11 November 2026 · wevisualise

Redesign, Rebuild or Migrate: Changing a Website Without Losing Rankings

Organic traffic falls after a relaunch far more often than it should, and the design is almost never the cause. The cause is URLs that changed without redirects, or redirects built the night before launch. This sets out the three scopes, a working migration method, a redirect mapping example and a pre-launch checklist.

Three scopes, defined precisely

Most projects go wrong because nobody agreed which of these was being bought. The distinction is not cosmetic. It determines how much search risk you are carrying.

What changes URLs change? Search risk
Redesign The visual layer: templates, typography, imagery, layout. Same platform, same content, same addresses. No Low
Rebuild The underlying build: new platform or new codebase, content carried across, often a new information architecture. Usually, in part Medium to high
Migration The location of the content: new domain, new subdomain, HTTP to HTTPS, or a wholesale URL restructure. Yes, by definition High

A redesign that leaves every URL intact is close to risk free from a search perspective. A rebuild that reorganises the navigation and renames every service page is a migration wearing a redesign's clothes. Ask early which one you are doing, and price the technical work accordingly.

Why relaunch traffic drops

Google Search Central lists site moves and migrations among the main causes of a drop in organic search traffic, alongside algorithmic updates, technical issues, security threats, spam violations and seasonality. That is the honest framing: a migration is a known risk category, not bad luck.

Google is also clear that some movement is expected. Its site move documentation states that you should "expect temporary fluctuation in site ranking during the move", and that "the visibility of your content in Search may fluctuate temporarily during the move. This is normal and a site's rankings will settle down over time." On duration, it says "a medium-sized website can take a few weeks for most pages to move in our index; larger sites can take longer."

The difference between a dip that recovers in three weeks and a decline that never recovers is almost always the redirect layer. When an old URL returns a 404 instead of a permanent redirect, the ranking signals attached to that page have nowhere to go. No amount of design quality compensates for that.

Benchmark before anything changes

You cannot prove a migration worked if you did not record what you had. Do this before development starts, not after.

Export a full Search Console Performance report for the old site. Google's guidance on debugging search traffic drops recommends using the Performance report with a 16 month date range and comparison filters, so pull the full window while the data still describes the old site. Record clicks, impressions and average position by page and by query.

Export analytics sessions and conversions by landing page for the same period. Record your top fifty landing pages by organic entrance, because those are the URLs where a redirect mistake will cost you the most. Save the crawl, the exports and the date you took them in one folder. This is your evidence file.

Verify both the old and the new property in Search Console. Google's site move documentation is explicit that you should verify both sites across all variants, including the www and non-www versions and HTTP and HTTPS, because these are treated as separate properties.

Crawl & inventory every URL

Google recommends compiling old URLs from sitemaps, server logs and analytics, then mapping each one to its equivalent on the new site. Three sources, because no single one is complete. A crawler finds what is linked, analytics finds what gets traffic, and server logs find what search engines and other systems actually request.

Build a single spreadsheet with one row per old URL and these columns: old URL, current status code, organic entrances in the last twelve months, number of referring domains if you have that data, the new URL, and the intended status code. Sort by traffic. If you run out of time, the top of that list is where the time must go.

Build the redirect map

Google Search Central states that permanent redirects show "the new redirect target in search results", while temporary redirects show the source page. It also states that for permanent redirects "the indexing pipeline uses the redirect as a signal that the redirect target should be canonical". That is the mechanism you are relying on, so use the right tool.

Server-side permanent redirects, HTTP 301 or 308, are what Google describes as "the best way to ensure that Google Search and people are directed to the correct page". Meta refresh with a zero second delay is treated as permanent and a delayed one as temporary. On JavaScript, Google's position is blunt: "Only use JavaScript redirects if you can't do server-side or meta refresh redirects."

Three rules from the same documentation that catch most teams out. Keep chains short: Google advises against chains longer than three hops, supports a maximum of ten, and notes that chaining "adds latency for users". Do not redirect a large set of old URLs to one irrelevant destination such as the homepage. And on how long to leave them in place, Google says "keep the redirects for as long as possible, generally at least 1 year".

A workable map looks like this.

