AI 提示詞： 數據分析與洞察

Design a marketing dashboard for [business type].

Channels I use: [list all active marketing channels]
Reporting frequency: [daily / weekly / monthly]
Dashboard audience: [CEO / marketing team / client / board]
Current tools: [Google Analytics, HubSpot, Semrush, etc.]
Biggest question leadership always asks: [what they want to know]
Current pain point: [what's wrong with your existing reporting]

Design:

1. TOP ROW: 5-7 KPIs that matter most
   - Metric name, definition, and target benchmark
   - Visualization type (number with trend arrow, gauge, sparkline)
   - Red/yellow/green thresholds
   - Why each KPI is here (what decision does it inform?)

2. CHANNEL BREAKDOWN
   - Supporting metrics organized by channel
   - Comparison views: this period vs. last period vs. goal
   - The exact chart type for each metric (line for trends, bar for comparison, pie for composition)

3. EARLY WARNING INDICATORS
   - 3 leading metrics that predict problems 2-4 weeks before they hit revenue
   - Alert thresholds: when to investigate vs. when to act
   - What each indicator means when it moves

4. REPORT AUTOMATION
   - Which data sources feed into this dashboard
   - Refresh frequency per data source
   - A 3-minute walkthrough script for presenting this dashboard
   - Questions to pre-answer before every reporting meeting

5. ANTI-VANITY METRIC CHECK
   - Which commonly tracked metrics should NOT be on this dashboard and why
   - The difference between a metric that feels important and one that drives a decision

Analyze these campaign results and tell me exactly what to do next.

[Paste campaign data: impressions, clicks, conversions, spend, CTR, CPC, ROAS by day/week]

Campaign type: [paid ads / email / social / content / SEO]
Campaign goal: [awareness / leads / sales / engagement]
Total budget: [spend]
Time period: [how long the campaign ran]
Benchmarks: [industry averages or your historical averages]
What you expected: [what you thought would happen]

Analyze across 4 dimensions:

1. PERFORMANCE DIAGNOSIS
   - What worked and what didn't (specific, with data evidence)
   - The single biggest lever to improve results
   - Trend analysis: is performance improving, declining, or plateauing?
   - Fatigue indicators: when did performance start dropping off?

2. BUDGET OPTIMIZATION
   - Current spend allocation vs. recommended reallocation (with exact percentages)
   - Which segments/audiences/creatives deserve more budget
   - Which should be paused or killed
   - Expected impact of reallocation on key metrics

3. HYPOTHESIS GENERATION
   - 3 specific hypotheses for why underperforming elements failed
   - How to test each hypothesis in the next campaign cycle
   - What additional data would help diagnose issues you can't explain

4. ACTION PLAN
   - Stop: what to kill immediately (with reasoning)
   - Start: new approaches to test based on what the data suggests
   - Continue: what to keep doing and why
   - Timeline: specific next steps for the next 7 and 30 days

Help me understand which marketing channels actually drive my results.

Channels and monthly spend:
[Channel 1]: $[spend] → [conversions attributed]
[Channel 2]: $[spend] → [conversions attributed]
[Channel 3]: $[spend] → [conversions attributed]
[Channel 4]: $[spend] → [conversions attributed]

Sales cycle: [average days from first touch to conversion]
Conversion tracking: [what I can currently measure and where I have gaps]
CRM/tools: [platforms I use]
Biggest attribution confusion: [the specific question I can't answer]
Business type: [B2B / B2C / ecommerce / SaaS / services]

Advise:

1. ATTRIBUTION MODEL SELECTION
   - Which model fits my business (first-touch, last-touch, linear, time-decay, position-based)
   - Why this model over others for my sales cycle length and channel mix
   - What each model would tell me vs. miss

2. IMPLEMENTATION PLAN
   - How to set this up in my current tools (specific steps)
   - UTM parameter strategy for consistent tracking
   - What data I need to start collecting now

3. CHANNEL INTERACTION ANALYSIS
   - Likely assist channels (touch points that rarely convert directly but enable other channels)
   - Channels that look good in last-touch but overstate their impact
   - Channels that look bad but are undervalued

4. DARK FUNNEL HANDLING
   - How to account for channels that can't be directly attributed (word of mouth, brand, podcast, PR)
   - Proxy metrics and survey-based attribution methods
   - 'How did you hear about us?' survey design

5. DECISION FRAMEWORK
   - How to present attribution findings to stakeholders who want simple answers
   - Budget reallocation recommendations based on this analysis
   - When to revisit and recalibrate the model

Build an AI visibility monitoring system for my brand.

