question survey (completion rate will be < 10%)

Question types ranked by signal quality:

1. Open-ended — highest signal, hardest to analyze at scale
2. Ranking — forces tradeoffs, reveals priorities
3. Multiple choice — easy to analyze, limited depth
4. Likert scale (1-5) — common but often produces meaningless middle-clustering
5. Yes/No — fast but low signal

Superhuman's Product-Market Fit Engine

Superhuman's PMF measurement, based on the Sean Ellis test, is one of the most practical frameworks for early-stage products.

The core question:

"How would you feel if you could no longer use [product]?"

Answers:

• Very disappointed
• Somewhat disappointed
• Not disappointed

The benchmark: If 40% or more of respondents say "very disappointed," you have product-market fit.

Superhuman's process:

1. Survey all active users with the core question
2. Segment responses by user type
3. For the "very disappointed" group: ask "What is the main benefit you get from [product]?" — this becomes your positioning
4. For the "somewhat disappointed" group: ask "How can we improve [product] for you?" — this becomes your roadmap
5. Ignore the "not disappointed" group — they are not your target customer
6. Rerun the survey every quarter and track the percentage over time

Why this works better than NPS:

• NPS measures satisfaction. The Sean Ellis test measures dependency.
• A user can rate you 8/10 on NPS but switch to a competitor tomorrow. A user who would be "very disappointed" without you is retained by definition.

MaxDiff for Feature Prioritization

When you have 10 features on the roadmap and need to know which 3 to build first, MaxDiff eliminates the "everything is important" problem.

How it works:

1. List 8-12 potential features
2. Show respondents groups of 4-5 features at a time
3. For each group, they pick the most important and least important
4. Repeat with different combinations
5. Statistical analysis produces a ranked list with relative importance scores

Why MaxDiff beats direct ranking:

• Direct ranking ("rank these 10 features") is cognitively impossible for more than 5-6 items
• Rating scales ("rate each feature 1-5") produce ceiling effects where everything is "very important"
• MaxDiff forces genuine tradeoffs by only asking about relative preference within small groups

Journey Mapping Interviews

Journey mapping reveals the full arc of a customer's experience — from first awareness to daily usage (or churn).

The interview script:

1. "How did you first hear about us?" (acquisition channel effectiveness)
2. "What were you using before?" (competitive landscape)
3. "What made you decide to try us?" (conversion trigger)
4. "Walk me through your first day using the product." (onboarding experience)
5. "When did it first feel valuable?" (time-to-value / aha moment)
6. "What almost made you stop using it?" (churn risk)
7. "What would make you recommend us to a colleague?" (referral trigger)

Map the answers onto a timeline. Mark emotional highs and lows. The lows are your optimization opportunities.

Jobs To Be Done (JTBD) Framework

JTBD reframes your product from "a set of features" to "a tool people hire to make progress in their lives."

The core question: "What job is the customer hiring our product to do?"

JTBD interview template:

1. First thought: "When did you first realize you needed something like this?"
2. Passive looking: "What did you do next? Did you search? Ask someone?"
3. Active looking: "When did it become urgent enough to actually look for a solution?"
4. Deciding: "What made you choose us over the alternatives?"
5. Consuming: "Walk me through the first time you used it."
6. Satisfaction: "Is it doing the job you hired it for?"

The four forces of progress:

Force	Direction	Example
Push (away from current situation)	Toward switch	"Our old tool kept crashing"
Pull (toward new solution)	Toward switch	"Your product looked so much easier"
Anxiety (about new solution)	Against switch	"What if we lose our data during migration?"
Habit (attachment to current situation)	Against switch	"We have used this tool for 3 years"

Your marketing must amplify Push and Pull while reducing Anxiety and Habit.

22 Essential Customer Questions

Keep this list handy. Not every question is right for every conversation, but these cover the full spectrum of customer insight:

Understanding the problem:

1. What is the hardest part of your day/week related to [domain]?
2. What tools are you currently using for [task]?
3. What do you wish those tools did differently?
4. How much time do you spend on [task] per week?
5. What happens when [task] goes wrong?

Understanding the buying decision: 6. Who else is involved in purchasing decisions like this? 7. What budget range are you working with? 8. What would need to be true for you to switch from your current solution? 9. What is your timeline for making a change? 10. What has stopped you from solving this problem before?

