The Perfect Website Replatform

Incomplete redirect mapping, far more often than anything about the new build. Traffic is lost when URLs that existed on the old site have no deliberate destination on the new one, so they resolve to 404s or get funneled to the homepage, which passes almost nothing. The failure is also delayed: rankings hold for days or weeks until search engines and AI crawlers finish recrawling, so teams often declare a migration successful before the loss appears. Build the URL inventory from crawl, analytics, Search Console, and server log data before the rebuild starts, because any URL missing from that list becomes a 404 you will not know about.

No. Shopify URL redirects are one to one, from a specific source path to a specific target, and there is no wildcard or pattern matching in the native redirect tool. That is the single most important constraint to design around when you move a large catalog onto Shopify, because a pattern that would be one rule on Apache or Cloudflare becomes one row per URL in Shopify. For large migrations that means generating the redirect file programmatically and importing it in bulk rather than writing rules. If you genuinely need pattern-based redirects, they have to run in a layer in front of Shopify, such as your CDN or edge proxy.

Shopify appends the original query string to the redirect target in most cases, so tracking parameters such as utm_source generally survive the hop. We tested this on a live proxied store and confirmed the parameter carried through to the target intact. Do not assume it though: if a campaign, an affiliate link, or an internal search URL depends on a specific parameter, test that exact URL shape before launch rather than testing a clean path and hoping the rest behave the same way.

Before the new site's URL structure is finalized, not after. Mapping done late becomes a reconciliation exercise against decisions already made, which is where teams discover that the new platform cannot produce the URL they need. Starting early also means the URL inventory can influence the build: if a template change would orphan a set of ranking pages, that is much cheaper to fix in design than in redirects. On a large catalog, treat mapping as a workstream with its own owner, not a task on the launch checklist.

Pick the window against your own seasonality rather than a general rule. The cost of a bad migration week is proportional to the traffic and revenue moving through that week, so the same mistake costs a gift retailer far more in November than in February. Look at your last two years of traffic and revenue by month, find the trough that is not immediately followed by your peak, and give yourself a full crawl cycle of monitoring inside that quiet period. Our launch window risk tool scores the months against that shape.