E-commerce website redesign: protecting conversions and search visibility

5 min readAndrey Antar
E-commerce website redesign: protecting conversions and search visibility

An old store has an advantage you can't see in a screenshot: it already works. People arrive through familiar links. Returning customers know where to find a product. Staff know how to update inventory. All of that belongs in an e-commerce website redesign, even if the meeting mostly covers the new homepage.

I'd start with the migration plan. What stays, what changes, how do we validate a purchase, and what would trigger a rollback? You can't promise there won't be a drop. Preparing for one is more useful.

The new version needs a baseline

Save the funnel before making changes: visits, product views, add-to-cart events, checkout starts, confirmed orders. For payments, use the actual transaction status rather than just a button click.

Look separately at devices, acquisition sources, and returning customers. Record promotions, ad changes, and stock availability. If a popular item sells out during the redesign, total sales won't tell you how well the new product page performed.

Verify that events mean the same thing in the new version. If add-to-cart was previously counted after a server response and is now counted on click, the metric could improve without a single additional purchase.

I cover the investigation itself in website UX audits. Here, the important thing is to make sure the baseline survives the migration and remains useful for comparison.

Where a good-looking filter runs out of data

Imagine a kitchenware store. This is a hypothetical example, not a client story. The new catalog has convenient filters for capacity and material. But the old database records capacity in milliliters for some products, free text for others, and nothing at all for the rest.

The interface faithfully displays the filtered results. A shopper reasonably concludes there aren't many suitable items. Some of the catalog simply never made it into the results.

To design an online store that works with its catalog, inspect the data before approving the filters. Which values exist, who maintains them, and can they be standardized? What will a shopper see when a value is missing?

The same applies to sorting, search, and variants. If a mug is sold out in one color, its product page shouldn't silently borrow the availability of another variant. A polished layout can't resolve a mismatch between what someone selected and what the store can sell.

A URL is part of the product

Our hypothetical mug may have lived at the same URL for years. Search results, bookmarks, and old articles point to it. The new platform makes a different path easier for the developer. To the project, that's a small string change. To an old link, it's the end of the road.

Prepare a URL map: where does each old address go? If a URL changes, a permanent redirect should lead to a relevant new page. Sending every removed product to the homepage is only convenient for whoever fills out the spreadsheet.

Google Search Central recommends mapping URLs, checking redirects, and monitoring the move. It also explicitly warns that rankings can fluctuate while Google recrawls the site. I'd ask anyone promising “no SEO drop” to explain that promise.

Check internal links, canonical URLs, the sitemap, images, and any indexing restrictions left over from staging. Keeping URLs isn't complete insurance, either: page content and site structure can change, too.

If you already have a new storefront in progress, send me the old and new versions. I'll review which working connections might get lost between them

One launch shouldn't have to test everything

A new platform, catalog, URL structure, and checkout can all launch on the same day. Afterward, it may be difficult to tell which change caused a problem.

Separate the transition where practical. Validate connected parts and keep working behavior where there's no reason to change it yet. Google also recommends avoiding unnecessary simultaneous major changes.

“Always change checkout last” isn't a useful universal rule. If checkout is the problem, investigate it. What matters is giving checkout its own validation rather than assigning it a place in the queue out of habit.

Agree on stop and rollback conditions in advance. What is unacceptable: missing orders, incorrect totals, unavailable delivery options? Who decides, who switches versions, and what happens to data created in the new system? Have that conversation before launch night.

Payment doesn't end at the thank-you screen

Checkout brings together the selected variant, quantity, discount, shipping, and total. Every value needs to reach the order without an unexpected substitution.

Test declined payments, retries, inventory changes, and returns from an external payment page. Do it with the integrations, not just a prototype. Payment confirmation and protection against duplicate processing belong to the back end and need specialist engineering.

Ideally, the team can follow a test purchase from selection through to the order appearing in its operational system. “The form submitted” is too early for a store to relax.

Store operations are moving, too

In the kitchenware example, someone has to fix the capacity values and add the missing materials. The catalog will keep changing after launch. If the new admin interface makes that work difficult, the data will soon fall behind the polished storefront.

Online store design needs to account for who updates products, handles exceptions, and keeps the terms current. That doesn't mean every internal workflow belongs in one project. But the boundary needs to be visible, with someone responsible on the other side.

That's why I'd discuss cost through the work involved: storefront, data, integrations, migration, validation, operations. Designing a few key page templates and preparing a live business for migration are different scopes.

Where I come in

I can connect product selection, interface design, and front-end implementation, validate states, and define data requirements. Complex back-end work and payments need a specialist team.

If the store is becoming a brand's direct sales channel, evaluate that business decision separately. I discuss the economics in building your own store alongside marketplace sales (in Russian). A new storefront doesn't create an audience by itself.

Send me the store and tell me what you want to change. I'll review what should survive, what needs checking before the move, and where separate ownership is needed

Sources and limitations

Google Search Central, Site moves, read November 4, 2026: URL mapping, redirects, sequencing major changes, and ranking fluctuations. The kitchenware store is hypothetical. No guarantee is made that conversions or rankings will remain unchanged.

Other articles

E-commerce website redesign without losing conversions — Andrey Antar