Test types

Test whole pages against each other.

Send part of your traffic to a different URL, such as a new landing page, and compare orders and revenue.

When a whole page is the change

Use a split URL test when the new version is built as its own page, rather than edits to an existing one.

  • A redesigned page

    A product or landing page rebuilt from scratch, live at its own URL.

  • A different landing page

    Send campaign traffic to a new page and see whether it sells more than the current one.

  • A rebuilt collection

    A new collection layout or filter set published at a separate address.

New split URL testDemo data

How it works

  1. 01

    Set the URLs

    The control is your current page. Each variant is a URL; visitors keep their query parameters, so campaign tags survive.

  2. 02

    Choose goals and visitors

    The same goals, targeting and traffic controls as any A/B test.

  3. 03

    Assign and redirect

    Each visitor is assigned once and keeps their version. The visit is recorded before the redirect, and a loop guard stops repeat redirects.

  4. 04

    Compare

    Orders, revenue and gross profit per visitor, with the same statistics as every test.

A fair split, checked every day

Redirect tests are where traffic splits break most often, so each day's split is checked against the plan. A mismatch is flagged before anyone acts on the result.

  • Revenue comes from paid orders, with refunds and cancellations taken off.
  • Run it with auto-allocation to send more traffic to the better page while the test runs.
  • The full method: how results are judged.
Traffic and sample ratioDemo data

Good to know

  • Visitors in a variant load two pages, the original and then the redirect: a small speed cost on that version only, as with every redirect-based tool.
  • Both pages need the Benchmyrk script so goals are recorded on either one. On Shopify the app embed covers every page of the theme.
  • Visitors who haven't allowed analytics stay on the original page and aren't counted.

Questions

Does a split URL test hurt SEO?

Crawlers that run JavaScript can be redirected like shoppers, but known bots are never counted in results. If the variant page shouldn't rank on its own, give it a canonical link to the original page.

Are UTM tags kept after the redirect?

Yes. Visitors are redirected with their query parameters, so campaign tags and other parameters reach the variant page.

Can a visitor bounce between versions?

No. Each visitor is assigned once and keeps their version, and a loop guard stops a page from redirecting the same visitor again and again.

Testing a new page design?

We'll go through how the split, the redirect and the results work on your store. Or install the Shopify app and set one up today.