Found
Ecommerce Migration Without Losing Product URLs, Feeds or Customer Journeys
Check product links, feeds, variants and checkout before moving your ecommerce store to a new platform.

Milan Gohil
·
8 min read
OVERVIEW
Here is a short version of what I write.
An ecommerce migration needs three connected checks: discoverability, product data and the buying journey. A working storefront can still hide broken product links, mismatched feeds or checkout failures.
Map product and category URLs and test redirects.
Verify product identifiers, variants, prices, availability and images.
Check feeds and structured data against the live store.
Test selection, cart, payment, confirmation and purchase tracking.
Moving an ecommerce store is not the same as moving a normal business website. You are not only moving pages. You are moving three connected systems:

A migration can succeed technically while still breaking one of these. The new Shopify, WooCommerce, Magento or custom store may load perfectly while:
old product URLs return 404s
Shopping ads point to outdated pages
variants become separate products incorrectly
product prices stop matching Google Merchant Center
users lose their cart during checkout
discount codes stop working
analytics no longer records purchases
So before choosing a launch date, test all three tracks separately.
Start by deciding what actually needs to change
Changing ecommerce platform does not automatically mean every product URL needs changing. Imagine the current store has:
/products/black-leather-office-chair
and the new platform can support the same public URL. There may be no business reason to turn it into:
/collections/office/products/chair-48192
just because the new system prefers a different default structure. Google recommends long-term, persistent URLs for ecommerce sites and consistent use of those URLs across internal links, canonicals and sitemaps. So separate:
PLATFORM CHANGE
from
URL CHANGE
They are not automatically the same decision.
Build one ecommerce migration control sheet
Do this before development is finished.

Current | New | Preserve? | Test before launch |
|---|---|---|---|
Product URL | New product URL | Yes/change intentionally | Redirect + page |
Category URL | New category | Yes/change intentionally | Products discoverable |
SKU | SKU | Usually preserve | Product mapping |
Merchant ID | Merchant ID | Preserve for same product | Feed |
Price | Price | Current value | Site/feed/schema match |
Availability | Availability | Current value | Site/feed/schema match |
Variant structure | Variant structure | Preserve relationship | Selection + feed |
Product image | Image | Move/replace | Page + feed |
Add to cart | New cart | Function | Real product test |
Discount | New discount system | Recreate | Apply/remove |
Shipping | New shipping rules | Recreate | Location tests |
Payment | New payment flow | Reconnect | Test transaction |
Analytics | New tracking | Reconfigure | Ecommerce events |
This becomes much more useful than:
“Product migration complete.”
Protect important product and category URLs first
Export the existing store. You want to know which URLs exist for:
products
categories/collections
brands
important filtered landing pages
buying guides
high-traffic content
Then add Search Console, analytics and backlink data where available. A product that generates:
organic traffic
revenue
external links
deserves more attention than a discontinued item nobody visits. Google also recommends ensuring products are discoverable through actual site navigation: category pages should link to subcategories and products rather than relying only on an internal search box. So migration QA should test the path:
HOME → CATEGORY → SUBCATEGORY → PRODUCT
not only whether the product URL exists.
If URLs change, map them product by product
Suppose:
Old: /product/mens-running-shoe-x1
New: /products/running-shoe-x1
Create an explicit mapping. The same applies to categories:
Old: /running/mens
New: /collections/mens-running-shoes
Use the relevant new destination rather than sending hundreds of old products and categories to the homepage. This follows the same migration principle Google recommends for general site moves: map old URLs to their appropriate new destinations and use permanent redirects for permanent moves.
Product IDs are different from product URLs
This distinction is easy to miss. Your customer may see:
/products/red-running-shoe
while Merchant Center knows the product internally as:
RS-482-RED-09
Google recommends keeping the Merchant Center id stable for the same product. Changing it causes Google to treat it as a different product and disconnects the previous Merchant Center/Google Ads performance history from the new ID. So during migration, do not casually regenerate all product IDs simply because the new ecommerce platform generates new database IDs. For the same product:
OLD MERCHANT ID RS-482-RED-09 → NEW MERCHANT ID RS-482-RED-09
where your system allows it.
Check variants carefully
Variants create another migration problem. One product might have:
T-SHIRT
Black / S
Black / M
Black / L
White / S
White / M
White / L
The old platform may model this one way and the new platform another way. Google's Merchant Center specification requires different variants to have their own unique product IDs while related variants can share an item_group_id. So before launch compare:
Attribute | Old | New |
|---|---|---|
Parent/group | — | — |
SKU | — | — |
Merchant ID | — | — |
Size | — | — |
Colour | — | — |
GTIN | — | — |
Price | — | — |
Availability | — | — |
Product URL | — | — |
Image | — | — |
Do this on representative products before migrating thousands.
Merchant Center must move with the website
An ecommerce migration does not end when the website works. Google Merchant Center product data contains landing-page URLs. Google requires the submitted link to lead to a valid product landing page on the verified domain and recommends using stable URLs rather than URLs that continually change. That means:
OLD PRODUCT URL CHANGES → UPDATE WEBSITE + UPDATE FEED / DATA SOURCE + CHECK MERCHANT CENTER
Do not leave Shopping traffic pointing toward an old redirect chain or dead product page.
Price and availability need to agree in three places
For an ecommerce product, you can effectively have three representations of the same offer:
Visible Product Page: £79.00 — In stock
Product Structured Data: £79.00 — InStock
Merchant Center Data: £79.00 — in_stock
Those should agree. Google Merchant Center checks product data against landing pages and structured data. Price or availability mismatches can cause product issues or disapprovals. Google can use automatic item updates to correct some temporary price or availability differences using data from the website, but Google explicitly says this is not a replacement for regularly providing accurate product data. After migration, test:
PRODUCT PAGE
=
SCHEMA
=
FEED
for representative products.
Make sure Product structured data survives the rebuild
The previous theme may already contain Product/Offer structured data. Changing themes or ecommerce platforms can remove or change it without anything looking visually broken. Google says Product markup can make pages eligible for merchant listing experiences that include information such as price, availability, shipping and returns. Test several product types after the migration:
Normal product
Sale product
Out-of-stock product
Variant product
Product with reviews
Do not only test your easiest product.
Product images are part of the migration too
Product image URLs often change when moving platforms. Check:
OLD IMAGE → NEW IMAGE Does it load? Is it crawlable? Is the correct image in the feed? Do variants show the correct image?
Merchant Center requires submitted product-image URLs to be accessible to Google's crawlers. If you are rebuilding the product-image library during 2026, it is also worth knowing that Google's 2026 Merchant Center specification says enforcement of a new minimum 500×500-pixel product-image requirement begins on January 31, 2027. That makes a migration a good time to identify product imagery that will soon need replacement anyway. Google: Merchant Center specification update 2026.
Now test the part SEO tools cannot protect: buying
This is where ecommerce migration becomes different from normal website migration. A crawler can tell you:
Product URL returns 200.
It cannot tell you whether the customer can successfully buy it. Test complete journeys.

