Quickstart
Install AB Test, enable the storefront embed, start your trial, and launch a URL redirect test in a few minutes.
Before you start
You need a Shopify store with a published theme. AB Test runs as an embedded admin app, so there is no separate account or login to create — installation and authentication are handled by Shopify's managed install and session tokens.
A URL redirect test is the fastest way to see the app work end to end, and it needs no code. This guide uses it as the example. Once you are comfortable, the same steps apply to template, theme, and JavaScript API tests.
AB Test never creates or modifies theme files. For a URL redirect test you only need two page URLs that already exist on your store.
1. Install and open AB Test
- Install AB Test from the Shopify App Store.
- After the managed install completes, open the app from Apps in your Shopify admin.
- The app loads embedded in the admin. There is no username or password step — you are authenticated through your Shopify session.
2. Enable the storefront app embed
AB Test needs its storefront extension active so it can run experiments for real visitors. Enable it once:
- Go to Online Store > Themes.
- On your published theme, click Customize.
- Open App embeds in the theme editor.
- Toggle on the AB Test app embed, then Save.
Also keep the Web Pixel that AB Test registers enabled. Together these two pieces do the work of every test:
- Assignment — the app embed assigns each visitor to a variant and holds them there consistently.
- Variant serving — for URL, template, and theme tests, the embed applies the assigned variant on the storefront.
- Event capture — the Web Pixel records the events used to measure conversion, and the embed stamps an anonymous visitor id (
_ab_vid) as a cart attribute so revenue is attributed to the right variant, even through third-party checkouts. - Open the app's plan or billing screen.
- Select a plan and approve the Shopify subscription charge to begin the trial.
- In AB Test, create a new test and choose URL redirect.
- Give the test a clear name so you can find it later.
- Set the control URL (your current page).
- Set the variant URL (the alternate page you want to test).
- Keep the traffic split at 50/50.
- Click Launch.
- Open your storefront in a normal browser window and note which page you land on (control or variant).
- Open the same entry URL in an incognito or private window.
- Because the two windows are separate visitors, they should sometimes land on different variants.
- Reload each window a few times — a given visitor stays on the same variant, which confirms assignment is sticky.
- URL redirect tests — full options for the test type you just launched.
- Audience targeting — limit a test to specific visitors.
- Results — read Bayesian results and segment breakdowns.
If the app embed or the Web Pixel is disabled, visitors will not be assigned to variants and results will not record. Confirm both are on before launching.
3. Start your free trial
AB Test includes a 15-day free trial. A paid plan is required to launch a test, so choose a plan before or at launch time:
The Base plan covers any 2 test types with 2 running tests; Premium unlocks all 4 test types and 5 running tests. See Plans for full limits. There is no free tier.
4. Create a URL redirect test
A URL redirect test splits traffic between two page URLs — for example two versions of a landing or product page.
Leaving audience targeting off sends the whole eligible audience into the test. You can narrow by device, new vs returning, country, UTM, referrer, or cookie later — see Audience targeting.
5. Verify assignment works
Confirm the split is live before you rely on the data:
Seeing both variants across fresh sessions means assignment and serving are working.
6. View results
As visitors flow through the test, open the test's Results view. Results use Bayesian statistics and show a probability-to-be-best for each variant, along with per-variant conversion rate and revenue. When a variant wins, you can apply it in one click.
Revenue is attributed per variant using the anonymous _ab_vid, so orders are booked correctly even when checkout happens off the storefront pixel. Read more in Results and Revenue attribution.
Tests auto-end at their scheduled end date. You can also end a test manually and apply the winner whenever the numbers are decisive.