Site & Infrastructure ArchitectureOctober 22, 202611 min read

Website Redesign SEO: How to Rebuild Your Site Without Losing Rankings (or AI Citations)

Website redesign SEO is the work of carrying a site's rankings and AI visibility across a rebuild: a Search Console baseline, a one-to-one redirect map, unchanged schema identifiers, open AI crawler access, staging blocks removed at launch, and checks at 1, 4 and 12 weeks. Without them, a traffic drop is hard to trace.

You paid for a website redesign. It launched on schedule, the new site looks better than the old one ever did, and a few weeks later the organic traffic line in your analytics has a cliff in it and you're wondering what happened.

It's rarely a design problem. It's a continuity problem. Your rankings don't live in the website design; they sit in URLs, links and signals. These are things that a design brief usually doesn't list, and a rebuild changes them whether anyone planned for it or not.

Website redesign SEO is the work of carrying those signals across the rebuild, in the right order, with a named person checking each one. This article explains how to redesign a website without losing SEO: finding the baseline, creating a redirect map, the two layers other guides skip (entity continuity and AI crawler access), what to do on launch day and in the first twelve weeks.

What a redesign changes underneath

Design

The visible layer, and usually the only one the brief lists.

  1. URLs and redirects

    Where each ranking page lives, and where its old addresses point.

  2. Content

    The words, headings and search intent that earned each ranking.

  3. Internal links

    How your pages point readers and crawlers to each other.

  4. Schema @id and sameAs

    The identifiers that tie your business together as one entity.

  5. Crawler access

    Which bots your host and CDN let through.

  6. Rendering

    Whether your content is in the initial HTML or appears only after JavaScript runs.

Rankings sit in the layers below the design. A rebuild changes them whether anyone planned for it or not.

Will a website redesign hurt your SEO?

It can, and when it does, the loss traces back to specific, checkable failures. Google's site-move guidance lists the common ones:

  • a noindex or robots.txt block left over from development,
  • redirects pointing at the wrong URLs,
  • other crawl errors,
  • a server that can't cope with the extra crawling, and
  • sitemaps still listing old URLs (Google Search Central).

Google also says to expect ranking fluctuation while it recrawls, and that for a small or medium site most pages take a few weeks to move.

How bad can it get? The largest public dataset I found measures domain moves, not same-domain redesigns, so read it as the worst-case scenario of the website migration SEO impact. Across 892 migrations, using third-party traffic estimates, the new domain took 523 days on average to match the old one. The fastest took 19 days, and 17% hadn't matched it after 1,000 days (Search Engine Journal, January 2025). That spread is the case for planning continuity before the build.

Domain moves: how long traffic took to recover

19 days
the fastest move to match the old domain's traffic
523 days
on average to match the old domain
17%
still hadn't matched it after 1,000 days
Domain moves across 892 migrations, measured with third-party traffic estimates (Search Engine Journal, January 2025). Not typical of a redesign that keeps its domain.

Redesign, rebuild or migration: which one are you doing?

It's important to name the type of change first, because it sets the plan.

Visual redesign, same platform

URLs change? No

What it needs Content, internal links and schema carried over unchanged, plus launch checks

New platform, CMS or hosting, same URLs

URLs change? No

What it needs Crawler access, rendering and robots.txt checks on the new stack

New URL structure

URLs change? Yes

What it needs URL inventory, one-to-one 301 map, internal link updates

Domain or subdomain change

URLs change? Yes

What it needs All of the above, plus the Change of Address tool in Search Console

Google's general rule is to change one thing at a time. A new domain, a new platform and a new layout in one launch make any drop impossible to trace, so split them. Whichever row you're in, the steps below double as a website migration SEO checklist. If you're still deciding whether a rebuild is the answer at all, start with The Problem Isn't Your Website.

Before you touch anything: record the baseline

A baseline is a dated copy of everything that's working, taken before the first design is approved. Without one, you can't tell a normal dip from a failure.

Search Console export. Search Console keeps 16 months of data (Google Analytics Help), so a baseline that isn't exported will roll off. Export the Performance report for pages and queries and the Generative AI performance report for pages (it has no query data), and save the Links report, because backlinks are what the redirects have to protect. Store the files outside Search Console.

A crawl of the live site. Record every URL with its status code, canonical, title, H1 and internal links. Include images, PDFs, CSS and JavaScript files, which Google says move the same way pages do.

