AI 提示詞: 客戶回饋

收集回饋很容易。如何處理才是難題。這些提示詞幫助您設計能獲得誠實回答的調查、分析回饋數據的模式,並將投訴轉化為產品改進。已在ChatGPT、Gemini和Claude上測試,讓您知道哪個模型更善於傾聽。

最近測試日期 Feb 15, 2026 · 模型: GPT-4o, Gemini 2.0, Claude 3.5 Sonnet, Grok 2
您想做的事 提示詞 最適合
設計能獲得有用回覆的調查問卷 調查設計師 Claude
從客戶回饋中提取可行動的洞察 回饋分析器 Gemini
專業地回覆線上評論 評論回覆專家 ChatGPT
將NPS分數轉化為具體改善措施 NPS行動計劃 Claude
為客戶的功能需求排定優先順序 功能需求優先排序 Gemini
建構持續的回饋系統 回饋循環建構 ChatGPT

調查設計師

設計能獲得有用回覆的調查問卷

I need a customer feedback survey for [product/service/experience].

Goal of the survey: [what you want to learn]
Audience: [who will fill it out]
Survey length target: [number of questions]
Distribution method: [email, in-app, post-purchase, etc.]
Last survey sent: [when, or never]

Design the survey:
1. 8-12 questions in optimal order (easy to hard, closed to open)
2. A mix of rating scales, multiple choice, and open-ended questions
3. The one question that reveals the most actionable insight
4. NPS question placement and follow-up question for detractors
5. An incentive strategy that increases completion without biasing results
6. A pre-survey message that explains why their feedback matters (under 50 words)

最適合: CLAUDE

Gemini能建立最結構化的調查框架,問題順序和措辭經過優化以減少偏差並最大化完成率。其跳題邏輯建議讓調查對每個受訪者都保持相關性。

已測試 Feb 15, 2026

進階技巧

將調查限制在5到7個問題以內。每多一個問題,完成率就下降約10%。一份100%完成率的5題調查比50%完成率的20題調查提供更多可靠數據。

回饋分析器

從客戶回饋中提取可行動的洞察

Analyze this customer feedback and tell me what to prioritize:

[Paste 20-50 pieces of customer feedback: reviews, survey responses, support tickets, social comments]

Product/service: [what they're giving feedback on]
Recent changes: [anything you've changed recently]
Known issues: [problems you're already aware of]

Provide:
1. Theme analysis: group feedback into 5-7 categories with frequency counts
2. Sentiment breakdown: positive, negative, neutral percentages
3. The #1 issue to fix immediately (biggest impact, most mentioned)
4. Hidden insights: feedback patterns that aren't obvious on first read
5. Exact customer quotes that best represent each theme
6. A prioritized action plan: what to fix this week, this month, this quarter

最適合: GEMINI

Claude能進行最深入的回饋情感分析,不僅分類正面/負面,還能識別未明確表達的潛在需求和期望。其主題歸類框架讓大量回饋變得可理解和可行動。

已測試 Feb 15, 2026

進階技巧

不要只看最常見的投訴。尋找「驚喜」回饋——客戶提到您從未想過的使用方式或需求。這些往往是最有價值的產品創新來源。

評論回覆專家

專業地回覆線上評論

Help me respond to these customer reviews:

[Paste 5-10 reviews with star ratings]

Business type: [what you do]
Brand voice: [professional/friendly/empathetic/casual]
Common misconceptions: [things reviewers often get wrong]
Changes you've made: [improvements based on past feedback]

For each review, write:
1. A personalized response (not generic 'thank you for your feedback')
2. Acknowledgment of their specific experience
3. For negative reviews: what you're doing to address the issue
4. For positive reviews: a genuine reaction that encourages loyalty
5. A subtle invitation to return or try something new
6. Flag any reviews that need private follow-up (and a DM template)

最適合: CHATGPT

ChatGPT能撰寫最自然、個人化的評論回覆,不會聽起來像公司制式回應。它在1星和5星評論之間完美調整語調,不會防禦性或過度感激。

已測試 Feb 15, 2026

進階技巧

在4小時內回覆負面評論,在24小時內回覆正面評論。負面評論的快速回覆顯示您的重視。正面評論的稍遲回覆感覺沒問題,因為快樂的客戶不會焦急等待。

NPS行動計劃

將NPS分數轉化為具體改善措施

Help me act on our NPS results.

