The SEO loss during a shop migration does not occur on launch day. It happens weeks beforehand, in a spreadsheet that nobody maintains: the redirect mapping. By the time the first 404 errors start appearing in Search Console on the evening of the go-live, the decision has long since been made. Only then does it become clear what was missing from the project plan right from the start.
Yet a platform change is not an SEO gamble. Google itself documents quite precisely how a migration should proceed, and adheres to these guidelines. Rankings fluctuate during the migration phase; only those shops that introduce technical errors suffer permanent losses. It is precisely these that we’ll be looking at.
What is really at stake during a migration
Your organic ranking is not an attribute of your brand. It depends on specific URLs: the category page that has been ranked in position 3 for ‘buy work gloves’ for the past four years, the product pages with inbound links, and the advice article that drives traffic every month. Each of these URLs carries a history of backlinks, user signals and indexing data. And when switching platforms, these URLs almost always change: Magento structures category paths differently to Shopware 6, Shopify enforces its /products/ scheme, and even when switching from Shopware 5 to Shopware 6, the technical URL structures change behind the scenes.
From Google’s perspective, this means a known page disappears and an unknown one appears. Put simply: a migration without a redirect strategy isn’t a move, but a demolition followed by a rebuild. The postman doesn’t know the new address, and regular customers find themselves facing a building site. This hits shops hardest that generate a large proportion of their turnover via organic category pages; those who sell primarily via adverts and newsletters will notice it later, but the impact will be more insidious.
The redirect mapping is the actual project
If you take one thing away from this article, let it be this: the redirect mapping is not a sub-item in the launch plan. It is the project. Everything else is platform configuration.
A clean mapping means: every old URL is assigned exactly one new destination, via a permanent 301 redirect, to the new page with the most relevant content. The old category page points to the new category page, the product page to the new product page, and the guide to the new guide. Google has officially confirmed on several occasions that permanent redirects do not lose any PageRank signal; the Site Move documentation states that the signals from the old URLs are transferred to the new ones. So link equity is transferred if you allow it.
Three mistakes reliably ruin this in practice:
The blanket redirect to the homepage. The classic mistake made under time pressure: anything without an obvious destination is redirected en masse to the homepage. Google treats such redirects as soft 404s, i.e. effectively as deleted pages. The signals from the old URL go nowhere. A redirect to the category page most closely related to the topic is justifiable in exceptional cases, but a blanket redirect to the homepage is never acceptable.
Redirect chains. The old URL redirects to an intermediate version, which redirects to yet another version, which eventually reaches its destination. Googlebot follows a maximum of ten redirects; three is the recommended limit, and each redirect costs loading time and crawl budget. Particularly tricky after several consecutive relaunches: the redirects from 2021 point to the 2023 structure, which is now moving again. When mapping, always point to the final destination, not the last intermediate stop.
The incomplete inventory. You can only redirect what you know. A complete URL inventory never comes from a single source: a crawl of the old shop (e.g. using Screaming Frog), Google Search Console, Analytics data from the last twelve months, and backlink targets from a tool such as Ahrefs or Sistrix. Only the intersection reveals the URLs that haven’t been linked to internally for a long time but still rank or carry links. Don’t forget parameter URLs and pagination.
And afterwards? Redirects aren’t just launch ‘decoration’ to be cleared away after three months. Google recommends keeping redirects active for at least a year. For URLs with valuable backlinks, the rule is: there is no reason ever to deactivate these redirects. ‘Tidying up’ old redirects, which is regarded as standard practice in many IT departments, is self-sabotage from an SEO perspective.
URL structure: keep what works
The second strategic decision is made before the mapping: what will the new URLs actually look like? There’s a rule here that regularly sparks debate: a migration is not the right time for a URL facelift.
Every avoidable URL change increases the risk without offering any benefit. Anyone who switches platforms and changes the domain and overhauls the information architecture and and, whilst doing so, merges half the categories, ends up with four areas of work on the go simultaneously and will subsequently be unable to tell which of them cost them traffic. If a reorganisation of the architecture is long overdue in terms of content, it belongs in a separate project phase, clearly distinct from the platform change.
Technically, this is rarely an obstacle. Shopware 6, for example, allows URL patterns for categories and products to be defined largely freely via SEO URL templates. In many cases, the existing structure of the old shop can simply be replicated, and redirect mapping is reduced to those instances where there are genuine differences. We have described in detail how such a migration works in practice and where the typical pitfalls lie, using the example of Magento to Shopware 6.
