找到潛在客戶不難,找到正確的潛在客戶才難。這些提示詞幫助您建立精準的潛在客戶名單、撰寫能得到回覆的外展訊息,並在浪費時間之前篩選潛在客戶。已在ChatGPT、Gemini和Claude上測試,讓您知道哪個模型能提升潛在客戶開發能力。
| 您想做的事 | 最適合 |
|---|---|
| 定義您的理想客戶畫像 | Claude |
| 在接觸前深入研究潛在客戶 | ChatGPT |
| 建立多觸點的外展序列 | ChatGPT |
| 快速判斷潛在客戶是否值得跟進 | Gemini |
| 建立系統化的轉介流程 | Claude |
| 在LinkedIn上有效地開發潛在客戶 | Gemini |
提示詞
定義您的理想客戶畫像
Help me build an ideal customer profile for [product/service]. What I sell: [describe your offering] Current best customers: [describe 3-5 of your most profitable/happiest customers] Deal size: [average contract value] Sales cycle: [how long deals typically take] Channels that work: [where you currently find customers] Build a comprehensive ICP: 1. Firmographic profile: industry, company size, revenue range, geography 2. Technographic profile: tools they use, tech maturity level 3. Trigger events: what happens in their business that creates buying urgency 4. Buyer personas: 2-3 decision-maker profiles with titles, goals, and frustrations 5. Disqualification criteria: red flags that signal a bad-fit prospect 6. A scoring rubric (1-10) to rank incoming leads against this ICP
最適合: CLAUDE
Claude能建構最多維度的ICP,超越基本的人口統計和公司規模。其框架包含行為信號、購買觸發因素和判定指標,讓銷售團隊能快速判斷一個潛在客戶是否真正符合。
已測試 Feb 15, 2026
進階技巧
從您最好的5個現有客戶開始逆向工程ICP。不要問「我們想賣給誰?」而要問「誰最快買、付最多、續約最久?」
在接觸前深入研究潛在客戶
I'm about to reach out to [prospect name/company]. Help me prepare. Their company: [name, industry, size if known] Their role: [job title] What I sell: [your product/service] What I know so far: [any existing information] My goal: [book a meeting / get a referral / start a conversation] Research and prepare: 1. Key business challenges this type of company likely faces right now 2. Industry trends affecting their decision-making 3. 5 personalization angles I can use in my outreach 4. Potential internal priorities based on their role 5. Questions to ask that show I've done my homework 6. A pre-call one-pager I can review 5 minutes before reaching out
最適合: CHATGPT
Gemini能進行最結構化的潛在客戶研究,將公開資訊組織成可行動的銷售情報。其報告格式讓銷售人員能在5分鐘內掌握客戶的關鍵資訊。
已測試 Feb 15, 2026
進階技巧
尋找「觸發事件」——融資、高管變動、收購、新產品發布。這些事件創造了購買的緊迫性。向剛融資的公司推銷比向穩定的公司推銷容易10倍。
建立多觸點的外展序列
Build a multi-touch outreach sequence for [target persona]. Product/service: [what you're offering] Channel mix: [email, LinkedIn, phone, video — what you'll use] Sequence length: [number of touches over how many days] Value proposition: [main benefit] Social proof: [case studies, numbers, logos] Create the full sequence: 1. Touch 1: First email with subject line and body (under 100 words) 2. Touch 2: LinkedIn connection request with personalized note 3. Touch 3: Follow-up email with new value angle 4. Touch 4: Video message script (under 60 seconds) 5. Touch 5: Breakup email that creates closure urgency 6. Timing recommendations: days between each touch and best send times
最適合: CHATGPT
ChatGPT能設計最自然流暢的多觸點序列,每封訊息之間有邏輯連貫性。其序列不會感覺像是獨立的推銷信,而是一段持續的對話。
已測試 Feb 15, 2026
進階技巧
在序列中混合渠道——郵件、LinkedIn、電話。單一渠道的序列回覆率平均為3%,多渠道的序列回覆率可達15到25%。
快速判斷潛在客戶是否值得跟進
Help me build a lead qualification process for [business type]. What I sell: [product/service and price range] Sales team size: [number of reps] Current lead volume: [leads per month] Biggest time waste: [types of leads that never close] Current qualification method: [what you do now, or nothing] Design a qualification system: 1. A BANT+ framework customized to my business (Budget, Authority, Need, Timeline + 2 custom criteria) 2. 10 discovery questions ranked by importance 3. A lead scoring matrix: what adds points, what subtracts them 4. A 'fast no' checklist: disqualify in under 2 minutes 5. Handoff criteria: when a lead moves from SDR to AE 6. A qualification call script that feels like a conversation, not an interrogation
最適合: GEMINI
