Every enterprise replatforming project has a slide that says "preserve SEO equity." Then the timeline compresses, the redirect map gets built from the sitemap two weeks before launch, and organic revenue drops for six months while everyone debates whether it was the migration or the algorithm.
In 2026 there is more at stake. Beyond rankings, a migration can break the pages AI tools cite, the structured data they read, and the product feeds that put you in AI shopping results. None of that shows up on a standard SEO launch checklist. Here are the tactical issues that cause the damage, in the order you will meet them.
Before you choose: rendering is a platform requirement
Many migrations in 2026 are moves to headless or composable architectures. That can be a very good decision. It also introduces the single biggest new risk to AI visibility.
Googlebot executes JavaScript. The major AI crawlers, including GPTBot, ClaudeBot, and PerplexityBot, do not. A storefront that renders product content in the browser can rank in Google and still be close to invisible to AI tools. If your current site is server-rendered and the new one is not, you can lose AI visibility at launch without any change in rankings to warn you.
The fix: write it into the requirements. Server-side rendering or static generation for product, category, and content pages belongs in the RFP and the statement of work, with a testable acceptance criterion: product name, price, availability, key specifications, and structured data must be present in the initial HTML response, verified with a non-rendering fetch.
Issue 1: A redirect map built from the sitemap
The sitemap lists the URLs you think matter. It misses old product URLs still earning links, PDF spec sheets cited by distributors, campaign landing pages, legacy category paths, and the specific pages AI tools have been citing. Those are often the URLs carrying the most accumulated trust.
The fix: build the URL inventory from every source.
- 12 to 16 months of server logs, including crawler and AI user-agent requests
- Search Console pages with impressions or clicks
- Backlink data for any URL with external links
- Analytics landing pages, including AI referral landing pages
- Pages cited in your own AI visibility audits
- Media files such as PDFs, spec sheets, and manuals
Map every URL with value one-to-one to its closest equivalent. Avoid redirect chains, and avoid pointing everything at the homepage, which search engines tend to treat as a missing page rather than a true replacement. Keep redirects in place for at least a year, and longer for URLs with strong external links.
Issue 2: URL cleanup ambitions
A new platform is a tempting moment to fix the URL structure. Sometimes it is worth it. Every changed URL, though, is a page that has to be re-crawled, re-evaluated, and re-trusted, and every changed category path ripples down to the products beneath it.
The fix: change URLs only when the benefit is specific. If the new platform can preserve existing product and category paths, preserve them. Where it cannot, document the pattern and test it at scale. Save structural cleanup for areas with a clear search or usability case, not general tidiness.
Issue 3: Structured data and feed regression
Old platforms often carry years of accumulated schema fixes and feed customizations that nobody documented. The new platform ships with defaults. On launch day, Product markup loses its GTINs, variant structure, and shipping details, and the feed reverts to bare titles.
Feeds carry a second risk. If product IDs change in Merchant Center or other channel feeds, those items can lose their performance history and effectively start over. AI shopping surfaces that read the same data inherit the disruption.
The fix: treat product data as its own migration workstream. Before launch, export structured data and feed output for a representative sample of SKUs from the old site and diff it against staging. Keep feed item IDs stable unless there is a compelling reason to change them. Plan the feed cutover with the same rigor as the DNS cutover. The broader data side is covered in Product Data Is the New Shelf.
Issue 4: Facets and parameters behave differently
Every platform handles filtering, sorting, and pagination its own way. A site that kept faceted URLs under control on the old platform can launch with millions of new crawlable combinations on the new one, simply because the defaults changed.
The fix: crawl staging the way a bot would. Run a full crawl of the staging environment with a crawler set to follow every link, and look for parameter explosions, inconsistent filter ordering, missing canonicals, and empty pages on no-result combinations. Carry your facet indexing decisions over explicitly. Our guide to faceted navigation and AI crawlers covers the policy itself.
Issue 5: Launch-day configuration mistakes
The most expensive migration SEO errors are also the most avoidable, and they keep happening:
- A staging robots.txt or site-wide noindex carried into production
- Canonicals pointing to the staging domain
- XML sitemaps listing old URLs, redirected URLs, or nothing at all
- Hreflang referencing URLs that no longer exist
- Redirects deployed as 302s instead of 301s
- Robots rules for AI crawlers lost, or new blanket bot blocks added by a security or CDN default
The fix: a launch-hour checklist with named owners. Verify each item on production within the first hour, and again at 24 hours, before anyone declares the launch a success.
Issue 6: Monitoring that stops at launch
Migration problems surface over weeks, not hours. Rankings fluctuate, crawl patterns shift, and AI tools keep citing old URLs until they recrawl.
The fix: 90 days of structured monitoring. Track daily for the first two weeks, then weekly:
- Redirect hits and 404s from logs, fixing any valuable URL that was missed
- Indexed pages and crawl stats in Search Console
- Organic and AI referral traffic by template type, compared with the pre-launch baseline
- Merchant Center disapprovals and feed errors
- A repeat of your AI visibility prompt audit at 30 and 90 days
Integration issues often surface in the same window, and they show up in search as broken pricing or stock data. Our piece on why enterprise eCommerce integrations fail covers the usual suspects.
The migration SEO timeline
- Platform selection: rendering requirements, URL flexibility, and feed capabilities in the evaluation criteria.
- Three to four months before launch: URL inventory from all sources, baseline metrics, and an AI visibility audit.
- Six to eight weeks before launch: redirect map complete, staging crawl, structured data and feed diffs.
- Two weeks before launch: redirect testing at scale and launch checklist owners assigned.
- Launch: hour-one and day-one verification.
- Launch plus 90 days: monitoring, fixes, and a post-migration review.
FAQ
Q: How long does organic traffic take to recover after an ecommerce migration? A well-executed migration with complete one-to-one redirects often sees only a short dip, sometimes a few weeks of fluctuation while search engines re-crawl. Migrations with incomplete redirects, URL structure changes, or rendering problems can take many months, and some lost traffic never returns. Preparation is the variable you control.
Q: Will a migration affect how AI tools describe our products? It can. If cited URLs break, if content moves behind client-side rendering, or if feeds lose attributes, AI answers can become less accurate or drop you until their sources update. Redirecting cited URLs, keeping content server-rendered, and preserving feed quality protect that visibility.
Q: Is moving to headless bad for SEO? No, provided the storefront server-renders the content that matters. Headless can give you more control over performance and structure. The risk comes from client-side-only rendering, which most AI crawlers cannot read.
The bottom line
Search equity is lost in migrations through small, preventable gaps: a URL missing from the redirect map, a template that renders only in the browser, a feed that reverts to defaults. Treat SEO and AI visibility as a workstream with its own requirements, owners, and acceptance tests, starting at platform selection rather than two weeks before launch.
If a replatform is on your roadmap, our eCommerce Audit establishes the search and AI visibility baseline, the URL inventory, and the platform requirements before any decision is locked in.