Brand/company: [name and what you do]
Core topics: [5-10 topics where you want to be cited]
Key competitors: [who currently gets cited instead of you]
Current AI visibility: [do ChatGPT / Perplexity / Google AI Overviews mention you? Where?]
Content assets: [what you've published that AI could reference]

Build a tracking system:

1. QUERY INVENTORY
   - 20 high-value queries to monitor monthly (the questions your ideal customer asks AI)
   - For each: which AI platforms to check (ChatGPT, Perplexity, Google AI Overview, Bing Copilot)
   - Categorize: brand mentioned / competitor mentioned / neither mentioned

2. BASELINE AUDIT
   - Run all 20 queries now across platforms
   - Document: who gets cited, what sources are referenced, what claims are made
   - Identify your 'citation gap': queries where competitors appear and you don't

3. CITATION SIGNAL ANALYSIS
   - What content your competitors have that's getting them cited (Reddit posts, blog articles, YouTube videos, Wikipedia mentions)
   - What content you need to create to earn citations
   - Which platforms feed most into each AI model (Reddit → ChatGPT, YouTube → Gemini, etc.)

4. MONTHLY TRACKING ROUTINE (15 minutes)
   - Spreadsheet template with queries, platforms, and citation status
   - How to efficiently check all 20 queries across platforms
   - What changes to track month-over-month
   - When to escalate (competitor gains, your citations disappear, misinformation)

5. ACTION PRIORITIES
   - Top 5 content pieces to create or update for maximum citation impact
   - Platform-specific strategies (Reddit answers, YouTube descriptions, structured blog content)
   - Timeline: realistic expectations for when citations start appearing

Score my content portfolio and tell me where to focus.

Content inventory (for each piece, include what you have):
[Paste a table: Title, URL, Monthly Traffic, Conversions, Bounce Rate, Avg Time on Page, Backlinks, Social Shares]

Business goals: [what content should ultimately drive]
Time available for content improvement: [hours per week]

Score each piece across 4 dimensions:

1. TRAFFIC PERFORMANCE (25 pts)
   - Organic traffic trend (growing, stable, declining)
   - Traffic relative to effort invested
   - Keyword rankings and position trajectory
   - Click-through rate from search results

2. ENGAGEMENT QUALITY (25 pts)
   - Time on page vs. content length (are people actually reading?)
   - Bounce rate in context (informational pages bounce high — that's okay)
   - Social shares and backlinks earned
   - Comments and direct responses

3. BUSINESS IMPACT (25 pts)
   - Conversions attributed to this content
   - Role in the customer journey (top of funnel vs. decision stage)
   - Revenue influence (direct and assisted)
   - Email signups or other micro-conversions

4. COMPETITIVE POSITION (25 pts)
   - How this ranks vs. competitor content on the same topic
   - Content freshness (last updated date vs. competitors)
   - Unique value that competitors can't easily replicate
   - AI Overview vulnerability: is AI summarizing this topic away?

Deliver:
- Ranked list: top performers, underperformers, and candidates for retirement
- Top 5 pieces to update (highest ROI per hour invested)
- Top 3 content gaps (topics you should cover but don't)
- Kill list: content that should be consolidated, redirected, or removed

Find and fix the leaks in my conversion funnel.

Funnel stages with conversion rates:
[Stage 1]: [name] — [number entering] — [conversion rate to next stage]
[Stage 2]: [name] — [number entering] — [conversion rate to next stage]
[Stage 3]: [name] — [number entering] — [conversion rate to next stage]
[Stage 4]: [name] — [number entering] — [final conversion rate]

Industry: [your industry]
Average order value / deal size: [revenue per conversion]
Traffic sources: [where visitors come from, with volume per source]
Known friction points: [what you already suspect is causing drop-off]

Analyze:

1. LEAK IDENTIFICATION
   - Which stage has the biggest absolute leak (most people lost)
   - Which stage has the biggest rate leak (worst conversion rate vs. benchmark)
   - Which leak, if fixed, would generate the most revenue? (this is the priority)

2. BENCHMARK COMPARISON
   - How my rates compare to industry averages at each stage
   - Which stages are healthy and which need attention
   - Segment analysis: do specific traffic sources have much worse funnel flow?

