Talk to Franz

SEO · 02

Relaunch SEO Consulting.

Franz Enzenhofer at a flipchart with a hand-drawn site structure, explaining it to a workshop group
Photo © Christa Gaigg / Atelier Authentic

A relaunch changes URLs, templates and code at once. No relaunch loses zero traffic. I keep the loss small and short, from the first wireframe to the weeks after go-live.

You hire me when

  • A relaunch, new CMS or new frontend is coming.
  • You merge sites or move domains.
  • Your agencies build, nobody checks what Google will index.
  • The relaunch happened and traffic dropped.

What you get

  • Comps, wireframes and specs reviewed per page type.
  • URL schema and the old-to-new migration matrix.
  • Redirect rules, QA-checked per mapping.
  • Staging audit, go-live sign-off, after-launch check.
  • HTTPS, server and domain moves, one change at a time.

Why a relaunch costs traffic

Google does not see a relaunch. It sees old URLs that now redirect and new URLs it has never ranked. It moves its signals over one URL at a time, after it has crawled both.

Google’s own documentation on site moves says it plainly: “To consider a site move complete, Googlebot will have to visit every URL on your old and new site at least once.” (Google Search Central, How to move a site)

So rankings move even when everything is done right. The goal I set in every relaunch kick-off: zero loss is impossible, 80 percent loss is unacceptable. The target is a controlled, minimal, short-lived loss, and then a stable, search engine optimized base to win new traffic from and kick-start your growth.

What breaks relaunches in 2026

Redirect chains and a forgotten staging robots.txt still happen. They are not where the traffic goes today. These are the failures I measured in my 2026 relaunch projects, and the fix I specified for each:

What breaksWhat I measuredWhat I specify
The new frontend shows Google lessPages on the new frontend minus 40 percent clicks; content only with JavaScript on most page typesMain content in the server HTML, 150 to 250 KB, the three SEO tests on every release
Deploys delete files Google needs95 percent of old JS and CSS files goneKeep build files 16 months, a short plain 404
A consent layer hides the siteGooglebot, robots.txt and sitemaps redirected to a consent page; a site with this setup lost 80 percentNo consent proxy, the banner as an overlay
URLs change by accident7 in 10 ranking URLs suddenly redirectOne page, one URL, links fixed at the source
Old URLs end on the home page or an error1.7k URLs sent to the home page; a third of redirects end in a 404A 301 to a page that answers 200, or a deliberate 410, for every known URL
Sitemaps nobody ownsA sitemap where 0 percent of URLs answer 200; a third of clicks outside every sitemapRedirect sitemap on day 0, new sitemaps on day 14, redirect sitemap gone on day 60
Images redirect to a blocked hostAbout one in eight of Googlebot’s fetches are redirectsImage URLs that answer 200 directly

The new frontend shows Google less than the old one

One domain, three backends: a new JavaScript frontend in front of two older systems. Clicks per month, February 2025 to May 2026:

  • Pages on the new frontend: from about 25k to 15k, minus 40 percent.
  • Old CMS: minus 10 percent.
  • Blog: plus 4 percent.

On a product page of the new frontend, key details were not in the server HTML at all. They loaded on click. The page shipped over 500 KB of HTML, under 2 percent of it text.

On most tested page types, the content was visible only with JavaScript.

What I specify:

  • Main content in the server HTML, visible without a click.
  • 150 KB of unzipped HTML per page, 250 KB at most.
  • The three SEO tests on every page type, before and after every release: PageSpeed Insights, the page with JavaScript turned off, and Google’s live test.

Every deploy deletes files Google still needs

Google crawls a page first and renders it later, with the CSS and JavaScript the stored HTML points to. Google on its render queue: “The page may stay on this queue for a few seconds, but it can take longer than that.” (Google Search Central, JavaScript SEO basics)

Of about 2.5k old JS and CSS files, 95 percent answered 404. I watched one file go from 200 to 404 within an hour of a deploy. Every missing file answered with a 500 KB copy of the start page instead of a short 404.

