Preview and QA a Theme Experiment
How to preview and QA a theme experiment in Elevate A/B Testing before launch, and how to browse an unpublished theme without being redirected.
A theme experiment changes every page for every assigned visitor. QA it before you launch, not after.
Previewing without being redirected
Elevate sends visitors to their assigned theme. That applies to you too, which gets in the way when you are trying to look at an unpublished theme.
Add eabUserPreview=true to any theme preview URL to exclude yourself:
https://yourstore.myshopify.com/?preview_theme_id=1234567890&eabUserPreview=trueUse &eabUserPreview=true if the URL already has parameters.
This applies to your browser until you clear your cookies for the store. Use it when:
Previewing a theme in the Shopify theme editor while an experiment is running
Sharing a preview link with your team or a client, so nobody is redirected mid demo
Running automated browser tests
To go back to being treated as a normal visitor, clear your cookies for your store domain, or open the store in an incognito window.
Tell anyone doing theme work about this parameter. An unexpected redirect during a client review is hard to explain.
QA checklist
Run every check on every variation theme, before launching.
Page types
[ ] Homepage. Hero, featured collections, navigation, footer.
[ ] Collection pages. Grid, filters, sorting, pagination.
[ ] Product pages. Images, description, variant selectors, Add to Cart, reviews.
[ ] Search results.
[ ] Cart. Items display correctly, upsells work, checkout button works.
[ ] Checkout. Complete a test order end to end.
Apps
Apps do not carry across between themes. Confirm each one works on the variation.
[ ] Reviews widget appears on product pages
[ ] Chat widget loads
[ ] Upsell and cross sell popups appear
[ ] Subscription options work, if you sell subscriptions
[ ] Trust badges and payment icons display
Devices
[ ] Desktop at full width
[ ] Mobile
[ ] Tablet, if a meaningful share of your traffic uses one
Content parity
[ ] Menus and navigation match
[ ] Announcement bars and banners are present on both
[ ] Any section that exists on the control also exists on the variation, unless its absence is the thing you are testing
This is the check people skip, and the one that most often invalidates a theme test. A missing testimonial block turns a theme test into a testimonial test.
Tracking
[ ] Browse the variation theme and confirm the visit is recorded
[ ] Confirm the correct variation is assigned
[ ] Complete a test order and confirm it appears in the experiment data
If the variation theme records nothing, do not launch. Visitors assigned to it will be served the theme but will not be counted, which quietly biases your result.
Before you launch
[ ] Every page type checked on every variation
[ ] Apps confirmed working
[ ] Mobile checked, not just desktop
[ ] Content parity confirmed
[ ] Tracking confirmed on every variation
[ ] Every variation theme is unpublished
[ ] Goal metric is set, and it is Revenue per visitor unless you have a specific reason otherwise
[ ] Traffic split and audience targeting are what you intended
After you launch
Check the store yourself in an incognito window in the first hour, and watch the numbers for the first 24 hours. Theme tests affect everyone, so a missed problem is expensive.
If something is wrong, pause the experiment rather than publishing or unpublishing themes underneath it.
Last updated