A/B Testing

A/B Testing lets you test different text variations on your site to discover which performs best. No code changes required — just target an element with a CSS selector and define your variants.

You can run up to 5 tests simultaneously. Each test can have up to 10 variants.

How It Works

Create a test and target a text element using a CSS selector
Define your control (original) and variant text options
Set traffic distribution between variants
Start the test and collect data
Declare a winner when you have statistical significance

Configuration

Test Setup

Option	Description	Example
Test Name	Internal name for the test	`Homepage CTA Test`
Target Selector	CSS selector for the element to test	`#hero-headline`
Status	Current test state	Draft, Running, Paused

CSS Selector Examples

Selector	Targets
`#hero-headline`	Element with id `hero-headline`
`.cta-button`	Elements with class `cta-button`
`[data-testid="signup"]`	Element with data attribute
`h1.main-title`	h1 with class `main-title`

Use a unique selector that matches only one element. If multiple elements match, only the first will be tested.

Variant Configuration

Each test needs at least 2 variants: a control and one or more variations.

Field	Description	Example
Name	Label for the variant	`Control`, `Variant A`
Content	Text that replaces the original	`Start Free Trial`
Weight	Percentage of traffic (0-100%)	`50`

Traffic Distribution

Weights must total 100%
Use "Balance Weights" to distribute evenly
Control is always the first variant

Example distribution:

Variant	Weight	Visitors
Control	50%	500
Variant A	25%	250
Variant B	25%	250

Goal Configuration

Define what counts as a conversion.

Goal Type	Description	Example
Button Click	Visitor clicks a button	`#signup-btn`
Element Click	Visitor clicks any element	`.pricing-card`
Form Submit	Visitor submits a form	`#contact-form`
Page Visit	Visitor reaches a specific page	`/thank-you`

Goal Settings

Option	Description	Example
Target Selector	CSS selector for click/form goals	`#checkout-btn`
Target URL	URL pattern for page visit goals	`/success/*`

Page visit goals support wildcards. /checkout/* matches any checkout subpage.

Results & Analysis

Once a test is running, you'll see:

Metric	Description
Impressions	How many times the variant was shown
Conversions	How many times the goal was achieved
Conversion Rate	Conversions / Impressions as percentage
Confidence	Statistical significance of results

Test Lifecycle

Status	Description
`draft`	Test is being configured, not live
`running`	Test is active and collecting data
`paused`	Test is temporarily stopped
`completed`	Winner declared, test finished

Wait for at least 100 conversions per variant before declaring a winner for reliable results.

Best Practices

Test one thing at a time — Isolate variables for clear insights
Run tests long enough — Wait for statistical significance (95%+)
Use clear hypotheses — Know what you're testing and why
Start with high-impact elements — Headlines, CTAs, value props
Document your learnings — Build a knowledge base of what works

Example Tests

Headline Test

Control: "The fastest way to build"
Variant A: "Build 10x faster"
Variant B: "Ship in days, not months"

Goal: Button click on #get-started

CTA Button Test

Control: "Sign Up"
Variant A: "Get Started Free"
Variant B: "Start Your Free Trial"

Goal: Form submit on #signup-form

Pricing Page Test

Control: "Start Free Trial"
Variant A: "Try Free for 14 Days"
Variant B: "No Credit Card Required"

Goal: Page visit to /checkout/*

On this page