Search traffic losses after a migration are common, usually significant, and almost always caused by the same small set of mistakes. The preparation below takes a day or two and prevents the months of recovery work that follows a botched launch.

Inventory every URL first

Crawl the existing site and export every URL. Cross-reference with analytics and search console data to identify which pages actually receive traffic and which have inbound links. This list is the foundation of the entire migration, and building it after launch is considerably harder.

Map old to new, explicitly

Every old URL needs a destination: the equivalent new page, or the closest relevant page. Avoid redirecting everything to the homepage — search engines treat that as a soft error and the equity is lost. Where content is genuinely removed, return a proper 410 or 404 rather than redirecting to something irrelevant.

Redirect correctly

  • Use 301 permanent redirects, not 302 temporary ones, for moved content.
  • Redirect directly to the final destination — chains dilute and slow.
  • Preserve query parameters where they affect content.
  • Keep redirects in place for a year at minimum; they are not a temporary measure.

The launch-day checklist

  • Remove the noindex directive and any robots.txt disallow used during staging. This is the single most common serious migration error.
  • Verify canonical tags point to the new URLs.
  • Submit the updated sitemap in search console.
  • Confirm analytics and conversion tracking fire on the new site.
  • Check structured data survived the rebuild.

Benchmark before, monitor after

Record current traffic, rankings for key terms, and indexed page count before launch. Without a baseline you cannot distinguish a migration problem from seasonal variation. Afterwards, watch crawl errors and indexed pages daily for the first fortnight — problems surface there before they appear in traffic.

Expect a dip, and know what is not normal

A modest temporary fluctuation while search engines recrawl is normal. A sharp sustained drop is not, and usually traces to indexing being blocked, redirects missing or chained, or a large volume of pages returning errors. Check those three first — they account for the overwhelming majority of cases.