Your schema, as it is now. Copy the JSON-LD from each live template, with every @id and sameAs value. Check the new templates against it.

An AI citation baseline. Ask ChatGPT and Perplexity, and search Google for, the questions your buyers ask, and save what each says about your business and which pages it cites, with the date. There's no export for this, so a dated screenshot is the record.

A rank and conversion snapshot. Your priority queries, plus the pages that produce enquiries, so the effect on the business is measured, not remembered.

The pages that carry the site. From the export, list the pages that bring the most clicks, enquiries and backlinks. Their content, headings, title tag and search intent carry over to the new site unchanged, unless a reason is written down. A ranking guide rewritten into three lines of homepage-style copy loses the ranking along with the words.

If you'd rather have the layer underneath diagnosed first, that's what the Infrastructure Audit does. You'll receive a prioritised fix list you can hand to whoever builds.

What redirects do you need when you rebuild a website?

Every old URL with impressions, clicks, internal links or backlinks needs a decision, written down before launch. That list is the redirect map, and it's the file your developer builds from. Google says 301 and other permanent redirects don't cause a loss in PageRank, so the work is in the destination (Google Search Central).

Deciding what each old URL needs

  1. Does the old URL have impressions, clicks, internal links or backlinks?

    • No No redirect needed. Let it go.

    Yes

  2. Does it still exist at the same URL?

    • Yes Keep it. No redirect.

    No

  3. Is there a close equivalent on the new site?

    • Yes 301 to that page, in one hop.
    • Merged 301 to the consolidated page.
    • No equivalent Return a 404 or 410. Never send it to the home page.

Don't send many old URLs to the home page. Google says that can confuse users and may be treated as a soft 404. Redirect straight to the final destination, and where that isn't possible keep chains to no more than 3 hops. Use server-side permanent redirects (301 or 308), keep them for at least a year, and don't block old URLs in robots.txt, because a crawler can't follow a redirect on a URL it isn't allowed to fetch. Update your own internal links to the new URLs too.

I used this map on my own site when I merged two overlapping articles into one. The old URL now resolves to Why isn't my website ranking?, the page I kept, and the internal links that pointed at the old one were updated to match. It's a small version of the same job: one decision per old URL, written down before anything moved.

What most redesign checklists miss: entity continuity and AI crawler access

The top-ranking guides I read in October 2026 cover baselines, redirects, sitemaps and launch checks well. None talk about the importance of keeping your schema identifiers stable, checking CDN and AI crawler access on the new host, or testing whether your content is in the initial HTML. Those are the layers AI systems depend on.

Entity continuity. An @id is the label that ties your schema mentions together: one business, one identifier, referenced by every page. sameAs links point that identifier at your profiles elsewhere. If the new templates generate fresh identifiers, every reference to the old ones points at something that no longer exists, and anything reading the markup meets what looks like a new entity. Require the new templates to output the same @id and sameAs values, and validate a sample of pages with Google's Rich Results Test on staging and again after launch. (What Is Entity SEO explains why identifiers matter.)

AI crawler access. A new host or CDN can change which crawlers reach you without anyone editing robots.txt. This is because robots.txt is a request, while a CDN setting is enforcement. Since 15 September 2026, Cloudflare's Block setting also applies to mixed-use crawlers such as Googlebot, Bingbot and Applebot, so choosing it stops them entirely, search included. To stop AI training and keep search, the setting is now Disallow AI Training, and "Block AI Bots" is being retired in favour of separate Search, Training and Agent controls (Cloudflare, 15 September 2026). Check the bot settings on the new platform, then check the live result with the AI Crawler Checker. For which bots to allow and why, see AI Training vs AI Citation.

How a crawler's request reaches your page

  1. AI crawler

    Requests the page.

  2. CDN or host bot settings

    Enforcement

    Can block the request before it reaches anything else.

  3. robots.txt

    A request

    Asks the bot to stay out. The bot decides whether to listen.

  4. Your page

    Reached only if the CDN lets the request through and the bot honours robots.txt.

CDN or host bot settings: On Cloudflare since 15 September 2026, Block also stops Googlebot, Bingbot and Applebot, search included.

Allowed in robots.txt doesn't mean the bot gets through: the CDN decides first.

JavaScript rendering. A rebuild on a new framework can move your content out of the initial HTML and into JavaScript. I often see this with newly built vibe-coded sites. Google renders JavaScript. In Vercel and MERJ's December 2024 measurement, none of the major AI crawlers did, including OpenAI's GPTBot and Anthropic's ClaudeBot. Gemini was the exception, because it uses Googlebot's infrastructure.

