Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.grainql.com/llms.txt

Use this file to discover all available pages before exploring further.

​
What is A/B Testing?

A/B testing shows different versions of your app to different users, then measures which performs better. Version A might have a blue button, version B a green button - data shows which gets more clicks. Why A/B test?
  • Remove guesswork from design decisions
  • Improve conversion rates with data
  • Validate changes before rolling out to everyone

​
Simple A/B Test

Start with a basic two-variant test: 
import { useConfig, useTrack } from '@grainql/analytics-web/react';

function Hero() {
  const { value: variant } = useConfig('hero_variant');
  const track = useTrack();
  
  const variantA = {
    title: 'Welcome to Our App',
    subtitle: 'The best tool for productivity',
    buttonText: 'Get Started'
  };
  
  const variantB = {
    title: 'Boost Your Productivity',
    subtitle: 'Join thousands of happy users',
    buttonText: 'Start Free Trial'
  };
  
  const content = variant === 'B' ? variantB : variantA;
  
  const handleClick = () => {
    track('cta_clicked', { variant: variant || 'A' });
  };
  
  return (
    <div>
      <h1>{content.title}</h1>
      <p>{content.subtitle}</p>
      <button onClick={handleClick}>{content.buttonText}</button>
    </div>
  );
}
Set up in Grain Dashboard:
  1. Create config key: hero_variant
  2. Set values: A for 50% of users, B for the other 50%
  3. Track which variant converts better

​
Multi-Variant Testing

Test more than two options: 
function PricingCard() {
  const { value: pricingVariant } = useConfig('pricing_variant');
  
  const variants = {
    control: { price: 29, label: 'Pro Plan' },
    test_1: { price: 24, label: 'Pro Plan - Special' },
    test_2: { price: 29, label: 'Professional' },
    test_3: { price: 27, label: 'Pro Plan - Limited' }
  };
  
  const { price, label } = variants[pricingVariant || 'control'];
  
  return (
    <div>
      <h3>{label}</h3>
      <p>${price}/month</p>
    </div>
  );
}
Test multiple variants to find the optimal price and messaging.

​
Component-Based Testing

Swap entire components: 
import { useConfig } from '@grainql/analytics-web/react';

function LandingPage() {
  const { value: heroVersion } = useConfig('hero_version');
  
  const HeroComponent = {
    'minimal': HeroMinimal,
    'detailed': HeroDetailed,
    'video': HeroWithVideo
  }[heroVersion || 'minimal'];
  
  return (
    <>
      <HeroComponent />
      <Features />
      <Pricing />
    </>
  );
}
Test completely different design approaches.

​
Tracking Test Results

Track when users see and interact with variants: 
function ABTest() {
  const { value: variant } = useConfig('test_variant');
  const track = useTrack();
  
  // Track impression
  useEffect(() => {
    track('test_viewed', {
      test_name: 'hero_cta',
      variant: variant || 'A'
    });
  }, [variant, track]);
  
  // Track conversion
  const handleConversion = () => {
    track('test_converted', {
      test_name: 'hero_cta',
      variant: variant || 'A'
    });
    
    // Conversion action (signup, purchase, etc.)
  };
  
  return <button onClick={handleConversion}>Convert</button>;
}
Now you can compare conversion rates: test_converted / test_viewed for each variant.

​
Personalized Testing

Combine A/B testing with user segmentation: 
function PersonalizedHero() {
  const { value: variant } = useConfig('hero_variant', {
    properties: {
      plan: userPlan,
      signup_date: userSignupDate
    }
  });
  
  // New users see variant B, existing users see variant A
  return <HeroComponent variant={variant} />;
}
Set rules in the dashboard to show different variants based on user properties.

​
Button Style Test

Test button designs: 
function CTAButton() {
  const { value: buttonStyle } = useConfig('button_style');
  const track = useTrack();
  
  const styles = {
    default: {
      backgroundColor: '#007bff',
      borderRadius: '4px',
      padding: '12px 24px'
    },
    bold: {
      backgroundColor: '#ff4500',
      borderRadius: '8px',
      padding: '16px 32px',
      fontWeight: 'bold'
    },
    minimal: {
      backgroundColor: 'transparent',
      border: '2px solid #007bff',
      borderRadius: '4px',
      padding: '12px 24px'
    }
  };
  
  const style = styles[buttonStyle || 'default'];
  
  const handleClick = () => {
    track('cta_button_clicked', { button_style: buttonStyle || 'default' });
  };
  
  return (
    <button style={style} onClick={handleClick}>
      Sign Up
    </button>
  );
}

​
Feature Launch Testing

Test new features with a subset of users: 
function Dashboard() {
  const { value: newDashboard } = useConfig('new_dashboard_enabled');
  
  if (newDashboard === 'true') {
    return <NewDashboard />;
  }
  
  return <LegacyDashboard />;
}
Start with 10% of users, gradually increase if metrics improve.

​
Best Practices

1. Test One Thing: Change one element at a time to know what caused the difference. 
// ✅ Good: Testing one element
const buttonText = variant === 'B' ? 'Start Free' : 'Get Started';

// ❌ Bad: Too many changes
const content = variant === 'B' 
  ? { text: 'Start Free', color: 'red', size: 'large' }
  : { text: 'Get Started', color: 'blue', size: 'medium' };
2. Track Both Impressions and Conversions: Know how many saw each variant and how many converted. 3. Run Tests Long Enough: At least 1-2 weeks to account for day-of-week variations. 4. Significant Sample Size: Need enough users to draw conclusions (typically 1000+ per variant). 5. Document Tests: Track what you’re testing and why in your dashboard.

​
Analyzing Results

After running your test, compare metrics in the Grain Dashboard:
  • Impressions: How many saw each variant
  • Conversions: How many completed the goal action
  • Conversion Rate: Conversions / Impressions
  • Statistical Significance: Is the difference real or luck?

​
Gradual Rollout

Once you have a winner, roll it out gradually: 
// Dashboard config progression:
// Week 1: 50% A, 50% B (testing)
// Week 2: 25% A, 75% B (B is winning)
// Week 3: 10% A, 90% B (confirming)
// Week 4: 0% A, 100% B (full rollout)
This catches any issues before full deployment.

​
Next Steps

Remote Config

Learn about configuration management

Feature Flags

Control features with flags

Analytics Dashboard

Build analytics dashboards