Current NPS score: [number]
Previous NPS score: [number and when]
Promoters feedback themes: [what happy customers say]
Detractors feedback themes: [what unhappy customers say]
Passives feedback themes: [what neutral customers say]
Response rate: [percentage who responded]

Build an NPS action plan:
1. Score interpretation: what our NPS really means in context
2. Top 3 actions to convert passives to promoters
3. Top 3 actions to reduce detractor volume
4. A promoter program: how to activate happy customers as advocates
5. Detractor rescue workflow: steps to save at-risk customers
6. Quarterly NPS improvement targets with specific initiatives for each

最適合: CLAUDE

Claude提供最誠實的NPS解讀,包括分數何時因低回覆率或偏差樣本而具有誤導性。其行動計劃優先考慮最具槓桿效應的變更。

已測試 Feb 15, 2026

進階技巧

親自打電話給貶損者。向給出低分的人打一通真人電話,能將30到40%的人轉變回滿意的客戶。AI可以寫腳本,但真人的電話才能發揮作用。

功能需求優先排序

為客戶的功能需求排定優先順序

Help me prioritize these feature requests from customers:

[List 10-15 feature requests with: who requested it, how many times it's been asked, customer tier/value]

Product roadmap capacity: [how many features you can build this quarter]
Strategic priorities: [company goals this year]
Competitor features: [what competitors have that you don't]
Technical constraints: [anything limiting what you can build]

Prioritize and provide:
1. A scoring matrix: impact (revenue, retention, acquisition) x effort
2. The top 5 features to build, in order, with business justification
3. Features to explicitly NOT build (and how to communicate that)
4. A 'quick wins' list: low-effort features that delight customers
5. A customer communication plan for each decision (yes, not yet, no)
6. A feedback loop: how to validate the priority order with customers before building

最適合: GEMINI

Gemini能建構帶有明確評分標準的最結構化的優先順序矩陣。影響vs工作量的分析很實用,能產生產品團隊可立即採用的框架。

已測試 Feb 15, 2026

進階技巧

問客戶「如果我們做了這個功能,您會停止使用什麼?」第一個問題揭示了真正的價值。第二個「您想要這個功能嗎?」永遠會得到無意義的「是」。

回饋循環建構

建構持續的回饋系統

Help me build a continuous customer feedback system.

Business type: [product/service/SaaS/e-commerce]
Customer base size: [number of active customers]
Current feedback sources: [surveys, support tickets, reviews, social — or none]
Team responsible: [who handles feedback]
Biggest feedback gap: [what you don't know about your customers]

Design a feedback ecosystem:
1. Feedback collection touchpoints: when and where to ask (map to customer journey)
2. Passive feedback sources: signals you can read without asking
3. A feedback aggregation system: how to centralize data from all sources
4. A weekly feedback review ritual: agenda, attendees, and action format
5. Closed-loop process: how to tell customers their feedback led to changes
6. Quarterly 'voice of customer' report template for leadership

最適合: CHATGPT

ChatGPT能設計包含實用工具推薦的最全面回饋生態系統。將回饋觸點清楚地映射到客戶旅程上,並提出現實的實施時程。

已測試 Feb 15, 2026

進階技巧

回饋必須是封閉的循環。建立收集→分析→執行→向客戶報告的循環。告訴客戶「我們聽到您的聲音了」能提升下一次回饋的品質和數量。

模型比較

基於實際測試結果 — 非假設推測。 查看測試方法

G

Gemini

Best for feedback analysis and feature prioritization. Categorizes unstructured data efficiently and builds clear scoring matrices. Less effective at writing empathetic customer-facing responses.

結果來源: Gemini 2.0 Flash · 已測試 Feb 15, 2026
C

ChatGPT

Best for review responses and feedback system design. Writes natural, personalized customer replies and maps comprehensive feedback ecosystems. Can be overly optimistic in sentiment analysis.

結果來源: GPT-4o · 已測試 Feb 15, 2026
C

Claude

Best for survey design and NPS interpretation. Creates unbiased surveys and provides honest data analysis. Sometimes over-qualifies findings — push it for clear recommendations.

結果來源: Claude 3.5 Sonnet · 已測試 Feb 15, 2026
G

Grok

Excels at cutting through complaint noise to identify the real issues customers are trying to communicate. Delivers blunt, actionable analysis without sugarcoating negative feedback patterns. Less effective at building empathetic feedback response frameworks and diplomatic survey designs.

結果來源: Grok 2 · 已測試 Feb 15, 2026

在 NailedIt 中試試

將上方的提示詞貼到 NailedIt,並排比較各模型的回應。

進階技巧

1

Ask 'why' after every rating. A 3-star rating tells you nothing. The follow-up 'what would make this a 5?' tells you everything. Always pair quantitative scores with one open-ended question that reveals the story behind the number.

2

Negative feedback is a gift — silence is the threat. Customers who complain are giving you a chance to fix things. The ones who leave without saying anything are the real problem. Build systems that make it easy to complain before they walk away.

3

Share raw feedback with the whole team. Don't filter customer quotes into sanitized summaries for your team. Let engineers, designers, and leadership read the actual words customers use. Raw emotion drives action faster than any report.