Understanding value: 11. What metric would improve if this problem were solved? 12. How would your day be different if this were fixed? 13. What would you do with the time you save? 14. How much is this problem costing you? (In time, money, or frustration) 15. Who else on your team would benefit?

Understanding the product experience: 16. What was your first impression when you tried the product? 17. What feature do you use most? Why? 18. What feature did you expect that was not there? 19. On a scale of 1-10, how easy was it to get started? 20. What would make you recommend us to a friend?

Understanding churn risk: 21. What, if anything, has frustrated you about the product? 22. If you had to stop using us tomorrow, what would you do instead?

Maze Question Bank: 350+ Research Questions

The Maze Question Bank is an open resource with 350+ pre-written research questions organized by research phase:

• Discovery: Understanding user needs and behaviors
• Validation: Testing concepts and prototypes
• Usability: Evaluating product experience
• Post-launch: Measuring satisfaction and finding issues

Use it as a starting point, not a script. Pick 5-8 questions per interview, tailored to your specific research goal.

Resource: Maze Question Bank

Michele Hansen's 4 Interview Scripts

From Deploy Empathy, these four scripts cover the complete customer lifecycle:

Script 1: New Customer Interview

Goal: Understand why they signed up and what triggered the decision.

Key questions:

• "What were you doing when you decided to sign up?"
• "What else did you try before us?"
• "What almost stopped you from signing up?"
• "What do you hope to accomplish in your first week?"

Script 2: Happy Customer Interview

Goal: Understand what value they get and how to replicate it for others.

Key questions:

• "What has changed since you started using [product]?"
• "If you had to describe [product] to a colleague, what would you say?"
• "What is the one thing you would never want us to remove?"
• "What would make you upgrade to a higher tier?"

Script 3: Churned Customer Interview

Goal: Understand why they left and whether it was preventable.

Key questions:

• "What prompted you to cancel?"
• "Was there a specific moment when you decided to leave?"
• "What are you using now instead?"
• "Is there anything that could bring you back?"

Script 4: Feature Request Interview

Goal: Understand the real need behind the request (often different from the stated request).

Key questions:

• "Can you walk me through the situation where you needed this?"
• "How are you handling this currently?"
• "If we built exactly what you described, how would your workflow change?"
• "How important is this compared to the other things you use [product] for?"

The critical principle across all four scripts: listen for the job, not the feature. A customer who says "I need a Gantt chart" might actually need "I need to show my boss that the project is on track." Those are very different solutions.

第四部分：使用者研究——一切的根基

如果你跳過這一節直接去優化 Landing Page，你就是在黑暗中優化。上面的每一個原則——定位、訊息、定價、CRO——都建立在你對使用者夠深入的理解之上。

工程師的用戶訪談指南

大多數技術背景的創辦人一聽到用戶訪談就頭痛。換個角度想：用戶訪談就是 debug，只是對象是你的商業模式。

你已經知道怎麼 debug：重現問題、隔離變數、形成假設、驗證假設。用戶訪談是同一套流程，只是應用在「為什麼人們會（或不會）買我的產品？」

給技術創辦人的規則：

1. 閉嘴、聽。 你的目標是理解他們的世界，不是解釋你的產品。
2. 問過去，不問未來。 「跟我說上次你……的經歷」 > 「你會不會用一個可以……的功能？」
3. 追蹤情緒。 當他們說「那真的很讓人挫折」，那就是你該深挖的訊號：「多說一點那個。」
4. 不要推銷。 你開始賣的那一刻，就是你停止學習的那一刻。
5. 全程錄音。 經過同意後，錄下通話。你在即時對話中一定會漏掉東西。

最小可行訪談流程：

• 每輪 5 次訪談（到第 3-4 次你就會聽到模式）
• 每次 30 分鐘
• 一個訪談者，一個紀錄者
• 24 小時內彙整，趁記憶還新鮮

自動化洞察收集

你不可能訪談每個人。用自動化方法補足：

• App 內問卷（1-2 題，在特定操作後觸發）：「什麼差點讓你放棄註冊？」
• NPS + 追問： 分數沒有「為什麼」就毫無用處。一定要問「你打這個分數的主要原因是什麼？」
• 取消時的離開問卷： 「我們可以做什麼不一樣的？」
• 引導後問卷（第 3-5 天）：「1 到 5 分，上手有多容易？」
• 功能使用分析： 人們實際做了什麼 vs. 你以為他們做了什麼？

