Creating an A/B Test

Go to Alpha Insights → A/B Test → Add New. Save a draft first, preview the variants, then start the test when you are ready. Starting a test bypasses page cache on targeted URLs until you pause or complete it.

Basic fields

Field	Purpose
Name	Admin label shown on the A/B Test list and report.
Slug	Stable key used in preview links and `wpdai_in_experiment()`. Autogenerated from the name; edit it before you start if you will call it from code.
Hypothesis	Optional note for your team (what you expect to change and why).
Primary goal	What the test is optimising for. Free tests always use Purchases. Pro can also use purchase value, revenue per exposure, average order value, add to cart, a specific page view, or a custom event type.
Status	Draft, running, paused, or completed. You can also Start / Pause / Complete from the list.
Traffic in test	Percent of eligible visitors assigned to an arm. The rest are holdout (Pro). Free is always 100%.
Schedule	Optional start and end datetimes (Pro). Leave empty to control the test manually.

Primary goals

Purchases — a WooCommerce transaction attributed to an exposed visitor. If analytics has not written a session cookie yet, Alpha Insights still attributes the order using the visitor ID stored on the order. This is the only primary goal on Free.
Purchase value (Pro) — same purchase attribution as Purchases, but the list compares revenue per exposed visitor (rate, lift, and the leading badge). Use this when a treatment might sell fewer orders at a higher value, or the reverse.
Revenue per exposure (Pro) — same purchase attribution as Purchases, but the list compares revenue ÷ exposures. Prefer this over total revenue when one arm has more traffic.
Average order value (Pro) — same purchase attribution as Purchases, but the list compares average order value. Use this when the hypothesis is about basket size, not conversion rate.
Add to cart (Pro) — session recorded an add_to_cart event after exposure.
Page view (Pro) — session viewed the path you enter (for example /thank-you/).
Custom event (Pro) — session recorded the event name you enter (for example form_submit). Use the same name as custom event tracking.

Purchase value and average order value still count a conversion when an attributed purchase happens. They change how lift is scored on the A/B Test list, not whether an order counts. Revenue and AOV columns on the date-filtered report stay Pro-only either way.

Recommended workflow

Create the A/B test as a draft and add Control plus Treatment (or more arms in Pro).
Narrow page targeting to the URLs you are changing. Empty URL rules mean the entire site.
Add CSS, JS, or PHP on the treatment — or leave snippets empty and branch in your theme with the developer helpers.
Save, then use the preview links while logged in as an Alpha Insights user. Preview does not record experiment events.
Start the test. Watch the list for exposures; open A/B Test → Report for charts and the full table.
Pause if something looks wrong. Complete when you have a decision. Do not delete a running test if you still need the history.

The A/B Test list

Each test is its own table. The header shows name, slug, status, traffic %, and CSS / JS / PHP badges when a variant has snippet code. Under that, each arm shows all-time exposures, add to carts, initiate checkouts, purchases, revenue, revenue per exposure, AOV, the selected goal’s rate, and lift versus control. Use Target conversion to preview lift against Purchases, Add to carts, Initiate checkouts, Purchase value, Revenue per exposure, or Average order value without changing the saved primary goal. Those numbers come from the same A/B testing data warehouse as the report (the report itself is date-filtered, usually last 30 days). Revenue, revenue per exposure, AOV, and the value goals are Pro.

Creating an A/B Test

Basic fields

Primary goals

Recommended workflow

The A/B Test list

Related