Investigation Patterns - What to Ask Your AI Analyst
Investigation Patterns: What to Ask Your AI Analyst
The Secret: Think Like an Executive, Not a Query Writer
Don't ask for data. Ask for insights, investigations, and understanding.
Core Investigation Patterns
1. The "Why" Investigation
Purpose: Uncover root causes of changes or issues
Pattern: "Why did [metric] [change]?"
Examples:
- "Why did revenue drop last month?"
- "Why are conversion rates improving?"
- "Why did customer satisfaction decrease?"
- "Why are costs rising faster than revenue?"
What Scoop Does: Multi-probe investigation → root cause identification → action plan
2. The "Analyze" Investigation
Purpose: Discover patterns and segments you don't know exist
Pattern: "Analyze my [entity/area]"
Examples:
- "Analyze my customer base"
- "Analyze sales performance"
- "Analyze product usage patterns"
- "Analyze team productivity"
What Scoop Does: ML-powered discovery → hidden segments → strategic insights
3. The "What Drives" Investigation
Purpose: Find non-obvious factors influencing outcomes
Pattern: "What drives [outcome]?"
Examples:
- "What drives customer retention?"
- "What drives deal size?"
- "What drives product adoption?"
- "What drives team performance?"
What Scoop Does: Statistical analysis → correlation discovery → predictive factors
4. The "Find Patterns" Investigation
Purpose: Uncover hidden patterns and anomalies
Pattern: "Find patterns in [data/behavior]"
Examples:
- "Find patterns in customer churn"
- "Find patterns in sales cycles"
- "Find patterns in support tickets"
- "Find patterns in usage data"
What Scoop Does: Pattern recognition → anomaly detection → business implications
5. The "Compare" Investigation
Purpose: Understand differences and changes
Pattern: "Compare [A] to [B]"
Examples:
- "Compare this quarter to last quarter"
- "Compare high-value customers to others"
- "Compare successful deals to lost deals"
- "Compare top performers to average"
What Scoop Does: Multi-dimensional comparison → key differentiators → insights
Advanced Investigation Patterns
The Evolution Investigation
Purpose: Track how things change over time
Examples:
- "How has customer behavior evolved?"
- "How are our sales patterns changing?"
- "Show me pipeline evolution over time"
- "Track how usage patterns have shifted"
Power: Uses snapshot data to show progression, not just current state
The Prediction Investigation
Purpose: Anticipate future outcomes
Examples:
- "Predict which customers might churn"
- "What will revenue look like next quarter?"
- "Which deals are likely to close?"
- "Forecast demand by product"
Power: ML models trained on your specific patterns
The Segmentation Investigation
Purpose: Find natural groupings in your data
Examples:
- "Segment my customers"
- "Group products by performance patterns"
- "Cluster sales territories"
- "Find user behavior segments"
Power: Discovers segments you didn't know existed
The Diagnostic Investigation
Purpose: Diagnose complex business issues
Examples:
- "Diagnose pipeline health"
- "Why is growth slowing?"
- "What's wrong with our conversion funnel?"
- "Diagnose customer satisfaction issues"
Power: Systematic investigation across all factors
Industry-Specific Investigation Starters
SaaS / Technology
- "Why is MRR growth slowing?"
- "Analyze churn patterns by cohort"
- "What drives expansion revenue?"
- "Find early indicators of account risk"
- "Compare feature adoption across segments"
E-Commerce / Retail
- "Why did cart abandonment increase?"
- "Analyze customer lifetime value patterns"
- "What drives repeat purchase behavior?"
- "Find seasonal patterns I should plan for"
- "Compare channel performance deeply"
Sales Organizations
- "Why are sales cycles lengthening?"
- "Analyze win/loss patterns"
- "What differentiates top performers?"
- "Find bottlenecks in our sales process"
- "Predict which deals will close"
Marketing Teams
- "Why did CAC increase?"
- "Analyze campaign performance patterns"
- "What drives conversion at each funnel stage?"
- "Find the optimal channel mix"
- "Compare audience segment behavior"
Financial Analysis
- "Why are margins shrinking?"
- "Analyze expense patterns"
- "What drives profitability by segment?"
- "Find cost optimization opportunities"
- "Compare budget vs actual deeply"
The Power of Vague Questions
Don't Be Too Specific!
Scoop's AI thrives on investigative freedom.
Too Specific: "Show me revenue by product by month for Q1" Better: "Why did Q1 revenue disappoint?"
Too Specific: "List customers who churned last month" Better: "Analyze recent churn patterns"
Too Specific: "Calculate average deal size by rep" Better: "What drives sales performance?"
Conversation Patterns
The Progressive Investigation
Start broad, then narrow based on discoveries:
- "Analyze my business performance"
- (Scoop finds issue in customer segment)
- "Why is that segment struggling?"
- (Scoop finds product-related cause)
- "What should we do about it?"
The Multi-Angle Investigation
Approach problems from different angles:
- "Why are we losing customers?"
- "What do churned customers have in common?"
- "Compare churned vs retained customers"
- "Predict who might churn next"
- "What interventions would help most?"
The Time-Based Investigation
Understand evolution and trends:
- "How has our business changed over time?"
- "What's different about recent months?"
- "When did these changes start?"
- "What coincided with these changes?"
- "Project these trends forward"
Power User Secrets
1. Let Scoop Surprise You
- "What's interesting in my data?"
- "What should I know about my business?"
- "Find something surprising"
- "What am I missing?"
2. Use Business Language
- Not: "SELECT SUM(revenue) GROUP BY..."
- But: "How's business doing?"
- Not: "Join customers with orders where..."
- But: "Tell me about customer behavior"
3. Ask About Problems, Not Metrics
- Not: "Show me the churn rate"
- But: "Why are we losing customers?"
- Not: "Calculate revenue by segment"
- But: "Which segments drive growth?"
4. Trust the Investigation Process
- Don't pre-filter or constrain
- Let Scoop determine what's relevant
- Ask follow-ups on discoveries
- Build on what you learn
Common Mistakes to Avoid
❌ Thinking Like SQL
"SELECT * FROM customers WHERE..." → "Analyze my customer patterns"
❌ Being Too Prescriptive
"Show me exactly X grouped by Y filtered by Z" → "Help me understand what's happening with [area]"
❌ Asking for Single Metrics
"What's the churn rate?" → "Why are customers leaving?"
❌ Ignoring Suggestions
Scoop suggests follow-ups for a reason - they lead to insights
Your First Week of Investigations
Day 1: Start with Why
Pick your biggest current concern and ask why it's happening
Day 2: Analyze Something
Choose a business area and let Scoop analyze it completely
Day 3: Find Drivers
Ask what drives your most important metric
Day 4: Compare Periods
Have Scoop compare your recent performance to historical
Day 5: Discover Segments
Let Scoop find hidden segments in your customers/products/data
Day 6: Predict Something
Ask Scoop to predict an important future outcome
Day 7: Get Strategic
Ask: "What should I focus on to improve performance?"
The Bottom Line
Stop thinking about queries. Start thinking about questions that matter to your business. Scoop isn't waiting for perfect syntax - it's waiting to investigate your most important business challenges.
Your AI analyst is ready. What mystery should it solve first?
[Start Your Investigation] | [Watch Investigation Demo]
Updated about 24 hours ago