忙碌創辦人的高速研究法

如果你只有 30 分鐘，而不是 30 小時：

1. 5 秒測試。 把首頁給 5 個人看 5 秒鐘。問：「這家公司做什麼？」如果他們答不對，你的訊息傳達有問題。
2. 客服信件挖掘。 你的客服信箱是研究金礦。把最近 100 封信分類。前三大類別告訴你產品和訊息在哪裡不一致。
3. 評論挖掘。 讀競品的 G2/Capterra 評論。負面評論告訴你該做什麼，正面評論告訴你該說什麼。
4. 銷售通話錄音。 如果你有銷售團隊，聽 5 通探索電話。記下每一個異議。你的 Landing Page 應該預先處理前三大異議。

問卷題目怎麼寫

爛問題產生爛數據，而且你不會發現——直到你已經根據它做了決策。

規則：

要做的	不要做的
每題只問一件事	問雙管問題（「你對我們的價格和功能滿意嗎？」）
用具體的語言	用模糊的量尺（「我們的產品有多好？」）
隨機排列選項順序	總是把你偏好的答案放第一個
加入「其他」和文字欄位	強迫人們選進你的分類
控制在 5 分鐘內	送出 30 題問卷（完成率會低於 10%）

問題類型，按訊號品質排名：

1. 開放式 —— 最高訊號，大規模分析最困難
2. 排序 —— 迫使取捨，揭示優先順序
3. 多選 —— 容易分析，深度有限
4. 李克特量表（1-5） —— 常見但容易產生無意義的中間聚集
5. 是/否 —— 快但訊號低

Superhuman 的 PMF 引擎

Superhuman 的 PMF 衡量方法，基於 Sean Ellis 測試，是早期產品最實用的框架之一。

核心問題：

「如果你再也不能使用 [產品]，你會有什麼感覺？」

選項：

• 非常失望
• 有點失望
• 不會失望

基準線： 如果 40% 以上的受訪者選「非常失望」，你就有了 product-market fit。

Superhuman 的流程：

1. 對所有活躍用戶問核心問題
2. 按用戶類型分群回答
3. 對「非常失望」群組：問「你從 [產品] 得到的主要好處是什麼？」—— 這成為你的定位
4. 對「有點失望」群組：問「我們怎麼改進 [產品] 讓它更適合你？」—— 這成為你的產品路線圖
5. 忽略「不會失望」群組——他們不是你的目標客戶
6. 每季重跑一次問卷，追蹤百分比的變化

為什麼比 NPS 更有效：

• NPS 衡量的是滿意度。Sean Ellis 測試衡量的是依賴度。
• 一個用戶可以在 NPS 上給你 8/10 分，但明天就跳槽去競品。一個「非常失望」的用戶，定義上就是被留住的。

MaxDiff 功能優先排序

當你有 10 個功能在路線圖上，需要知道先做哪 3 個，MaxDiff 消除了「每個都重要」的問題。

運作方式：

1. 列出 8-12 個潛在功能
2. 每次給受訪者看 4-5 個功能
3. 每組中，他們選出最重要的和最不重要的
4. 用不同組合重複
5. 統計分析產出一個有相對重要性分數的排名表

MaxDiff 為什麼比直接排名好：

• 直接排名（「把這 10 個功能排序」）對超過 5-6 個項目來說認知負荷太高
• 評分量表（「每個功能評 1-5 分」）會產生天花板效應，每個都「非常重要」
• MaxDiff 透過只在小組內問相對偏好，迫使真正的取捨

旅程圖訪談

旅程圖揭示客戶體驗的完整弧線——從第一次知道你到日常使用（或流失）。

訪談腳本：

1. 「你第一次是怎麼聽到我們的？」（獲客管道效果）
2. 「你之前在用什麼？」（競爭格局）
3. 「是什麼讓你決定試試看？」（轉換觸發點）
4. 「帶我走過你使用產品的第一天。」（引導體驗）
5. 「什麼時候你第一次覺得它有價值？」（time-to-value / aha moment）
6. 「什麼差點讓你停止使用？」（流失風險）
7. 「什麼會讓你推薦我們給同事？」（推薦觸發點）

