Analytics and Insights

Flagix provides comprehensive analytics to help you understand how your feature flags are being used and measure the impact of your experiments. Flagix currently shows only project wide analytics. Per-flag analytics will be available in future releases.

Overview

The analytics dashboard gives you insights into:

• Flag Usage: How often flags are evaluated
• Active and Stale flags: Which flags are active and stale
• Variation Impressions: Which variations are being used most
• Experiment Results: A/B test performance

Flag Usage Analytics

How Evaluation Tracking Works

Flag usage is tracked automatically when you call Flagix.evaluate():

// This automatically records an impression for the flag evaluation
const isEnabled = Flagix.evaluate("new-feature");

// The following data is captured:
// - Flag key: "new-feature"
// - Variation returned: true/false (or your variation value)
// - User ID: from your evaluation context
// - Timestamp: when the evaluation occurred

Note: You don't need to manually track flag evaluations. Every evaluate() call is automatically recorded in your analytics.

Usage Dashboard

The usage dashboard shows:

• Total Impressions: Number of times flags were evaluated
• Active Flags: Flags that have received traffic in the selected time range
• Stale Flags: Flags with fewer than 100 impressions (may indicate unused features)
• Variation Distribution: Which variations are being served to users
• Impression Trends: How flag usage changes over time

Key Metrics

Conversion Tracking

Understanding Conversions

Conversions allow you to measure specific user actions (like purchases or signups) in relation to feature flag variations. This is essential for A/B testing and measuring the business impact of your features.

Key Concept: You use Flagix.track() to record custom events that represent goals or conversions in your application.

Setting Up Conversion Tracking

Track important user actions in your application:

// Simple conversion tracking
Flagix.track("purchase_completed", {
  revenue: 99.99,
  currency: "USD",
  plan: "premium"
});

// Another example: signup
Flagix.track("signup_completed", {
  method: "google-oauth",
  plan: "free"
});

Important: When you call Flagix.track(), it records a conversion event. The system automatically associates this event with flag evaluations that happened before it (for the same user), which is how we calculate A/B test conversion rates.

Conversion Funnels

Track multi-step user journeys:

// Step 1: User views pricing page
Flagix.track("pricing_viewed", {
  source: "homepage"
});

// Step 2: User starts checkout
Flagix.track("checkout_started", {
  plan: "premium"
});

// Step 3: User completes purchase
Flagix.track("purchase_completed", {
  plan: "premium",
  revenue: 99.99
});

By tracking these steps, you can measure:

• How many people view pricing → start checkout
• How many start checkout → complete purchase
• Conversion rates for each step of your funnel

How Conversions Work in A/B Tests

When you run an A/B test:

// User is assigned a variation
const checkoutType = Flagix.evaluate("checkout-redesign");

// ... user interacts with the feature ...

// User completes a purchase (conversion event)
Flagix.track("purchase_completed", {
  revenue: 99.99
});

Flagix automatically:

1. Records that this user saw the checkout-redesign flag (impression)
2. Records that this user completed a purchase (conversion event)
3. Calculates the conversion rate for each variation

This is why we have both evaluate() and track():

• evaluate() records which variation users saw
• track() records what actions users completed
• Together, they calculate: "Of users who saw variation A, X% completed the goal"

Conversion Metrics

A/B Testing Analytics

Setting Up A/B Tests

To run an A/B test in Flagix, create an experiment rule on your flag that splits traffic between two or more variations. Then, track conversions:

// Assign user to a variation
const checkoutType = Flagix.evaluate("checkout-redesign");

// User interacts with the feature...
// ... user completes purchase (or your goal) ...

// Record the conversion
Flagix.track("purchase_completed", {
  revenue: 99.99,
  plan: "premium"
});

// That's it! Flagix automatically:
// 1. Recorded which variation the user saw
// 2. Recorded the conversion event
// 3. Will calculate conversion rates per variation

Understanding Experiment Results

Your A/B test results show:

{
  "variation-a": {
    "participants": 1000,      // Users who saw this variation
    "conversions": 120,        // Users who converted
    "conversionRate": 0.12,    // 12% conversion rate
    "lift": 20,                // 20% better than control
    "significance": 0.96       // 96% statistical confidence
  },
  "variation-b": {
    "participants": 1000,
    "conversions": 100,
    "conversionRate": 0.10,
    "lift": 0,                 // This is the control
    "significance": 0
  }
}

Statistical Significance

Understanding the results:

Note: Statistical significance is calculated using the Wilson score interval method for all variation sizes. With very small sample sizes (1-5 users), confidence intervals will be wide. For reliable results, aim for at least 100+ participants per variation..

Experiment Reports

Monitor your experiments in the dashboard:

• Total Participants: Users exposed to each variation
• Top Performer: Which variation has the best conversion rate
• Performance Breakdown: Detailed metrics for each variation
• Conversion Rate Trend: How conversion rates change over time

The system automatically calculates lift and statistical confidence using a Z-test for proportions.

Real-time Analytics

Real-time Flag Updates

Flagix uses Server-Sent Events (SSE) to push flag updates to your client in real-time. When you update a flag in the dashboard, connected clients receive the update instantly:

// The Flagix client automatically listens for flag updates
// When a flag changes, it emits a 'flagUpdate' event

Flagix.on('flagUpdate', (flagKey) => {
  console.log(`Flag '${flagKey}' was updated`);
  
  // You can now re-evaluate the flag with the new configuration
  const newValue = Flagix.evaluate(flagKey);
  console.log(`New value for ${flagKey}: ${newValue}`);
});

This ensures your application always uses the latest flag configurations without requiring a page reload.

Performance Monitoring

Monitor your SDK performance locally:

// Measure evaluation performance
const startTime = performance.now();
const result = Flagix.evaluate("feature-flag");
const endTime = performance.now();

const evaluationTime = endTime - startTime;
console.log(`Flag evaluation took ${evaluationTime}ms`);

// Flagix evaluations are cached locally, so they should be very fast
// First initialization may take slightly longer due to API calls

Analytics API

Accessing Analytics Data

Analytics data is available through the Flagix dashboard.

To access analytics:

1. Via Dashboard: Visit your project's Analytics section in the Flagix UI
2. Usage Metrics: the analytics tab shows flag usage, impressions, and variation distribution
3. A/B Test Results: the analytics tab shows experiment performance and conversion rates

Supported Time Ranges

• 7d: Last 7 days (default)
• 30d: Last 30 days
• 3m: Last 3 months