The most common reasons ecommerce platform migrations fail are also the most predictable. That’s because they’re not about the new platform they’re about what hasn’t been done yet to prepare for it. What follows is the first clear, actionable, field-tested process in the business to help you find out what you don’t know, as well as what it’s going to take to get ready for it.
Start with a full URL audit, not a rough one
First, you must identify all indexed and linked URLs from the old website before making any changes to the new one. This includes obtaining URLs from the XML sitemap, the index coverage report on Google Search Console, your analytics platform, and a complete site crawl. You must double-check and cross-reference all these sources since they do not always provide identical lists.
You may be surprised how many seemingly dormant URLs are still indexed or linked, indirectly or directly, from e.g. newsletters, PDFs, or other off-site content. Old blog posts, discontinued product pages, seasonal category pages – all those have likely accumulated some link equity, and some are probably still receiving visitors who arrive via those hidden backlinks.
Given your final list of old URLs, you must prepare a redirect map. It’s just a spreadsheet. In column A, list the old URL. In column B, list the corresponding new destination. The only rule here is that each redirect must point to the most relevant new URL, not the homepage, and not a catch-all category page when a direct product match is available. If users don’t land on the new location of the expected content, they leave. Search engine bots conclude the same thing and may stop checking back for the content’s reappearance.
If you’re changing URL structure, the number of old URLs is likely to balloon exponentially due to adding and removing slash-separated levels on the URL path. Mapping each old URL to the corresponding new URL by hand can be a bit exhausting. There’s no alternative, though. Stick some interns on it. It’s that important. This file plays the most significant role in determining the fate of your search visibility.
Pick your migration window and have a rollback plan
Avoid migrating during your peak sales time. This may seem obvious, but it’s the most common mistake we’ve seen. The “we’ll get to it eventually” migration puts all momentum on hold until a seasonal spike forces your hand. Instead, plan for a genuinely quiet window. Ideally, you will have a nice long runway before your next busy period, which allows you to capture, surface, and fix issues before they hit the bottom line.
Be sure to agree on a rollback plan with whoever is in charge of the technical day-to-day execution before you start. When something inevitably goes horribly wrong in that first 24 hours, how quickly can you be back on the old infrastructure? What’s the specific trigger to revert – a certain error rate, checkout failure rate, a sudden drop in Search Console impressions? Don’t wait until you’re panicking in the control room amidst a launch sequence and stakeholder pressure to figure this out.
Build and stress-test a staging environment
Your test environment should be a clone of the live environment as much as possible including real products, real pricing, and a fully-configured checkout. Testing against a few test products is a waste of time and resources.
Firstly, test your entire site using a tool like Screaming Frog. You’ll want to identify missing meta titles and descriptions. You’ll also want to find pages with thin or duplicate content, empty or broken pages. Look for broken internal links. Check redirects from old URLs. Watch out for redirect loops. Ensure 301 and 302 headers are being used correctly. If you’ve got a staging site on a subdomain, be sure noindex robots directives on the dev site aren’t mistakenly set to disallow in robots.txt (that happens more often than you’d think).
You’re also looking at your canonicals in detail here – category and product pages often generate a lot of duplicates on an ecommerce site and sort orders, paginated listings (and with Magento in particular, filtered category views) will require some thought. If sort order URLs and catalog/view URLs are showing up in your crawl despite being canonical’d out, you’ve got them wrong. If paginated review/result URLs are showing up in your crawl, you probably have them wrong. Canonicals handle passing duplicate content value to the correct page, but not the duplication in the crawl.
Magento brings its own particular complications in that category/views and filter URLs are often not canonical, even when filtered to the same view, so leaving that configuration to someone who has worked with the platform is generally a good idea. If you’re moving to Magento or to a new server, do yourself a favor and hire a specialist magento agency for the first six months. It will save you a world of pain. One of the most common issues I see is redirects and rewrites not working or incorrect configurations on the new server serving old pages. It often goes unnoticed until the new site goes live and the old server is turned off.