3. DIAGNOSIS per leaky stage
   - 3 most likely causes of drop-off (with reasoning)
   - Micro-conversion additions between stages to pinpoint friction
   - User experience issues to investigate (page load, form length, clarity)

4. FIX PLAN
   - Top 3 fixes for the biggest leak, ranked by effort vs. impact
   - Expected improvement range for each fix (realistic, with confidence level)
   - A/B test designs to validate each fix before full rollout

5. 2-WEEK EXPERIMENT
   - A specific experiment to improve the weakest stage by 15%+
   - What to measure, when to measure, and how to know if it worked
   - Fallback plan if the experiment doesn't move the needle

Help me build and interpret a cohort analysis for [business/product].

Cohort definition: [how to group users — signup month, acquisition channel, plan type, first purchase category]
Key metric to track: [retention rate / revenue per user / feature adoption / repeat purchase rate]
Time period: [how far back to analyze]
Data I have: [describe available data fields and where they live]
Goal: [reduce churn / increase LTV / improve activation / optimize channel spend]

Build:

1. COHORT TABLE DESIGN
   - Row structure: what defines each cohort
   - Column structure: time periods to track
   - The exact formulas to calculate cohort metrics (for Sheets/Excel/SQL)
   - Color-coding rules for quick pattern recognition

2. HOW TO READ THE TABLE
   - What healthy cohort curves look like vs. warning signs
   - The 'banana chart' visualization and how to interpret it
   - Specific patterns to look for (early churn, delayed activation, seasonal effects)

3. INSIGHT EXTRACTION
   - 5 questions to answer from this cohort data
   - What to compare: channel cohorts vs. time cohorts vs. behavior cohorts
   - How to identify your 'best customers' by cohort behavior
   - Signals that predict which new users will become high-value

4. ACTION FRAMEWORK
   - If early churn is high: specific interventions for the first 7/30/90 days
   - If retention curves flatten late: expansion and upsell strategies
   - If certain cohorts dramatically outperform: how to acquire more of those users
   - Budget reallocation based on true cohort LTV (not just acquisition cost)

5. PRESENTATION FORMAT
   - How to visualize cohort findings for different audiences (executive vs. team)
   - The 3 most compelling slides from this analysis
   - How to update this analysis monthly in under 30 minutes

Generate my monthly marketing report from this data.

[Paste key metrics: traffic, leads, conversions, revenue, spend by channel, month-over-month changes]

Reporting month: [month/year]
Previous month data: [for comparison]
Goals for this month: [targets that were set]
Report audience: [CEO / board / marketing team / client]
Key context: [anything unusual this month — campaigns launched, market changes, team changes]

Create:

1. EXECUTIVE SUMMARY (4 sentences max)
   - Lead with the headline: did we hit goals or not?
   - The single most important insight
   - The single biggest concern
   - One-sentence recommendation

2. GOALS vs. ACTUALS TABLE
   - Each goal with target, actual, and status (green/yellow/red)
   - Trend arrows showing direction
   - Brief explanation for any red or yellow items

3. CHANNEL PERFORMANCE
   - Each channel: spend, results, ROAS/CPA, trend vs. last month
   - Highlight the best-performing and worst-performing channels
   - Budget reallocation recommendation if applicable

4. TOP 3 WINS (with evidence)
   - What worked, why it worked, and how to replicate it
   - Specific data points that prove the win

5. TOP 3 CONCERNS (with recommended actions)
   - What's not working and what you plan to do about it
   - Don't just flag problems — propose solutions with timelines

6. NEXT MONTH PREVIEW
   - Planned initiatives and expected outcomes
   - Risks and dependencies
   - Resource or budget requests (if any)

7. ONE CHART
   - The single most compelling visualization from this month's data
   - What it shows and why it matters
   - How to present it in 30 seconds