First, decide which URLs need to change. Then map only those. Not the other way round.
What else needs to be moved apart from URLs
Redirects preserve link equity, but not relevance. A page ranks because its content and signals match the search query. During replatforming, it is surprisingly common for the very elements that underpinned the ranking in terms of content to be lost.
The title tags and meta descriptions of the old shop have evolved over the years and are often manually optimised. If the new platform replaces them with generic templates (“Product name | Shop name”) during import, every page loses its snippet profile. The same applies to structured data: product markup with price, availability and reviews determines whether rich snippets appear and influences whether AI searches select your shop as a source. Google AI Overviews have been live in German search results since March 2025, with AI Fashion following in October 2025. A shop that loses its markup during a site migration weakens its position in precisely the channel that is currently growing. What it means to remain visible and citable for AI searches is a topic in its own right.
Added to this is the unspectacular, painstaking work: internal links within content that still point to old URLs and will all go through a redirect after the launch. Canonical tags that point to old shop URLs. hreflang markup for multilingual shops. Image URLs, which also need to be mapped for shops with significant Google Images traffic. And loading time: if the new shop is noticeably slower than the old one, you’ve just migrated against your own interests when it comes to Core Web Vitals.
One tool on this list deserves its own paragraph because it’s so counterintuitive: the transitional sitemap containing the old URLs. At launch, you’ll naturally submit the sitemap with the new URLs. However, it’s also worth temporarily providing a second sitemap containing the old URLs. Why would you feed Google addresses that no longer exist? Precisely for this reason: Google specifically crawls the old URLs, finds the 301 redirects there, and processes the migration faster than if it were to revisit every old URL at some point in the future. Side effect: by checking the index status of the two sitemaps in the Sitemap Report, you can see how far along the migration actually is.
None of this is complicated. It’s a checklist that someone has to take responsibility for. In projects without a clear SEO lead, nobody takes responsibility for it.
Launch and the first 90 days
Going live is not the end of the SEO project, but the halfway point. Immediately after launch, it is decided whether errors will persist for hours or months.
On the launch day itself: run a crawl using the old URL list and check that every URL reaches the correct destination via a single 301 redirect. Submit the new sitemap to Search Console. If there’s been a domain change, also register the move via the Change of Address tool; for URL changes within the same domain, this tool isn’t available; in such cases, the redirects handle all communication with Google.
And then there’s the launch mishap that’s rife across the industry: the noindex tag from the staging environment. Every test environment is set to “do not index” for good reason. If this setting is inadvertently carried over to the live system at go-live, Google begins removing the shop from the index page by page, silently and without an error message. This often only becomes apparent once traffic has already plummeted. Robots.txt and meta-robots tags should therefore be checked within the first few hours after launch, not during the first week.
After that, the monitoring phase begins. In the event of a domain change, two properties run in parallel in Search Console: the old one shows how the index is being depopulated, whilst the new one shows how it is being built up. If the domain remains the same, there is a single property, and you monitor index coverage, 404 reports and the performance of the most important pages there. Fluctuating rankings in the first few weeks are to be expected and are explicitly described in Google’s documentation. What you really need to respond to: rising 404 numbers (add any missing redirects), soft 404 errors (resolve bulk redirects) and pages that, after several weeks, are neither ranking nor indexed. A weekly review of the top 50 money keywords is no substitute for a tool, but it turns abstract indexing trends into a business metric.
When is the job done? When organic clicks have reached the previous level and index coverage is stable. For a medium-sized website that has been migrated properly, Google talks in terms of weeks, not quarters. If, after two or three months, the curve remains significantly below the previous level, this is not a matter of fate, but a symptom: Somewhere, one of the tasks described above remains unfinished.
Migration without SEO loss is a craft, not a risk
Back to the table at the start. Whether your migration is SEO-neutral isn’t determined by some ingenious trick, but by a few hundred lines of mapping that someone has conscientiously maintained, and by a checklist that someone takes responsibility for right through to the end. It’s tedious and unglamorous. But it’s also the reason why we believe replatforming projects can be planned without any permanent loss of visibility: the mechanics are documented, the errors are known, and none of them are new.
If you’re about to switch platforms and want to know where your shop is vulnerable from an SEO perspective, let’s take a look at it together before the first URL is migrated. That’s exactly what our e-commerce consultancy is for. The sooner the mapping is in place, the cheaper the rest of the project will be.