Resist the urge to redesign everything at once
We understand the temptation. You are already in contact with the platform, so why not update the homepage, the navigation, and the brand colors, while refreshing everything?
The issue is that if rankings or conversions decrease after launch, you won’t know what caused it. Was it the URL change? The platform’s new rendering engine? The product page layout? The new confusing navigation? You bundled the redesign with the platform migration, multiplied your risk, and made it impossible to isolate the problem.
If a redesign is really necessary, do it weeks or months after the migration. Migrate first, stabilize everything, make sure rankings and sales are identical, and then do the redesign. It’s not just good science to isolate variables, it’s the only way you can hope to get it right.
Launch day: the checklist that actually matters
When you flip the switch, there’s a specific sequence to follow, and skipping steps here is where a lot of otherwise well-planned migrations come undone.
Step one: Submit your new XML sitemap to Google Search Console immediately. Double-check your robots.txt file hasn’t accidentally been left in a “disallow all” state, which happens more often than you’d think when staging configurations get pushed to production by mistake. Re-verify both your old and new site properties in Search Console so you can compare index coverage and search performance across the transition. Update any structured data, and confirm your canonical tags are pointing to live, correct URLs rather than staging URLs left over from testing.
Step two: Update everything outside the website itself. Product feeds for Google Shopping and any affiliate networks need the new URLs, or your shopping ads will start pointing to broken pages within hours of launch. Paid marketing campaigns need the same update. Nothing wastes ad spend faster than sending click traffic to a 404.
The first 48 hours are non-negotiable
Initiate a complete crawl of the live site in the first two hours after launch and conduct additional crawls at 24 and 48 hours post-launch. Probe for identical issues as in staging, but this time those pages are exposed to production traffic.
404 errors are your top priority. Every one of those errors is a broken internal link or a redirect that didn’t get transferred over, and every one of those errors is a small amount of friction for a real customer that wanted to buy something. Since the average shopping cart abandonment rate is already 69.57% according to Baymard Institute’s research, even the addition of a few errors on a product or checkout page will drive those hesitant buyers away permanently. You’re in the danger zone for lost customers and can’t afford to lose any more.
Verify those redirects carefully too. They need to resolve in a single hop, not spend three or four redirects before reaching the page. Redirect chains slow things down and can occasionally break depending on server response. Make sure your ecommerce tracking and analytics events are triggering correctly as well. A botched migration that you aren’t even measuring is far worse than one that you can at least see the total disaster. If the add-to-cart and purchase events simply vanish, then you’re missing the most important data.
Turn the migration into a performance upgrade
Since you’re already rebuilding pages, use that to finally resolve issues you’ve been ignoring. Optimize and right-size images, reduce render-blocking scripts, and tidy up overweight code that’s been slowing your site. Benchmark your Core Web Vitals before the migration and after. You’ll then have actual performance data to tell you if load speed increased and if an increase correlates with better rankings or conversions.
Page speed directly impacts both rankings and whether people stay to convert. A migration that results in faster load times but steady online revenue will in fact be a win if you know to measure that specific result, rather than just assume it must be helping somehow.
Watch the data for a month, not a week
Achieving and maintaining search visibility and sales takes more than three days. It is recommended to monitor organic traffic, keyword rankings, and conversion rates daily for at least 30 days after the new site launches. Fluctuations may occur as search engines recrawl and reindex your new URLs, but significant drops should not be ignored.
Until proven otherwise, assume that any traffic drop was due to something technical. Here’s a quick checklist: Review your redirect map; Revisit crawl errors in Search Console or any new manual actions; Review your canonical tags; and double-check for errant noindex or nofollow directives. It’s almost always one of those things, not a sudden shift in the search algorithm or semantic capacity that just happened to coincide with your site launch.
The pattern behind migrations that go well
The migrations that go well all follow this pattern: thorough preparation, isolated variables, a tested staging environment, and disciplined monitoring after launch. They aren’t difficult on their own. Teams get hung up when these steps are skipped due to time constraints or by assuming the platform will handle them for you. If you stick to the above list, in order, you have the best opportunity to keep your rankings and client base before, during, and after a migration.