What a browser shows

Bookkeeping for small businesses

We close your books every month, so tax time is a non-event.

What GPTBot or ClaudeBot gets

<!doctype html>
<html>
  <head>
    <title>Loading…</title>
  </head>
  <body>
    <div id="root"></div>
    <script src="/assets/index-4f2a.js"></script>
  </body>
</html>
Example page. In Vercel and MERJ's December 2024 measurement, none of the major AI crawlers rendered JavaScript; Gemini was the exception, because it uses Googlebot's infrastructure.

The two-minute test is to request the page without a browser and search it for a sentence from the page. Pick one with no apostrophes or quotes, because the raw HTML often encodes them:

curl -s https://yoursite.com/your-page | grep -c "a sentence from the page"

If the count is 0, that sentence isn't in the initial HTML. The Vibe Coded Website Checklist covers the same checks for sites built with AI tools.

llms.txt. If your old site had an llms.txt file, restore it at the root with working links; What Is llms.txt covers what it does and doesn't do.

Launch day: the website redesign SEO checklist

Run these eight checks on launch day, and give each one an owner before the site goes live. They start with the mistake that can drop your entire site from search results, because it takes minutes to prevent and far longer to fix.

  1. Remove staging blocks. Publish the production robots.txt, not the staging one, and remove noindex from page templates and X-Robots-Tag headers. During the build, keep staging behind a password as well as noindex: a URL blocked only by robots.txt can still be indexed if something links to it.
  2. Test redirects. Spot-check URLs with the URL Inspection tool in Google Search Console, and test the full list with a crawler.
  3. Canonicals and sitemaps. Give each new URL a self-referencing canonical tag. Submit the new sitemap.xml, listing new URLs only, and leave the old one submitted for a few weeks so crawlers find the redirects sooner. Submit both in Bing Webmaster Tools as well: Bing's index feeds Microsoft Copilot.
  4. Search Console and analytics. Confirm verification still works on the new site, and that analytics are firing.
  5. Schema and AI access. Check that sample pages validate with the same @id values, the site passes the AI Crawler Checker, and the curl test finds your content.
  6. Core Web Vitals. Check Google's page-speed and stability measures. Test on staging multiple times using different devices, because a heavier design can slow pages.
  7. Server capacity. Google crawls a new site more heavily than usual after a migration, so confirm the host can take the load.
  8. A rollback plan. Keep the old site and its configuration ready to restore, and agree in advance what would trigger a rollback and who makes the call.

How long does SEO take to recover after a redesign?

Google says most pages on a small or medium site take a few weeks to move, and larger sites take longer. Your own recovery depends on what changed, so check at 1, 4 and 12 weeks against the baseline, in the same order each time: access, indexing, redirects, then performance.

What counts as normal? Some ranking movement in the first few weeks is expected while Google recrawls. What isn't normal: indexed pages falling overall instead of shifting from the old sitemap to the new one, a rise in 404s, or your protected top pages still down at the 4-week check. Any of those means something didn't carry over, so go to the checks below.

Week 1. Crawl the live site and compare it with the baseline crawl. In Search Console, the new sitemap's indexed count should rise as the old one's falls. Check server logs for Googlebot and for URLs returning errors, and re-run the AI Crawler Checker.

Week 4. Compare the Performance report with your export, grouped by page type, and re-run your AI citation questions against the dated baseline.

Week 12. Repeat the comparison, update remaining external and profile links to the new URLs, and keep the redirects live.

If something dropped. Check in this order: a production noindex, robots.txt, canonicals pointing at staging, old URLs returning 404 instead of redirecting, and content cut from pages that ranked. Google's guide to debugging drops in Search traffic is the next step.

Should you redesign now, or fix underneath first?

If the reason for the redesign is that the site doesn't bring any traffic in, check what's underneath it first. A site that looks right and can't be found doesn't need knocking down. It needs the right architecture built underneath it. If a rebuild is the answer, What Is a Bespoke Website explains what it means to build for your business rather than adapt a template.

If the redesign is going ahead, write the continuity plan before the design is approved and hand it to whoever builds: the baseline, the redirect map, the schema identifiers, the crawler settings, and an owner for each launch check.

