Shopify review migration QA checklist before a theme launch

Shopify product reviews are not just decorative social proof. They affect product-page trust, merchandising decisions, support questions, conversion rate, and in some setups the structured data that search engines read from the page.

A theme launch, review app change, product-handle cleanup, or catalog migration can quietly disconnect ratings from the right products. The storefront may still look polished while review counts are wrong, old moderated reviews reappear, rich-result markup disappears, or the review widget slows the product page. This checklist isolates those risks before launch.

Start with a source-of-truth inventory before touching the theme. List where reviews live today, which app owns them, how they are exported, which product identifier they use, and where each field should land after launch.

Do not only count total reviews. Record ratings, review body, reviewer name, email hash or customer reference, product ID, product handle, SKU, variant reference, photos, videos, questions and answers, merchant replies, moderation status, language, and original date.

• Export a current review dataset before theme work begins.
• Map each review app field to the new app, theme widget, metafield, or API destination.
• Identify products that changed handles, SKUs, variants, or canonical URLs.
• Separate published, hidden, flagged, imported, incentivized, and pending reviews.
• Keep a product-level review count baseline for launch comparison.

Most review migration failures are matching failures. A review can attach to the wrong product, duplicate across variants, disappear because the old handle changed, or publish again because moderation state was not migrated.

Use stable Shopify product IDs when possible. If the migration depends on handles or SKUs, test products that were renamed, merged, split, archived, or moved into new collections. Those edge cases usually reveal the real risk.

• Match by Shopify product ID first; use handles or SKUs only with a clear fallback rule.
• Check products with duplicate SKUs, retired variants, bundle products, and imported marketplace listings.
• Confirm hidden, rejected, pending, and flagged reviews keep the same moderation state.
• Preserve original review dates and merchant replies unless there is a legal or support reason to remove them.
• Decide how reviews should behave when a product is unpublished, redirected, or replaced.

Do not QA only one best-selling product with five clean reviews. The theme needs predictable states for products with no reviews, one review, hundreds of reviews, photo reviews, long comments, short comments, low ratings, translated content, and unavailable variants.

Review widgets often break layout through small details: star rows wrap on mobile, filters overlap sticky add-to-cart bars, app CSS overrides the theme, or lazy-loaded reviews shift the page after the user starts reading.

• Test zero-review, low-review, high-review, and photo-review products.
• Check mobile, tablet, desktop, dark mode if supported, and common browser widths.
• Verify sorting, filtering, pagination, helpful votes, media galleries, and review submission states.
• Confirm review snippets do not overlap price, variant selectors, subscription widgets, or sticky CTAs.
• Test translated storefronts when reviews, rating labels, or app UI appear in multiple languages.

Review migrations can change structured data even when the visible design looks right. If aggregate ratings or review schema were previously emitted by the theme, app, or custom code, confirm the new implementation outputs valid, non-duplicated markup.

Keep the markup honest. Search engines do not need review schema on every page, and fake or mismatched ratings create risk. The goal is to make the product page accurate, crawlable, and internally consistent.

• Validate Product, AggregateRating, and Review schema on priority product pages.
• Make sure rating count, average rating, and visible reviews agree with the structured data.
• Check that review tabs, lazy loading, app iframes, or hydration do not hide essential content from crawlers.
• Keep duplicate app widgets from emitting duplicate or conflicting schema.
• Confirm noindex, canonical, sitemap, redirect, and hreflang rules still match the product URL plan.

A review app can add scripts, CSS, fonts, media galleries, analytics events, and third-party requests to every product page. Measure the performance cost after the final widget is installed, not from a blank staging theme.

Also confirm consent and privacy behavior. If the review provider collects email, images, customer identity, or analytics events, the theme should not bypass consent banners or publish data the team meant to keep private.

• Compare product-page weight, request count, layout shift, and interaction delay before and after the widget loads.
• Check whether app scripts load only where the review experience appears.
• Confirm review submission, image upload, and email capture flows respect consent requirements.
• Test app outages or blocked third-party scripts so the PDP still remains usable.
• Track review interactions only once in analytics and tag manager.

Before the theme is published, choose a review migration sample that represents the whole catalog. Include best sellers, products with no reviews, products with many photos, old products with redirects, translated products, and any product family where trust signals are commercially important.

Save before-and-after screenshots and exports. A rollback plan is much easier when the team knows which app settings, theme sections, snippets, product IDs, and review exports need to be restored.

• Compare review count and average rating for at least 20 priority products.
• Check one product from each major collection, template, region, and language.
• Capture screenshots for PDP top summary, review list, review form, schema test, and mobile layout.
• Keep the final export, import log, app settings, theme version, and owner list in the launch folder.
• Define the rollback trigger: wrong ratings, missing reviews, schema errors, broken submission, or major performance regression.