What I specify: keep every build file for 16 months. A missing file gets a short, plain 404.

A consent proxy in front of the CMS can redirect Googlebot, robots.txt and the sitemaps to a consent page. Google then sees a consent form instead of the site.

I have seen a site with this setup lose about 80 percent of its organic traffic after a relaunch. My slide for the client said: without intervention this is not a risk, it is a prediction.

What I specify: no consent proxy in the new CMS. The banner is an overlay in the page, and robots.txt and the sitemaps answer directly.

The new platform changes URLs nobody decided to change

After the product pages moved to the new frontend, the letter case of their URLs flipped: the old spelling now redirected to a new one.

  • 7 in 10 ranking product URLs now answered with a redirect.
  • Almost 9 in 10 clicks on those pages landed on the old spelling.
  • About 600 blog posts still carried 2.3k links in the old spelling.

A catch-all route turned any unknown path into a user profile, short language codes included. That blocked the planned language paths, while about 40 percent of all clicks went to parameter language URLs that were in no sitemap.

What I specify: one page, one URL, and a URL that carries traffic never changes. Wrong links get fixed at the source. A redirect only limits the damage.

Old URLs end on the home page or on an error

  • A retired section: about 1.7k URLs with 1.2 million impressions in 90 days, all redirected to the home page. Its sitemap still listed them.
  • Almost a third of the permanent redirects I probed ended in a 404, and every 404 showed the start page, with a canonical to it.
  • I crawled 2k URLs as Googlebot before a relaunch: over 1k redirects, almost none of them permanent, the rest 302. Dozens of chains, dozens of redirects into a 404.

What I specified there: every one of the 100k known URLs (16 months of Search Console, 1.7 million clicks) gets a 301 to a page that answers 200, or a deliberate 410. About 15 pattern rules cover about 70 percent of them. The rest is a list a person reviews. No known URL ends in a 404 by accident.

My mapping QA app caught one greedy pattern rule that had produced 7k junk targets, before launch.

Sitemaps nobody owns

  • One sitemap listed 1.4k URLs, and none of them answered 200. Almost none of its 30 sitemaps had ever been submitted in Search Console.
  • Pages that earned a third of all clicks were in no sitemap at all.

What I specify for launch day:

  1. Day 0: the old sitemaps go. A redirect sitemap lists the old URLs, so Google recrawls them and finds the redirects fast.
  2. Day 14: the full new sitemaps go live.
  3. Day 60: the redirect sitemaps go. The redirects stay.

Images redirect to a host Google may not crawl

Redirects were about 150k crawl requests, about one in eight of everything Googlebot fetched. Two thirds of them were product images, sent to an old image host Google may not crawl.

What I specify: image URLs that answer 200 directly, no image redirects.

How I run a relaunch

Franz Enzenhofer in a workshop room, explaining a point to the people at the table

Photo © Christa Gaigg / Atelier Authentic

  1. Before design: URL rules per page type. A URL decision costs nothing in a spec and a lot after launch. As I wrote in my book, if you change a URL, “you are, from an SEO perspective, deleting the old page and creating an entirely new page with the same content.” (Understanding SEO, URL rules)
  2. During build: spec and wireframe review. I review comps, wireframes and tech specs per page type, together with your design and dev agencies. In one 2026 relaunch that was about 35 wireframe screens in three views each (phone first screen, phone full page, desktop), each annotated with the markup it needs.
  3. The migration matrix. Every old URL from every source I can get: sitemaps, Search Console, server or CDN logs, the Wayback Machine, the CMS. Each row gets one target and one expected answer: 200, a 301 to a page that answers 200, or a deliberate 404 or 410.
  4. Staging audit. The three SEO tests on every page type, page weight, rendering, and the matrix run against staging, before anyone fixes a launch date.
  5. Go-live day. The same checks on the live host, the same day: every redirect, robots.txt, noindex, canonicals, sitemaps, and the old JS and CSS files. I write the scripts for each relaunch.
  6. The weeks after. Search Console and logs, old and new URLs side by side, checked again after two weeks and after two months. When traffic drops, I separate what the relaunch did from what Google did anyway. In one 2026 case, splitting the traffic by URL scheme and market showed that a Discover loss was a market re-weighting by Google, not the relaunch.