Old URL New URL Status Note
/services/branding.html /brand-identity 301 Direct equivalent, highest traffic page
/about-us/team.php /about 301 Two pages merged into one
/blog/06/seo-tips /insights/seo-tips 301 Date removed from path
/products/widget-blue-v1 /products/widget-blue 301 Superseded variant
/services/ /services 301 Trailing slash normalised at server level
/spring-offer-lp none 410 Genuinely retired, no equivalent exists

The last row matters. Where nothing equivalent exists, a 410 is more honest than forcing a redirect to a page that does not answer the same question. Google's site move guidance calls for correct 404 or 410 responses for content that has been deleted.

What else has to survive the move

Redirects get the attention. These four get forgotten.

Titles. Google generates the title link in search results from several sources, the <title> element among them, and asks for titles that are descriptive, concise, unique per page and free of boilerplate. A rebuild that replaces every hand-written title with a template pattern such as "Page name | Site name" will change how a large number of results are presented in one day. Carry the old titles across, then improve them deliberately afterwards.

Canonicals. Google ranks canonicalisation signals by strength: redirects are "a strong signal that the target of the redirect should become canonical", rel="canonical" annotations are also strong, and sitemap inclusion is weak. It recommends self-referencing canonical links on canonical pages and absolute rather than relative URLs. Check that your new templates emit a self-referencing canonical and that it points at the live URL, not the staging domain.

Structured data. Google's general structured data guidelines require markup to describe content visible on the page, and note that missing required properties make an item ineligible for rich results. If your old product or article pages carried valid markup and the new templates do not, you lose those result types.

Sitemaps. Google states that "submitting a sitemap is merely a hint", that it should contain the URLs you want to appear in search results as fully qualified absolute URLs, and that it uses the lastmod value "if it's consistently and verifiably accurate". Generate the new sitemap before launch, submit it once the redirects are live, and remove the old submission afterwards.

Staging checks & the launch sequence

Run these in order. Numbers one to nine happen before launch.

  1. Confirm the staging site is blocked from indexing with a noindex rule, and that robots.txt does not block crawling of it. Google warns that if robots.txt blocks the page, "the crawler will never see the noindex rule, and the page can still appear in search results", so the two are not interchangeable.
  2. Crawl staging and compare the URL list against the inventory from the old site. Every old URL should appear in the map, and every new URL should be reachable from the navigation.
  3. Verify each new template outputs a unique <title>, a self-referencing canonical, and the structured data the old template carried.
  4. Test the redirect map against staging in bulk, not by hand. Check the status code and the final destination, and flag anything returning 302, 404, or more than one hop.
  5. Confirm internal links point at final destination URLs rather than relying on the redirect layer.
  6. Check that the robots.txt you intend to publish is the production version, not the staging version.
  7. Confirm analytics and consent tooling are installed on the new templates and firing.
  8. Generate the new XML sitemap and check it lists only canonical, indexable URLs.
  9. Agree the rollback plan and who authorises it.
  10. At launch, remove the staging noindex rule, publish the production robots.txt, and confirm the redirects are live on the production server.
  11. Submit the new sitemap in Search Console and remove the old one.
  12. If the domain has changed, use the Change of Address tool. Google requires that you own both properties in the same account, that 301 redirects are already in place, and notes it works at domain level only and cannot be used for HTTP to HTTPS moves. Once submitted, it prioritises the new site and forwards signals for 180 days.

The monitoring window

For the first fortnight, check daily. Google's site move documentation points to the Sitemaps report to watch indexing shift from old URLs to new, the Index Coverage report to see old-site indexing fall as new-site indexing rises, and server logs to see how Googlebot is crawling. Its traffic drop guidance adds the Crawl stats and Page indexing reports as the place to look for a corresponding spike in errors.

Watch for three specific things: a rise in 404s from URLs that were in your inventory, redirect chains introduced by a later change, and pages appearing with the staging domain in a canonical tag. After the first fortnight, move to weekly for eight weeks. Compare against the benchmark you exported, not against last week.

Do not judge the migration in week one. Do not stop watching in week three.

What to do next

  1. Decide which of the three scopes you are actually buying, and put the answer in the project brief this week.
  2. Export your Search Console Performance data and your top organic landing pages now, before any development begins.
  3. Ask your supplier to show you the redirect map as a deliverable with a due date, not as a launch-day task.
  4. Book the post-launch monitoring in the calendar, with a named owner, before the launch date is agreed.
  5. Check whether your current staging environment is indexable. It is a five minute check and a common source of duplicate content.

Losing a quarter of your organic traffic to a relaunch is a preventable outcome, and the prevention is unglamorous: an inventory, a map, and someone checking status codes. If you are planning a change and want the technical work scoped alongside the design, our web design team plans migrations that way.

Sources