Change CMS without losing SEO

We retain indexed URLs, clicks, impressions, landing pages, external links, conversions, canonicals and indexation rules.

AI measurement stays separate: Google exposes AI Overviews and AI Mode impressions by page, country, device and date; Bing exposes citations and a sample of grounding queries. Those metrics are not comparable.

Every retained URL stays or receives its closest successor. Duplicates may consolidate into one stronger page; removed content without an equivalent should honestly return 404 or 410.

Internal links, menus, sitemaps and campaigns point directly at the new addresses.

We first protect the answer that works: title, facts, media, evidence and useful links. Editorial improvements can follow in measurable batches.

Google recommends changing one major thing at a time when possible. It is less dramatic and far easier to read.

We keep old properties accessible, submit the new sitemap and inspect priority destinations.

We also check the “Search generative AI” setting on every old and new property and its inheritance from a parent property. A correct robots.txt does not prove that AI inclusion is active.

Redirects preserve gclid, gbraid, wbraid, _gl and utm_* parameters. Cross-domain measurement and consent are tested before cutover.

Page resource URLs may change; durable entity identity should not. Organization retains BCE/VAT, iso6523 and sameAs. Search Console, Bing and IndexNow are verified for every allowed host.

PDFs, office documents and other indexable files return an absolute canonical HTTP Link header that agrees with the redirect plan. Old and new hosts are tested directly instead of inferring their state from an HTML canonical.

Every language has a crawlable URL without a cookie or interaction. If Accept-Language, location or a cookie adapts the response, the CDN cache key and Vary stop French, Dutch or English content appearing on the wrong URL.

Third-party scripts and tag containers are frozen and versioned during comparison. Every tag names its owner, purpose, data read, destinations and consent rule; browser requests are compared with the deployed code.

Googlebot fetches only the first 2MB of a URL, including headers. The title, canonical, robots rules and essential structured data therefore stay early in lean HTML. Mobile content remains equivalent to desktop and deferred content appears through the viewport: Google neither clicks nor scrolls, and infinite scroll keeps persistent paginated URLs.

Indexable pages return 200 with useful content and links already present; Google may skip JavaScript rendering for a non-200 status. robots.txt remains below 500KiB and stores no critical rule at its end. Its cutover is tested separately because after a 5xx Google may pause crawling and then reuse the last valid version for weeks.

Campaigns move with the site: Google Ads final URLs, mobile URLs, tracking templates and cross-domain redirects are updated together. For a shop, the new Merchant Center domain is claimed and product feeds, images, prices and availability are resubmitted. The new Search Console property starts its own BigQuery export before cutover; its first export does not backfill history.

Third-party editorial sections, directories and white-label pages are inventoried under Google’s site reputation policy updated on 28 August 2026. A new address is also checked separately in Preferred sources: only domains and subdomains are eligible, and no automatic transfer of user choices is promised.