把答案映射到時間軸上。標記情緒高點和低點。低點就是你的優化機會。

Jobs To Be Done（JTBD）框架

JTBD 把你的產品從「一組功能」重新定義為「人們雇用來在生活中取得進展的工具」。

核心問題： 「客戶雇用我們的產品來完成什麼工作？」

JTBD 訪談模板：

1. 第一個念頭： 「你什麼時候第一次覺得需要這類東西？」
2. 被動搜尋： 「然後你做了什麼？有搜尋嗎？有問人嗎？」
3. 主動搜尋： 「什麼時候它變得夠緊急，讓你真的開始找解決方案？」
4. 決定： 「是什麼讓你選了我們而不是其他方案？」
5. 使用： 「帶我走過你第一次使用的過程。」
6. 滿意度： 「它有在做你雇用它做的工作嗎？」

進步的四個力量：

力量	方向	範例
推力（遠離現狀）	促進轉換	「我們舊的工具一直當機」
拉力（朝向新方案）	促進轉換	「你們的產品看起來簡單很多」
焦慮（對新方案的）	阻礙轉換	「遷移的時候會不會丟資料？」
慣性（對現狀的依附）	阻礙轉換	「我們用這個工具已經三年了」

你的行銷必須放大推力和拉力，同時降低焦慮和慣性。

22 個必備客戶問題

把這份清單存起來。不是每個問題都適合每次對話，但它們涵蓋了客戶洞察的完整光譜：

理解問題：

1. 你每天/每週跟 [領域] 相關最困難的部分是什麼？
2. 你目前用什麼工具處理 [任務]？
3. 你希望那些工具有什麼不一樣？
4. 你每週花多少時間在 [任務] 上？
5. 當 [任務] 出錯時會發生什麼？

理解購買決策： 6. 還有誰會參與這類的採購決策？ 7. 你的預算範圍大概是？ 8. 什麼條件成立，你才會從現有方案轉換？ 9. 你做出改變的時間表是？ 10. 是什麼阻止你之前就解決這個問題？

理解價值： 11. 如果這個問題被解決了，什麼指標會改善？ 12. 如果這被修好了，你的一天會有什麼不同？ 13. 你會拿省下來的時間做什麼？ 14. 這個問題花了你多少代價？（時間、金錢或挫折感） 15. 團隊裡還有誰會受益？

理解產品體驗： 16. 你第一次試用產品時的第一印象是什麼？ 17. 你最常用什麼功能？為什麼？ 18. 有什麼功能是你預期會有但沒有的？ 19. 1 到 10 分，上手有多容易？ 20. 什麼會讓你推薦我們給朋友？

理解流失風險： 21. 有什麼讓你對產品感到不滿的嗎？ 22. 如果你明天必須停用我們，你會改用什麼？

Maze 問題庫：350+ 研究問題

Maze Question Bank 是一個開放資源，包含超過 350 個按研究階段分類的預寫問題：

• 探索期： 理解用戶需求和行為
• 驗證期： 測試概念和原型
• 易用性： 評估產品體驗
• 上線後： 衡量滿意度和發現問題

把它當起點，不是腳本。每次訪談挑 5-8 個問題，根據你的研究目標客製。

資源：Maze Question Bank

Michele Hansen 的 4 個訪談腳本

出自《Deploy Empathy》，這四個腳本涵蓋完整的客戶生命週期：

腳本一：新客戶訪談

目標：理解他們為什麼註冊，什麼觸發了決定。

關鍵問題：

• 「你決定註冊的時候在做什麼？」
• 「在我們之前你還試了什麼？」
• 「什麼差點讓你放棄註冊？」
• 「你希望第一週完成什麼？」

腳本二：滿意客戶訪談

目標：理解他們獲得什麼價值，以及如何為其他人複製。

關鍵問題：

• 「開始用 [產品] 之後，什麼改變了？」
• 「如果你要向同事描述 [產品]，你會怎麼說？」
• 「有什麼功能是你絕對不希望我們拿掉的？」
• 「什麼會讓你升級到更高的方案？」

