Six weeks after launch, organic traffic is down by a third and nobody can say why. The agency mentions that Google has been volatile. Someone raises a core update. The real answer is usually sitting in the redirect file, decided by whoever happened to pick up the migration ticket that week.
I have audited more than 1,300 Shopify stores, and replatforming damage has a signature. Nothing has been penalised. The damage is structural, and it happens because Shopify makes decisions about your site that your old platform left to you.
Shopify picks your URLs, and you do not get a vote
Brands move to Shopify for the roadmap, and releases like Winter '26 are a fair reason to. What arrives with it is a URL structure you do not control.
Products live under /products/. Collections live under /collections/. Static pages live under /pages/, and blog posts live under /blogs/, plus a blog handle plus the article handle. None of those prefixes is configurable.
Coming off Magento, WooCommerce, BigCommerce or anything custom, that means every URL on the site changes. Not the important ones. All of them.
Most migration plans treat this as a redirect task. It is an information architecture task that happens to need redirects, and the distinction becomes obvious about two months after launch.
There is a second URL problem sitting underneath that one. Shopify will serve the same product at /products/handle and at /collections/some-collection/products/handle. Most themes canonicalise to the short version, which is correct, but plenty of themes still generate the long version inside collection grids and related-product blocks. Your internal links then point at URLs you never intended to rank. Crawl your own theme output before you blame anything external.
The redirect map is usually built from the wrong list
The standard approach is to export the top few hundred pages from analytics and map those. That list is wrong in a specific way. It contains only pages that got traffic inside the window you exported, which leaves out everything seasonal, everything that ranks but converts too poorly for anyone to look at, and every URL that earns links without earning sessions.
Build the list from four sources and take the union.
- A full crawl of the old site, every URL returning 200, captured before anything is switched off
- Search Console top pages across the full 16 months, not the last three
- Analytics landing pages across 12 months
- A backlink export, so you know which URLs other people point at
Then map one to one. What kills sites is the catch-all. Redirecting a few thousand retired URLs to the homepage feels tidy, and Google's own guidance on how to move a site is direct about it, warning not to "redirect many old URLs to one irrelevant single URL destination, such as the home page of the new site" because it "might be treated as a soft 404 error". The equity does not transfer, and you have taught the crawler that a large part of your site is worthless. Where no relevant destination exists, a 404 or a 410 is the honest answer, and it does less harm.
All of this has to happen before the cutover. Once the old site is switched off, you cannot go back for it, and reconstructing a redirect map out of the Wayback Machine is not a job anyone should have to do.
Faceted pages are where the rankings quietly die
On most older platforms, filtered category pages were real, crawlable URLs. Something like /womens-boots/black, or a category path with a colour parameter hanging off it. Over several years, a lot of those picked up rankings for exactly the long-tail terms that convert, because the page matched the query almost word for word.
Shopify's native filtering does not work that way. Filters apply through filter.v.option parameters, and those are not built to be indexed. No setting turns them into static, indexable pages.
If two hundred filtered pages were ranking, you now have two hundred rankings with nowhere to land. A redirect audit will not surface it, because those old URLs get pointed at the parent collection, and the parent collection does not target those terms.
Decide this before the switch. Pull every filtered URL with an impression in Search Console, work out which terms have real demand behind them and build those as proper collections with their own titles, copy and internal links. The rest can collapse into the parent. It is tedious work, and it recovers more traffic than anything else on this list.
Your theme rebuilds the internal linking badly
Old sites accumulate internal links. Category descriptions with links inside them, hand-built cross-links between ranges, breadcrumbs, buying guides pointing at products. None of that is in the product export.
A new theme hands you a menu, a collection grid, and whatever the related-products block decides to do. Products that sat two clicks from the homepage with eight contextual links pointing at them end up four clicks deep with one.
None of this appears in a redirect audit. Every URL resolves, the site works, and it has stopped telling search engines which pages matter.
The repair is unglamorous. You export the internal link graph from the old site, find the pages that lost the most inbound links, and put those links back by hand.
The content that never makes it across
Product data migrates. Content usually does not.
Category descriptions, size guides, FAQ blocks, specification details, the buying advice someone wrote in 2019 that has been quietly ranking ever since. On the old platform, these sat in CMS fields. Shopify gives you a rich text description and metafields, and whether any of it survives depends entirely on whether somebody scoped it.
A crawl catches this one as well. Extract word counts per URL on the old site and compare them against the same pages on staging. Any page that has lost most of its body copy is a page about to lose its rankings, and the crawl shows you that weeks before Search Console does.
What to lock down before you switch
Roughly in order of how much damage each one prevents.
- Full crawl of the old site, archived, before anything is decommissioned
- Redirect map built from the four-source union, mapped one to one, no catch-all to the homepage
- A decision on every filtered URL that has ever recorded an impression
- Titles and H1s identical on day one, so you are changing one variable at a time. Save the Shopify SEO improvements for the month after launch
- Internal link graph compared old against new, with the gaps rebuilt deliberately
- Word counts compared old against new, page by page
- Staging password lifted and the live theme checked for a stray noindex, since a password-protected Shopify store cannot be crawled at all
- Redirects tested on staging, then spot-tested an hour after DNS moves again
This is also the moment to be honest about what else is in the release. Replatforming is the largest change most brands make to their tech stack, and it belongs on its own. Do not replatform, redesign and rebrand in one go. When traffic moves, you will have no way to tell which of the three caused it, and the next quarter becomes a debate rather than a fix. Bundling changes to save a release window is one of the more expensive DTC tech stack mistakes I see.
If you are already live and already down, start with the filtered URLs. Export every query in Search Console that lost impressions since the cutover, find the ones that used to land on a filtered page, and build those collections properly. That is usually where most of the missing traffic went.