Claude能建立最務實的資格審核框架,在全面性和速度之間取得最佳平衡。其評分系統不僅看表面資訊,還能識別「隱藏的不合格」信號。
已測試 Feb 15, 2026
進階技巧
問「您打算什麼時候解決這個問題?」如果答案是「明年」或「某天」,這不是一個合格的潛在客戶——只是一個有興趣的人。時間表是最可靠的資格審核指標。
建立系統化的轉介流程
Help me build a referral system for [business]. Current customer base: [approximate number] Average deal value: [price] Current referral rate: [percentage of new business from referrals] Incentive budget: [what you can offer] Best customers: [describe your happiest customers] Design a referral engine: 1. The right moment to ask for referrals (and the wrong moments) 2. Three referral ask scripts (casual, formal, email) 3. An incentive structure that motivates without cheapening the relationship 4. A referral tracking system I can build in a spreadsheet 5. A 'warm introduction' email template the customer can forward 6. A quarterly referral campaign that reminds without nagging
最適合: CLAUDE
Gemini能建立最系統化的轉介流程,從時機、話術到追蹤,每個步驟都有明確的模板。其轉介請求腳本自然不尷尬。
已測試 Feb 15, 2026
進階技巧
在客戶成功的高峰期要求轉介——剛完成入職、達成首個里程碑、或續約時。這時他們最有動力幫您推薦。不要等到隨機的時間點。
在LinkedIn上有效地開發潛在客戶
Help me use LinkedIn to generate leads for [product/service]. Target audience: [job titles and industries] My LinkedIn presence: [follower count, posting frequency, profile strength] Goal: [meetings booked per month] Time available: [hours per week for LinkedIn] Content I can share: [blog posts, case studies, insights] Build a LinkedIn lead gen strategy: 1. Profile optimization checklist (headline, about section, featured) 2. A daily LinkedIn routine (15 minutes) that generates inbound interest 3. Content strategy: 3 post types that attract decision-makers 4. Comment strategy: how to engage on prospects' posts without being salesy 5. Connection request templates for warm and cold outreach (5 each) 6. A conversion path: from connection to conversation to meeting
最適合: GEMINI
ChatGPT能撰寫最自然的LinkedIn外展訊息,不像典型的推銷訊息。其方法以建立關係為核心,而非直接推銷——這正是LinkedIn上最有效的策略。
已測試 Feb 15, 2026
進階技巧
先互動對方的內容(留言、分享)2到3週再發連結請求。「冷」連結請求的接受率約為20%,「暖」連結請求(對方認識您的名字)的接受率可達60%以上。
基於實際測試結果 — 非假設推測。 查看測試方法
Gemini
Best for qualification frameworks and LinkedIn strategies. Produces structured, implementable systems with clear scoring criteria. Less effective at writing personalized outreach copy.
結果來源: Gemini 2.0 Flash · 已測試 Feb 15, 2026ChatGPT
Best for outreach sequences and prospect research. Generates the most natural multi-channel messaging with broad industry knowledge. Can produce generic outreach if not given enough personalization context.
結果來源: GPT-4o · 已測試 Feb 15, 2026Claude
Best for ICP development and referral programs. Builds nuanced buyer profiles and relationship-based lead gen strategies. Sometimes overcomplicates qualification criteria — push for simplicity.
結果來源: Claude 3.5 Sonnet · 已測試 Feb 15, 2026Grok
Excels at creative, unconventional outreach angles that break through inbox noise and pattern interrupts. Writes outbound messages with personality that feel refreshingly human. Less systematic at building scalable lead gen infrastructure and scoring frameworks.
結果來源: Grok 2 · 已測試 Feb 15, 2026Qualify out faster than you qualify in. The fastest path to more revenue isn't more leads — it's spending zero time on bad ones. Build your disqualification criteria first and share them with marketing so bad leads never reach your pipeline.
Outbound works best when inbound is also running. Cold outreach converts 3x better when the prospect has already seen your brand through content or ads. Run a small awareness campaign alongside outbound so your name isn't completely unfamiliar.
Track 'speed to lead' as your #1 metric. Responding to an inbound lead within 5 minutes is 10x more effective than responding within an hour. AI can write perfect outreach, but if you're slow to follow up, none of it matters.