The continuity plan, stage by stage

  1. Baseline · Before the design is approved

    • Search Console export
    • Crawl of the live site
    • Schema as it is now
    • AI citation baseline
    • Rank and conversion snapshot
    • The pages that carry the site

    Owner

  2. Build · Staging locked, templates checked

    • Staging behind a password as well as noindex
    • New templates output the same @id and sameAs values
    • Redirect map written
    • Core Web Vitals tested on staging

    Owner

  3. Launch day · Eight checks

    • Staging blocks removed
    • Redirects tested
    • Canonicals and sitemaps
    • Search Console and analytics
    • Schema and AI access
    • Core Web Vitals
    • Server capacity
    • Rollback plan

    Owner

  4. Week 1

    • Live crawl compared with the baseline crawl
    • New sitemap's indexed count rising as the old one's falls
    • Server logs checked for Googlebot and errors
    • AI Crawler Checker re-run

    Owner

  5. Week 4

    • Performance report compared with your export, by page type
    • AI citation questions re-run against the dated baseline

    Owner

  6. Week 12

    • Comparison repeated
    • External and profile links updated to the new URLs
    • Redirects kept live

    Owner

Rebuild with discoverability built in from day oneSee the Project→

The redesign isn't what loses the rankings. Skipping the plan for what has to carry over is.

Frequently asked questions

Will redesigning my website hurt my SEO?

It can. Google expects ranking fluctuation while it recrawls a significantly changed site. Losses that last trace to specific failures: URLs changed without 301 redirects, ranking content cut, or a staging noindex or robots.txt block left on the live site.

What redirects do I need when I rebuild my website?

A 301 from every old URL that has impressions, clicks, internal links or backlinks, pointing to its closest equivalent on the new site. Retired pages with no equivalent should return a 404 or 410, and old URLs shouldn't all redirect to the home page.

How long does SEO take to recover after a redesign?

Google says most pages on a small or medium site take a few weeks to move. For domain moves, the largest public dataset (892 migrations) averaged 523 days to match old-domain traffic, and 17% hadn't after 1,000 days (Search Engine Journal, January 2025).

What is a website migration?

A website migration is any change to the URLs, domain, platform or hosting a site is served from. There are four common kinds: a new URL structure, a move to a new domain or subdomain, a new CMS or platform, and a new host or CDN. A visual redesign that keeps every URL on the same platform isn't a migration. A move to a new platform or host is, even when every URL stays the same, because it can change which crawlers get through and what's in the initial HTML. URL and domain changes need a one-to-one 301 redirect map, and a domain change also needs the Change of Address tool in Search Console. Platform and hosting changes need crawler access, rendering and robots.txt checks on the new stack.

How do I redesign a website without losing SEO?

Export 16 months of Search Console data, map every URL with traffic or links to a one-to-one 301 redirect, keep schema identifiers and crawler access unchanged, remove staging blocks at launch, and check results at 1, 4 and 12 weeks.

Does a website redesign affect AI citations?

It can. Most major AI crawlers, Gemini aside, don't run JavaScript (Vercel and MERJ, December 2024), and a new host, CDN or framework can change crawler access or hide content from them. Test the raw HTML and the bot settings after launch.

Should I change my URLs in a redesign?

Keep them wherever you can. Google advises changing one thing at a time, so a redesign that also changes URLs adds risk. Where a URL must change, it needs a one-to-one 301 redirect.

How much does it cost to rebuild a website without losing SEO?

It depends on the size of the site and how much changes. Either way, the continuity work (baseline, redirect map, schema and crawler checks) belongs in the build quote, not in a repair bill after launch. My Infrastructure Project includes it, at $15,000.

›Sources
Aimee Q Devlin—Systems & Infrastructure Architect based in San Miguel de Allende, Mexico

Aimee Q Devlin

Aimee Q Devlin is a Systems and Infrastructure Architect based in San Miguel de Allende, Mexico. She works with founders and operators of established businesses whose sites aren't ranking, converting, or being cited by AI—and builds the infrastructure that fixes it properly. She developed the PRISM Framework, an AEO framework for making founder-led businesses visible to ChatGPT, Perplexity, and the AI engines shaping discovery in 2026. The PRISM Diagnostic is where engagements begin.

About Aimee →

Something isn't working. Let's find out what.

You don't need to have it diagnosed before we talk. That's what the PRISM Diagnostic is for. 60 minutes, USD $350.

See the PRISM Diagnostic