Create a Page Experiment
Step by step walkthrough for creating a page experiment in Elevate A/B Testing, including page type selection, scoping, and setting up variation templates.
This walkthrough covers all five steps of creating a page experiment, from naming it to launching.
Before you start: Elevate A/B Testing installed, and a theme that supports alternate templates. Any Online Store 2.0 theme does. You do not need to build the alternate template first. Elevate A/B Testing can duplicate your current one for you during setup.
From your dashboard, go to Experiments, select New Experiment, then select Page Experiment.
Step 1: General Info & Goal
Name the experiment and choose the metric that decides the winner.
Enter an Experiment name. This is the only required field. Name it for what you changed, such as "PDP reviews above the fold", so it is still readable in your results list three months from now.
Add a Description and a Hypothesis if you want them. Both are optional and both are worth writing. The hypothesis is what you expect to happen and why, recorded before you see the data.
Attach Assets if useful. Images and documents up to 10 MB, videos up to 100 MB. Use this for mockups, design files, or the brief the change came from.
Under What do you want to measure?, select your experiment goal.
The goal defines how the winning variation is decided.
Revenue per visitor
You want overall revenue impact. Recommended default for most page experiments.
Conversion rate
The question is purely whether more visitors buy.
Average order value
The change is intended to affect basket size, such as upsell placement on the cart page.
Add-to-cart rate
You are testing how well the page drives initial interest, and want a faster signal than purchase.
Checkout start rate
The change targets the middle of the funnel.
Profit per visitor
You want margin rather than revenue. Requires product cost data.
Select Browse all metrics for the full list, or Add Custom Metric to track something Elevate A/B Testing does not measure by default, such as a newsletter signup or a click on one specific button. See Custom Events.
Every other metric is still calculated and shown in your results. The goal metric only decides how the winner is chosen.
Choose one goal and leave it alone once the experiment is running. Changing the goal after the fact and picking whichever metric happens to look good is the most common way to talk yourself into a result that is not there.
Select Continue.
Step 2: Set Variations
This is the step unique to page experiments. You choose which page type to test, how wide the experiment reaches, and what each variation renders.
Choose the page type
Under Select test type, choose one:
Home Page
Your storefront homepage
Collection Page
Collection and category pages
Product Page
Product detail pages
Cart Page
The cart page
Other Pages
Custom Shopify pages such as landing pages, about, or FAQ
The page type you choose determines what appears next. Each experiment covers one page type. Multi-page testing, which will let a single experiment span more than one page, is coming soon.
Scope the experiment
Home Page and Cart Page have no scoping step. There is one of each, so the experiment runs there.
Product Page offers four options under What products do you want to track?
Show on all products
Every product page enters the experiment. Use for sitewide layout changes.
Show on specific products
Only the products you select. Use to test on a high traffic product without touching the catalogue.
Exclude specific products
Every product page except the ones you select. Use for large catalogues where a handful of products need to stay untouched.
Show on specific templates
Only products already using the templates you select. Use when a product type such as denim or hoodies has its own template and you want to test that group.
Collection Page offers two options under What collections you want to track?: Show on all collections, or Show on specific collections. Choosing specific opens Select Collections, where you search by collection name and select the ones to include, then Save.
Other Pages requires you to pick the pages explicitly. Select Add pages to open Select Shopify Pages, search by page title or handle, select one or more, then Close.
Set the number of variations
Under Add your variations, use the Number of Variations dropdown. You can run up to five variations including your control, so four alternatives at most.
Start with two. A two variation experiment reaches a conclusion far sooner than a four variation one, because traffic is split fewer ways.
Set up the control
The Control card is your current live template. It is set for you and needs no configuration.
Two things on this card are useful:
Edit in Shopify opens the template in the Shopify theme editor.
Preview opens the control page as a visitor sees it.
If the thumbnail on the control card looks wrong or out of date, open the ⋮ menu and select Refresh preview.
Set up each variation
Name the variation something you will recognise in your results, such as "Minimal layout" rather than "Variation 1". Then choose how the variation is built:
Duplicate
Copies your current template into a new one you can modify
You are building the alternative from scratch. This is the usual choice.
Select
Points the variation at an alternate template that already exists in your theme
You have already built the alternative, or you are testing two templates that both already exist.
Add Custom Component
Adds a pre-built component such as a trust badge or social proof element to a duplicated template
You want to add one specific element rather than redesign the template. Currently in beta.
Duplicate and Add Custom Component both end at Select Template, where you choose which template to copy. Your current live template is at the top of the list, and any templates from previous experiments appear below it.
Elevate A/B Testing names the copy with an eab- identifier, so duplicating templates/index.json produces something like templates/index.eab-XXXXXXXX.json, where the suffix is generated for that variation. Note the name. You will need it if you use Shopify Translate and Adapt.
After duplicating, open the new template in the Shopify theme editor and make your changes. Finish the variation before you launch. A half configured template loses on its own merits and tells you nothing about the design you meant to test.
Select Continue.
Step 3: Traffic Allocation
Set what share of eligible visitors enters the experiment, and how those visitors split between control and variations.
Traffic defaults to an even split. Uneven splits are valid but cost you time, because the result is limited by whichever side has less traffic.
See Traffic Allocation for the full explanation, including how this differs from traffic isolation.
Step 4: Audiences
Optionally narrow who enters the experiment, by device, location, traffic source, visitor type, or UTM parameters.
Leave this empty unless the hypothesis is about a specific audience. Every filter you add shrinks your sample and lengthens the experiment.
See Audience Targeting for the full list of conditions.
Step 5: Summary
Review everything before you commit. Check in particular that each variation points at the template you intended, and that the scope matches what you meant to test.
Select Save as Draft at any point to come back later. The experiment sits in Draft and nothing is live.
After you create it
Creating the experiment does not start it. Before you launch:
Preview each variation and work through the QA checklist. See Preview and QA a Page Experiment.
If your store uses Shopify Translate and Adapt, translate the variation templates. A new template starts with no translations, so visitors in other languages see untranslated content until you add them.
Launch when QA passes.
Next steps
Preview and QA a Page Experiment
Reading Results and Ending a Page Experiment
Troubleshooting Page Experiments
Last updated