How is this different from Google Optimize?

Google Optimize was sunset in 2023. Experiment Flow is a modern replacement with better multi-armed bandit support, faster setup, and more affordable pricing.

Do I need to install anything?

Just add one script tag to your site. No npm packages, no build steps, no dependencies.

How does statistical significance work?

We use two-proportion z-tests with 95% confidence intervals. Results only show as significant when there is less than 5% chance the difference is due to randomness.

Can I use this server-side?

Yes. The REST API works with any language. There are official SDKs for Node.js, Python, Go, and Ruby.

Documentation

Getting Started

1. Sign up for an account and get your API key from the dashboard.

2. Add the SDK to your site:

<script src="proxy.php?url=https%3A%2F%2Fexperimentflow.com%2Fsdk.js"
        data-optimize-key="YOUR_API_KEY"></script>

Creating Experiments

Create experiments via the dashboard or API:

curl -X POST https://experimentflow.com/api/experiments \
  -H "X-API-Key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"name": "CTA Test", "variants": ["control", "variant_a", "variant_b"]}'

JavaScript SDK

Get Variant

optimize.getVariant(experimentId, function(variant) {
  console.log('Assigned to:', variant);
});

Track Conversion

optimize.convert(experimentId, value, goalName);

DOM Changes

optimize.activate(experimentId, {
  'variant_a': {
    '.headline': { text: 'New Headline' },
    '.cta': { style: { backgroundColor: 'green' } }
  }
});

AI Personalization

Neural contextual bandits that learn which content converts best for each user segment in real-time.

Get Personalized Ranking

Rank actions based on user context. Returns actions sorted by predicted conversion probability:

optimize.personalize('hero-cta', {
  context: {
    device: 'mobile',
    referrer: 'google',
    returning_user: true
  },
  actions: [
    { id: 'free-trial', features: { urgency: 'low' } },
    { id: 'book-demo', features: { urgency: 'high' } },
    { id: 'see-pricing', features: { urgency: 'medium' } }
  ]
}, function(result) {
  // result.ranking = actions sorted by probability
  // result.event_id = use for reward tracking
  showCTA(result.ranking[0]);
});

Send Reward Signal

When the user converts, send a reward to train the model:

// After user clicks CTA or converts
optimize.personalizeReward(eventId, 1.0);

// Or use the last personalization event automatically
optimize.personalizeReward(null, 1.0);

Quick Helper: Get Best Action

optimize.getBestAction('hero-cta', context, actions, function(bestAction, eventId) {
  displayAction(bestAction);

  // Track when user converts
  button.onclick = function() {
    optimize.personalizeReward(eventId, 1);
  };
});

Auto-Reward on Click

// Automatically send reward when user clicks any matching element
optimize.autoReward('.cta-button', 1.0);

Multi-Armed Bandits

For automatic optimization using Thompson Sampling:

// Get best arm
optimize.selectArm(banditId, function(arm) {
  showVariant(arm);
});

// Record reward (e.g., on conversion)
optimize.reward(banditId, 1);

Shopify Integration

Experiment Flow provides first-class Shopify support with automatic add-to-cart tracking, purchase tracking, and specialized A/B testing helpers.

Installation

Add the SDK to your theme.liquid file, just before the closing </head> tag:

<script src="proxy.php?url=https%3A%2F%2Fexperimentflow.com%2Fsdk.js"
        data-optimize-key="YOUR_API_KEY"></script>
<script>
  // Initialize Shopify integration
  document.addEventListener('DOMContentLoaded', function() {
    optimize.initShopify();
  });
</script>

Automatic Tracking

Once initialized, Experiment Flow automatically tracks:

Add to Cart - Intercepted via AJAX (both fetch and jQuery)
Checkout Start - When user clicks checkout button
Page Context - Product type, collection, and page type

A/B Test Add-to-Cart Button

optimize.testAddToCart(experimentId, {
  'variant_a': {
    text: 'Add to Cart',
    style: { backgroundColor: '#000' }
  },
  'variant_b': {
    text: 'Buy Now - Free Shipping!',
    style: { backgroundColor: '#22c55e', fontWeight: 'bold' }
  }
});

A/B Test Product Page Elements

// Test with product context automatically included
optimize.testProduct(experimentId, {
  'variant_a': {
    '.product-title': { style: { fontSize: '28px' } }
  },
  'variant_b': {
    '.product-title': { style: { fontSize: '36px', color: '#1a1a1a' } },
    '.product-badge': { text: 'BEST SELLER', show: true }
  }
});

Test Price Display

optimize.testPrice(experimentId, {
  'variant_a': {
    format: '${price}'
  },
  'variant_b': {
    format: 'Only ${price} - Save 20%!',
    style: { color: '#dc2626', fontWeight: 'bold' }
  }
});

Track Purchases (Thank You Page)

Add this to your checkout thank you page (or use Shopify's Additional Scripts):

<script>
  // Shopify provides order data on thank you page
  {% if first_time_accessed %}
    optimize.trackPurchase({
      total_price: {{ checkout.total_price }},
      order_id: '{{ order.name }}'
    });
  {% endif %}
</script>

Custom Add-to-Cart Tracking

If you need manual control over tracking:

// Manual add to cart tracking
optimize.trackAddToCart({
  price: 2999,  // price in cents
  product_id: '12345'
});

// Get current cart data
optimize.getCart(function(cart) {
  console.log('Cart total:', cart.total_price / 100);
});

Listen for Events

// Listen for add to cart events
document.addEventListener('optimize:addtocart', function(e) {
  console.log('Product added:', e.detail);
});

REST API

Endpoints

POST /api/experiments - Create experiment
GET /api/experiments - List experiments
GET /api/stats/{id} - Get experiment stats
POST /api/decide - Get variant assignment
POST /api/convert - Track conversion
POST /api/track - Track custom event

Authentication

Include your API key in the X-API-Key header or as api_key query parameter.

Statistical Significance

We use a two-proportion z-test to determine statistical significance at 95% confidence. Results show:

Conversion rate per variant
P-value and z-score
Confidence level
Required sample size estimate
Winner declaration when significant