Compare User Journeys

Compare User Journeys

When to Use

• Understand why some users convert and others don't
• Identify behaviors that predict retention or churn
• Find what makes power users different from average users
• Debug why a specific cohort or segment underperforms

Instructions

Step 1: Define Segments to Compare

Ask the user which segments to compare. Common comparisons:

• Converters vs. non-converters (free trial → paid)
• Retained vs. churned (active after 30 days vs. not)
• Power users vs. casual users (top 10% engagement vs. median)
• High LTV vs. low LTV (revenue-based segments)
• Activated vs. not activated (completed onboarding vs. dropped off)

Step 2: Define Journey Window

• Time window: First 7 days? First 30 days? Last 90 days?
• Key outcome event: What defines "success"? (e.g., "Purchase Completed", "Day 7 Active")

Step 2.5: Define Segment Filters

If comparing segments that don't exist as saved cohorts in Mixpanel, define them using query filters:

Common segment definitions:

Converters vs. Non-converters:

• Segment A (Converters): Users who did "Purchase Completed" in time window
• Segment B (Non-converters): Users who did NOT do "Purchase Completed"
• Implementation: Use separate queries with event filters

Power users vs. Casual users:

• Segment A (Power): Users with high engagement (e.g., 20+ sessions in 30 days)
• Segment B (Casual): Users with low engagement (e.g., <5 sessions in 30 days)
• Implementation: Filter by user profile property or event frequency

Retained vs. Churned:

• Segment A (Retained): Users who returned (did return event in retention window)
• Segment B (Churned): Users who didn't return
• Implementation: Use retention query or event occurrence filters

How to apply segment filters in queries:

For event-based segments:

report={
  "metrics": [...],
  "filters": [
    {
      "property": "event_name",
      "operator": "equals",
      "value": "Purchase Completed"
    }
  ]
}

For user profile segments:

report={
  "where": "user[\"plan_type\"] == \"paid\""
}

For behavioral segments (did/didn't do event):

• Run separate queries for each segment
• One with the event present, one without
• Compare the results side-by-side

Pro tip: If these segments will be used frequently, ask the user to create saved cohorts in Mixpanel first, then reference them by cohort ID in queries.

Step 3: Query Both Segments

For each segment, get:

Behavioral data:

• Top 10 most common events
• Event frequency (events per user)
• Event sequences (which events happen before the outcome)
• Feature adoption (which features do they use)

Timing data:

• Time to first key action
• Session frequency
• Days active in the window

Use Mixpanel queries:

• Run-Query with segment filters
• Flow analysis to see paths
• Funnel analysis for conversion steps

Step 4: Identify Differentiators

Compare the segments to find:

Behavioral differences:

• Events that one group does significantly more (2x+ difference)
• Events unique to one group
• Features heavily used by one group but not the other

Timing differences:

• How quickly do they take key actions?
• How frequently do they engage?
• Do they use the product in bursts or consistently?

Path differences:

• Do successful users follow a specific sequence?
• Where do unsuccessful users drop off?

Step 5: Present Findings

Structure output as:

Segment Comparison: [Segment A] vs [Segment B]

Key Differentiators

Unique to Segment A (successful users):

• Event X (used by 70% of Segment A, 5% of Segment B)
• Event Y (used by 60% of Segment A, 10% of Segment B)

Unique to Segment B (unsuccessful users):

• Event Z (error or friction signal)

Recommended Actions

1. Guide more users to [key differentiator behavior]
2. Reduce friction around [drop-off point]
3. Highlight [underused feature] to [unsuccessful segment]

Supporting Data

• Include flow diagrams showing path differences
• Link to Mixpanel reports for each segment

Example Interaction

User: "Why do some trial users convert and others don't?"

Assistant:

1. Asks: "What's your trial length and what defines conversion?"
2. User says: "14-day trial, conversion = upgraded to paid"
3. Runs queries:
   • Segment A: Users who converted (last 90 days)
   • Segment B: Users who didn't convert (last 90 days)
   • Compares events, features, timing
4. Finds differentiators:
   • Converters use "Export" feature 5x more (80% vs 15%)
   • Converters invite teammates 3x more (60% vs 20%)
   • Converters hit "wow moment" within 48h (median: 1 day vs 8 days)
5. Outputs:
   • High priority: Accelerate time to "wow moment" (onboarding flow)
   • Medium: Encourage Export feature trial (add tooltip, email)
   • Low: Prompt team invites earlier (could drive virality)

Common Journey Comparisons

Trial converters vs. non-converters

• What features drive conversion?
• When do converters realize value?
• What blockers prevent conversion?

Retained vs. churned

• What habits do retained users form?
• Where do churned users disengage?
• What's the "point of no return" (predict churn)?

Power users vs. casual

• What makes someone a power user?
• Can casual users be activated into power users?
• What features unlock high engagement?

High LTV vs. low LTV

• Which features correlate with revenue?
• Do high LTV users have different onboarding?
• What expansion behaviors exist?

Best Practices

• Use large enough samples (100+ users per segment for confidence)
• Control for time (compare users in same time window)
• Look for causal patterns (not just correlation)
• Focus on early behaviors (leading indicators, not lagging)
• Quantify differences (2x, 5x, 10x — not just "higher")
• Suggest experiments (test whether guiding users to do X improves outcomes)