How-to guides

Run an A/B test on a deal

Test two to four versions of a deal against each other: set up the variants, choose the split, read the winner rule, and keep the version you choose.

Applies to: TreStack Bundles, every plan
Updated: 5 November 2026

At a glance

Variants: Two to four, A to D, one open test per deal
Winner shown after: 200 visitors and 10 orders per variant, and 72 hours
Price charged: The price of the variant the shopper saw

An A/B test shows each browser one version of a deal and compares the versions on conversion rate. You decide which version stays.

Set it up

In the deal list, open the deal's Actions menu and choose Setup A/B test, or use Setup A/B test above the preview in the editor. Variant A is your deal and variant B starts as a copy.
Use the variant buttons above the preview to switch to B, and change one thing, such as one bar's discount or the title. Add variant adds C and D.
Edit shared settings, such as products, markets and active dates, on A only. They apply to every variant.
To change the split, open Settings, "A/B test settings". Even allocation splits traffic equally. Custom allocation sends 5 to 95 percent to the variants other than A. Choose Save settings.
Choose Publish, or Save on a deal that is already live. The test starts then.

For a test with one change picked from suggestions, see Smart A/B tests.

Read the result

The deal shows an A/B badge in the deal list. Expand it for a row per variant, with how long the test has run and its numbers.

A "Significant winner" badge shows only when every variant has at least 200 visitors and 10 orders, the test has run 72 hours, and the difference holds up at 95 percent (stricter with three or four variants). Until then the numbers are a running count. The test never stops on its own.

To end it, choose Set winner on a variant's row, which sends all traffic to it, or Stop A/B test in the deal's menu and keep any variant, significant or not.

Good to know

Each variant charges its own price. Two shoppers can see and pay different prices for the same deal while the test runs.
A browser keeps its variant. Another device, a private window or cleared site data can land on another one. Changing the split also moves some visitors.
Only shoppers who allow analytics count as visitors. A small difference may never become significant.
Ended tests stay in the A/B test archive, with their numbers.

Every rule, including how traffic is split, is in A/B testing a deal.

Show a deal in one market, or price it differently there

Limit a deal to one Shopify market, leave markets out, or run a second deal with its own prices for one market, and how prices follow the shopper's currency.

Export deal analytics as CSV

Download a deal's daily numbers as a CSV file for a spreadsheet: picking the deal and the dates first, what each column holds, the export limit, and why the totals differ from Shopify reports.