Your First Experiment

From the Experiments view, click Create Experiment in the top right. A wizard opens with four tabs: General, Targeting, Variants, and Results.

In General, give the experiment a clear name like “Product page — orange CTA vs blue control”. Vague names (“Button test”) become useless once you have ten of them.

Add a hypothesis in the description field. The format that works best is:

“We believe [change] will [increase/decrease] [metric] because [reason].”

Example: “We believe an orange CTA button will increase add-to-cart rate because it contrasts more strongly with our white theme than the current blue.”

Add a URL targeting rule. The simplest first-experiment setup is:

• URL part: Path
• Operator: Equals
• Value: /products/your-product-slug

This restricts the experiment to one specific product page. You can target broader patterns (like /products/* for all product pages) later — see URL Targeting for the full set of operators.

Paste the full URL of the page you’re targeting (e.g., https://yourstore.com/products/your-product-slug) into the Preview URL field. The app uses this to:

1. Show a banner confirming whether your targeting rules would activate on that page.
2. Capture screenshots of each variant before you launch.

If the banner says “Preview URL doesn’t match targeting rules,” fix the targeting before continuing.

In the Variants tab, click Add Variant. A new variant appears below the control. Open it to edit.

Each variant has three independent fields you can fill in: CSS, JavaScript, and Redirect URL. Use one for a simple test or combine them for a more complex change. For your first experiment, use the CSS field — see Variant Types Overview for when to reach for the others.

For an orange-CTA test, your variant CSS might look like:

.btn-primary,
form[action="/cart/add"] button[type="submit"] {
  background-color: #f5620a !important;
  border-color: #f5620a !important;
  color: #ffffff !important;
}

Use your browser’s DevTools to find the right selector — open the page, right-click the button you want to change, and copy a stable class or ID. See Selector Cookbook for patterns.

By default each variant gets an equal share of traffic — 50% / 50% for two variants. Leave it at the default for your first test. You can use uneven splits later for risk-mitigation rollouts (see Traffic Allocation).

Back on General (or in the Metrics section, depending on platform), select the primary metric the experiment will be judged on.

• Shopify users: built-in goals like Add to Cart, Checkout Started, Purchase, and Revenue per Session are pre-wired and require no setup.
• HTML users: pick from the conversion goals you defined in Settings → Conversion goals, or define a new one. See Conversion Goals.

Whatever you pick is the metric the “Declare winner” button will judge.

When you click Start Experiment, Split Test Pro shows a Pre-Launch Checklist before going live. The checklist surfaces a reminder to manually verify your variant renders correctly on the target page (since automated visual diffing isn’t part of the launch flow), and an optional Get AI Review button that runs a Claude-powered check of your experiment setup.

The checklist doesn’t block you — even with the warning, you can proceed via Start Test Anyway. The decision to go live is always yours. See Pre-Launch QA for what to look for before you click.