A/B testing lets you send shoppers to two or more versions of a storefront destination and measure which performs better. Instant handles variant assignment, tracking, and results inside the same project where you build pages.
A/B testing is available on the Pro plan and up. Each eligible plan includes a monthly pool of A/B test sessions — unique visitors enrolled in your experiments — that resets each month. Pro includes 10,000 sessions per month and Business includes 100,000. See Read A/B test results for how sessions are counted.
Open the A/B testing panel
In the project sidebar, select A/B testing.
Click New A/B Test.
You can also start a test from a layout's ⋯ menu in the Layouts & layers panel with New A/B test, which pre-fills that layout as the first variant.
If you are out of A/B test sessions for the month, the panel prompts you to upgrade instead of opening the form.
Fill in the test form
The form opens with these sections. When you are done, click Start test in the top bar.
General
Name — use something you will recognize in your test list later (for example, "PDP hero — image left vs right").
Localization
Market — the Shopify market the test runs in.
Language — the locale within that market, shown when the market has more than one.
Instant resolves each variant's URL against the selected market and language, and shows the resulting Market URL. Loading markets needs the Shopify read_markets permission. For detail, see A/B test goals and targeting.
Variants
Each test starts with two variants (Variant 1 and Variant 2). For each one:
Give the variant a name.
Choose the storefront destination it points to (a product, page, blog post, collection, or other Shopify content).
Use Add variant to add more, up to six in total (labelled A through F). Multi-variant tests (more than two variants) depend on your plan — if your plan doesn't include them, Add variant opens an upgrade prompt. Your plan's features are shown in Settings → Plans. When you have three or more variants, the first is treated as the Control.
Traffic split
Set the percentage of enrolled traffic each variant receives. Each value must be between 1 and 99. With two variants the split auto-balances to total 100%; with more variants, set each percentage yourself.
Goal
Pick one primary goal — Conversion rate (the default), Revenue per visitor (RPV), Average order value (AOV), or Click-through rate (CTR). Instant uses this goal to rank variants and decide a winner. See A/B test goals and targeting.
A/B Test link
The form shows the A/B test link for the experiment, with a Copy button. This is how shoppers enter the test — direct traffic to the link and Instant assigns each visitor a variant.
Start the test
Click Start test. The experiment goes live immediately and begins enrolling visitors who arrive through the A/B test link. Share that link wherever you drive traffic (ads, email, social) so Instant can assign variants and count sessions toward your goal.
While the test runs
Direct traffic to the A/B test link so visitors are enrolled and counted.
Watch your remaining A/B test sessions in the A/B testing panel — high-traffic tests use the monthly pool quickly.
Avoid editing a variant's destination mid-test; changing what a variant points to mixes data. If you need a different setup, duplicate the test or start a new one.
You can End test early at any time from the test's page, and choose which variant future traffic goes to. If you run out of sessions, enrollment pauses until the next month or a plan upgrade.
Next steps
Configure your goal, market, language, and split in detail: A/B test goals and targeting.
When results come in, read them and pick a winner: Read A/B test results.
If A/B testing is locked on your account, see Upgrade your plan.