The journey should be tested with real store scenarios rather than simply confirming that each screen exists. Baymard's large-scale checkout research continues to find that checkout usability problems themselves can cause abandonment, so introducing unnecessary friction during a platform migration can affect a journey that previously worked.
Test more than one order
One successful order does not prove the store is ready. Use a small migration test matrix.
Scenario | Test |
|---|---|
Standard item | Normal purchase |
Product variant | Size/colour selection |
Sale item | Original + sale price |
Discount code | Apply/remove |
Out of stock | Correct behaviour |
Multiple quantity | Cart update |
Different region | Shipping/tax |
Guest customer | Checkout |
Existing customer | Login/account |
Mobile | Full purchase |
Refund/cancellation | Back-office workflow |
Add any scenarios specific to the business. For example:
subscriptions
pre-orders
gift cards
B2B pricing
local pickup
international shipping
product personalization
A store should be tested around what it actually sells.
Customer accounts need their own migration decision
If the current platform contains existing customer accounts, decide what happens to:
Data | Migration decision |
|---|---|
Customer profiles | — |
Addresses | — |
Order history | — |
Passwords | — |
Saved cards/payment tokens | — |
Wishlists | — |
Store credit | — |
Loyalty points | — |
Gift-card balances | — |
Subscriptions | — |
Some of these may not be transferable between platforms in their existing form. That is not necessarily a reason to avoid migration. It means the business needs a plan. For example:
Existing customers need to reset their passwords after launch.
That is manageable when communicated intentionally. It becomes a customer-support problem when nobody knew it would happen.
Do not forget what happens after payment
The ecommerce journey does not finish at the payment-success screen. Test that a successful order reaches:
Customer: Order confirmation
Store: Order management system
Warehouse: Fulfilment process
Payment Provider: Correct transaction
Email System: Transactional email
Crm / Marketing: Correct customer/order event
Analytics: Purchase recorded
A migration can produce a successful frontend checkout while silently failing further downstream.
Rebuild ecommerce analytics intentionally
Changing ecommerce platforms commonly changes:
data layer
checkout URLs
event triggers
product IDs
cart behaviour
So analytics should be tested again rather than assumed to survive. Google Analytics' recommended ecommerce implementation includes events for stages such as product selection, add_to_cart, begin_checkout and purchase. Your exact setup may differ, but the business should still be able to understand:
PRODUCT VIEW → ADD TO CART → CHECKOUT → PURCHASE
before and after migration. Otherwise you may launch successfully and then lose your ability to compare performance.
Compare performance before the move
Save a baseline before launch.

Measure | Before | After |
|---|---|---|
Organic product traffic | — | — |
Organic category traffic | — | — |
Merchant Center approved products | — | — |
Feed errors | — | — |
Add-to-cart rate | — | — |
Checkout starts | — | — |
Purchases | — | — |
Revenue | — | — |
Checkout errors | — | — |
Mobile conversion | — | — |
You do not need to expect identical numbers immediately. You need something to compare against. Without the baseline:
“Sales feel lower.”
With the baseline:
“Product traffic is stable, but checkout completion dropped after migration.”
Now you know where to investigate.
Before approving launch, review these three tracks

Search | Product data | Buying journey |
|---|---|---|
Important URLs mapped | Stable product IDs | Product selection tested |
Redirects tested | Variants correct | Cart tested |
Categories crawlable | Feed URLs updated | Discounts tested |
Internal links updated | Price matches | Shipping tested |
Canonicals correct | Availability matches | Tax tested |
Sitemap updated | Images working | Payment tested |
Indexing rules checked | Product schema valid | Confirmation tested |
Search Console ready | Merchant Center checked | Analytics tested |
If one column is still substantially unfinished, the store is not ready simply because:
“Development is complete.”
The goal is not to preserve the old store forever
Moving platform can be the right opportunity to improve:
category structure
product presentation
filters
checkout
content
technology
performance
operations
But separate intentional improvement from accidental loss. Keep the useful things stable where there is no reason to change them. Change the things that genuinely need changing. Then verify all three systems:
CAN PEOPLE STILL FIND THE PRODUCT?
DO GOOGLE AND OTHER SYSTEMS STILL UNDERSTAND THE PRODUCT?
CAN THE CUSTOMER STILL BUY THE PRODUCT?
If all three answers are yes, you have migrated much more than a website. You have migrated the store.
Sources and further reading
All Blogs
All Blogs & Articles
© 2026 Webshunya Studio®. All rights reserved.






