Clustering automatically discovers natural groupings in your data—customer segments, behavior patterns, or performance tiers—that you might not have thought to look for. Scoop's AI handles the complex math and delivers clear, actionable segment descriptions in plain English.

What Is Clustering?

Clustering vs. Filtering

Clustering considers combinations of attributes across multiple columns simultaneously, finding patterns you might never have thought to look for.

Tip: Use clustering when you want to discover "unknown unknowns"—patterns you didn't know existed in your data.

How Clustering Works in Scoop

The Workflow

Select Dataset → Click "Find Common Groups" → AI Analyzes → Segments Revealed
      ↓                    ↓                        ↓              ↓
   Your data          One click              EM algorithm    Named segments
                                            + Rules models   with descriptions

Step-by-Step

1. Select Your Dataset: Load the dataset to analyze (deals, customers, tickets, etc.)
2. Click "Find Common Groups": Access from Data Science Studio tab
3. AI Builds Clusters: Scoop determines optimal number of clusters automatically
4. Clusters Named: AI translates statistical patterns into human descriptions
5. Results Delivered: See summaries, profiles, and defining rules

What Happens Behind the Scenes

Tip: You don't need to understand the algorithms. Scoop handles variable selection, scaling, and cluster assignment automatically.

Understanding Cluster Results

Cluster Profiles

Each cluster includes:

How Clusters Are Named

Scoop creates meaningful names from the data:

Names are derived from actual patterns—not random labels.

Cluster Rules

Each cluster comes with defining rules:

Cluster: "Fast Closers"
Rules:
- Deal Size < $50K
- Sales Cycle < 30 days
- Decision Makers = 1-2
- Industry IN (Tech, Retail)

These rules tell you exactly why records belong to this segment.

Example: Sales Opportunity Clusters

Sample Results

After running clustering on opportunities:

What Each Cluster Tells You

Example: Support Ticket Clusters

Sample Results

Insights

• Quick Resolutions: Document common fixes for knowledge base
• Escalated Criticals: Route to senior agents immediately
• Low Engagement: Implement proactive outreach
• Complex Investigations: Allocate specialist time

Example: Customer Segmentation

Sample Results

Actions by Segment

Using Cluster Results

Filter by Cluster

Once clusters are identified, use them as filters:

• View all records in a specific cluster
• Compare metrics across clusters
• Track cluster distribution over time

Export for Action

Track Changes

Monitor how clusters evolve:

• Are "At Risk" customers increasing?
• Is "Quick Wins" cluster growing?
• Movement between clusters over time

Best Practices

When to Use Clustering

Data Preparation

Interpreting Results

Common Applications

Troubleshooting

Too Many/Few Clusters

Scoop automatically determines optimal cluster count. If results seem off:

• Check if data has clear natural groupings
• Verify data quality and completeness
• Consider filtering to relevant subset

Clusters Don't Make Sense

If segments aren't meaningful:

• Review which columns are included
• Check for data quality issues
• Validate with domain experts

Clusters Too Similar

If segments overlap significantly:

• Data may not have strong natural groupings
• Consider adding more discriminating attributes
• Focus on the most distinct clusters

Related Topics

• Automated Dataset Analysis - Full AI analysis workflow
• Understanding Recipes - Pre-built analytics templates
• Process Analysis - Visualize flows
• Creating KPIs - Custom metrics by segment