This site, relaunched on 21 October 2026

On 21 October 2026 I moved fullstackoptimization.com from WordPress to a static site on Cloudflare Pages. Every URL with impressions in 16 months of Search Console data stayed at its path or redirects to its new home in one hop.

The matrix has about 550 rows. They come from six sources:

SourceRows
Wayback Machine, every archived URL that answered 200about 340
Old page and image sitemapsabout 150
Cloudflare logs, last 30 daysabout 40
WordPress system URLs: old sitemaps, feeds, author page, login, shopabout 15
New site files: sitemaps, robots.txt, favicon4
Search Console, 16 months, not found anywhere else1

The sitemaps alone would have missed most of it. The inventory found 250 WordPress thumbnail URLs, file names like autocomplete-680x512.png, whose redirect would have ended on a 404. 270 explicit redirect rules now send every thumbnail size to its original image. Try one: it answers with a 301 to /i/autocomplete.png.

On launch day a script checked 650 things on the live www host, and 0 failed. It fetched every row of the matrix without following redirects and compared status and target.

Every redirect target had to answer 200, with no chain. Every page and every sitemap URL had to be indexable, with one self-referencing canonical, a title, an h1 and valid JSON-LD.

A second run covered the other hosts and protocol variants: 650 checks, 0 failed.

The traffic verdict takes longer. For medium-sized sites Google says it can take “a few weeks or more” before the new URLs show instead of the old ones (source). The checks above are what can be proven on launch day.

Questions I get before every relaunch

Do redirects cost us rankings?

Stop thinking in PageRank. Nobody outside Google can measure it, and in 2026 it is the wrong question. Ask this instead: does every old URL land, in one hop, on a page that answers the same need, is indexable and is linked from the new site? Traffic is lost through URLs without a redirect, wrong targets and the weeks Google needs to recrawl.

How much traffic will we lose?

When URLs change, I budget 20 to 50 percent less traffic for the first months. The work is to keep it at the low end, and short. When URLs stay the same and only the platform changes, the loss comes from what the new platform does differently: rendering, page weight, assets, links.

How long until traffic is back?

For medium-sized sites, Google says it can take “a few weeks or more” for the new URLs to show, and longer for large sites. Missing redirects do not heal with time. They have to be found and fixed.

Can we change domain, CMS and design at the same time?

You can. Google advises against it: “Plan your changes to your site one after the other, not everything at the same time.” (source) So do I. One change at a time means that when a number moves, you know why.

Do we need the Change of Address tool in Search Console?

Only for a new domain or subdomain. Google: “You only need this tool when moving from one domain or subdomain to another”. (source) Not for HTTP to HTTPS, not for www to non-www, not for new paths on the same domain.

When should we call you?

Before the first wireframe. If the relaunch is already live and traffic fell, the Post Relaunch Workshop is for that.

Sources for the Google quotes on this page, read on 2026-10-24: How to move a site and JavaScript SEO basics, Google Search Central. The client numbers come from my own audits, workshops and handovers in 2026.

From real jobs

  • This site, 2026: WordPress to static, about 550 old URLs mapped, 650 checks on launch day, 0 failed.
  • 2026: about 35 wireframe screens annotated in three views each, migration matrix, segmented sitemaps.
  • 2026: CMS relaunch, about 100k old URLs mapped, a 200-point checklist.
  • 2026: half-way to a new JavaScript frontend; rendering, page weight, URL changes, redirects and sitemaps measured per page type.
  • 2026: 2.5k old build files checked, redirect and 404 audit, language URL plan.
  • A manufacturer: relaunch SEO alongside its design agency and its dev agency.
  • Redirects across all subdomains of one large site.
  • Several relaunches of one site over years: migration prep and release testing.
  • Relaunch workshops before go-live, for many years.