腳本三：流失客戶訪談

目標：理解他們為什麼離開，是否可以預防。

關鍵問題：

• 「是什麼促使你取消的？」
• 「有沒有一個具體的時刻讓你決定離開？」
• 「你現在改用什麼？」
• 「有什麼能讓你回來嗎？」

腳本四：功能需求訪談

目標：理解需求背後的真正需要（通常跟表面的需求不同）。

關鍵問題：

• 「你可以帶我走過你需要這個功能的場景嗎？」
• 「你目前怎麼處理這件事？」
• 「如果我們完全照你描述的做出來，你的工作流會怎麼改變？」
• 「比起你用 [產品] 的其他事情，這個有多重要？」

四個腳本的關鍵原則：聽工作，不聽功能。 一個說「我需要甘特圖」的客戶，實際上可能需要的是「我需要讓老闆看到專案在正軌上」。那是兩個非常不同的解方。

---

Putting It All Together

The conversion optimization stack, bottom-up:

┌─────────────────────────────┐
│   Conversion Rate (output)  │
├─────────────────────────────┤
│   CRO Experiments           │
│   (A/B tests, UX tweaks)    │
├─────────────────────────────┤
│   Pricing Architecture      │
│   (tiers, trials, annual)   │
├─────────────────────────────┤
│   Landing Page & Messaging  │
│   (headlines, CTAs, proof)  │
├─────────────────────────────┤
│   Positioning               │
│   (who, what, why different)│
├─────────────────────────────┤
│   User Research             │
│   (interviews, surveys,     │
│    JTBD, PMF measurement)   │
└─────────────────────────────┘

The action plan:

1. This week: Run 5 user interviews using the scripts from Part 4. Run the Sean Ellis survey.
2. Next week: Rewrite your positioning using the April Dunford framework based on interview findings.
3. Week 3: Rewrite your homepage headline and CTA based on new positioning. Launch a price test.
4. Week 4: Install analytics and heatmaps. Start your first A/B test.
5. Ongoing: One CRO test per week. Five user interviews per quarter. Sean Ellis survey every quarter.

The companies that win at conversion are not the ones with the best design or the cleverest copy. They are the ones who talk to their users the most, test the most, and compound improvements over time.

Start at the bottom. Build up. Keep compounding.

整合起來

轉換率優化的架構，由下往上：

┌─────────────────────────────┐
│   轉換率（成果）              │
├─────────────────────────────┤
│   CRO 實驗                   │
│   （A/B 測試、UX 微調）       │
├─────────────────────────────┤
│   定價架構                    │
│   （方案、試用、年繳）         │
├─────────────────────────────┤
│   Landing Page 與訊息傳達     │
│   （標題、CTA、社會證明）      │
├─────────────────────────────┤
│   定位                        │
│   （給誰、做什麼、為何不同）    │
├─────────────────────────────┤
│   使用者研究                   │
│   （訪談、問卷、JTBD、PMF）   │
└─────────────────────────────┘

行動計畫：

1. 這週： 用第四部分的腳本做 5 次用戶訪談。跑 Sean Ellis 問卷。
2. 下週： 根據訪談結果，用 April Dunford 框架重寫定位。
3. 第三週： 根據新定位重寫首頁標題和 CTA。啟動一個價格測試。
4. 第四週： 安裝分析工具和熱力圖。開始你的第一個 A/B 測試。
5. 持續進行： 每週一個 CRO 測試。每季 5 次用戶訪談。每季一次 Sean Ellis 問卷。

在轉換率上贏的公司，不是設計最好或文案最巧的。而是跟用戶聊最多、測試最多、長期複利最久的。

從底層開始。往上建構。持續複利。

---

Resources

All resources referenced in this guide are curated from the Marketing for Founders collection:

• April Dunford — Obviously Awesome (positioning)
• Michele Hansen — Deploy Empathy (customer interviews)
• Sean Ellis — PMF survey methodology
• Van Westendorp — Price Sensitivity Meter
• Maze — Question Bank (350+ research questions)
• Superhuman — How Superhuman Built an Engine to Find Product-Market Fit

參考資源

本指南引用的所有資源皆整理自 Marketing for Founders 合輯：

• April Dunford —— 《Obviously Awesome》（定位）
• Michele Hansen —— 《Deploy Empathy》（客戶訪談）
• Sean Ellis —— PMF 問卷方法論
• Van Westendorp —— 價格敏感度模型
• Maze —— Question Bank（350+ 研究問題）
• Superhuman —— How Superhuman Built an Engine to Find Product-Market Fit