Your Shopify preview looks right but the live store does not
If a Shopify theme preview differs from the live store, first compare theme, page and context. Publishing to see whether it fixes itself changes what buyers receive; it does not identify the cause.
Perfect on your screen. Apparently a different store on your colleague’s phone.
Before declaring war on the cache… establish whether you are looking at the same thing.
Name both versions
Record the working theme and the version receiving ordinary traffic. “The new one” and “the good one” work beautifully until somebody creates another copy and you have three good ones.
Shopify’s preview guide provides theme and status identification and country and language controls. Use those signals to record the comparison.
Then open the same kind of page in both versions. A polished homepage does not validate a product page; one product does not establish how every other reference is presented.
A theme copy is not a business copy
Shopify’s testing-environment guide explains that theme duplication supports design work without isolating every store setting. That distinction prevents an unpleasant surprise.
Hypothetical example: while preparing a new theme, you edit a product’s text in admin for a test. Do not treat that as private draft content. Establish where data lives and what uses it before changing it.
The theme duplication help describes a way to work on a theme. It is not a complete catalog backup or a separate store.
Make a change list with two columns: intended modification and the business area it belongs to. If you cannot establish whether a setting is theme-specific, clarify before experimenting in a selling store.
Compare conditions rather than memories
I propose this record for differences:
| Detail | What to record in both views |
|---|---|
| Theme | Name and identity of the reviewed version |
| Page | Specific route and product or collection |
| Context | Country, language and buyer type |
| Screen | Device and approximate width |
| Expected change | Intended appearance or behaviour |
| Outcome | Actual difference and the step where it occurs |
The goal is a reproducible difference, rather than similar-looking screenshots. A preview in another language or displaying another variant may legitimately differ.
For B2B, include session and company context where relevant. Do not attribute an unexplained content difference to design.
Review a buying journey, not just the changed block
When modifying a product page, review variant selection, quantity and entry into the cart. For navigation changes, check reaching products through the mobile menu.
This is a proposed matrix, not a store test already completed:
| Journey | Acceptance question |
|---|---|
| Entry from an internal page | Is the intended content reached? |
| Variant selection | Is the chosen option understandable? |
| Add to cart | Are product and quantity correct? |
| Mobile navigation | Can users operate the menu and return? |
| Integrated app block | Is it present and functional in the reviewed version? |
Every visual interaction does not require a real purchase. For payments, integrations or externally consequential changes, agree on an appropriate environment and procedure. Preview alone does not establish isolation.
When preview still does not match the theme
Generate a link using the correct theme’s preview action rather than trusting an old address. Official guidance distinguishes link types and access conditions; share the one appropriate for the reviewer.
If differences persist, retain both addresses and the comparison record. A customisation or integration may need review. “It is the cache” is a hypothesis to test, rather than a magic phrase for closing an issue.
Do not uninstall apps or change infrastructure based on suspicion. A code issue calls for technical review supported by evidence, rather than a forum patch describing another store.
What to prepare before publication
Define the intended version, reviewed journeys and who will verify the result afterwards. Keep a reference to the previous version and document changes outside the theme too.
Returning to the previous theme does not automatically reverse every business decision. Shared content or settings need their own change record.
The question on updating a customised theme addresses preserving modifications during updates. This guide tackles a different question: why the reviewed experience differs from the published one.
For help, see Shopify theme development. Bring a reproducible example and expected outcome. “This page does this and should do that” supports work. “It looks weird”… still needs a conversation.
Documentation checked in November 2026. Examples and matrix are hypothetical; no client-store publication or tests are claimed.
Frequently asked questions
- Does saving a draft theme mean changes are published?
- No. Identify the reviewed theme and the active theme before comparing. Saving a change and publishing the theme are different actions.
- Does duplicating a theme isolate the entire store?
- No. The copy supports theme work rather than isolating all business data and settings. Establish each change’s scope before treating it as a test environment.
- Should I publish to see whether the difference disappears?
- Not as a diagnostic experiment. Reproduce the difference and review the intended journey first. Publishing exposes the change; it does not explain why the views differed.