SEO Migration: Move the Site Without Losing the Rankings
A migration is the highest-risk day in a website's life. Done badly it costs 30 to 60 per cent of organic traffic overnight and months to recover. Done properly nobody outside the project notices anything happened, which is the entire point.
Ski.com replatformed with zero rankings lost. That is the outcome we are selling.
THE SERVICE
What an SEO migration is
An SEO migration is the work that preserves search visibility when a site changes in a way search engines notice: a new domain, a new platform, a new URL structure, a redesign that changes templates, a move to HTTPS, or several sites merged into one. It covers benchmarking what you have now, mapping every old URL to its closest equivalent, preserving content and structured data, protecting the paid campaigns and tracking that point at the old URLs, coordinating the launch sequence, and monitoring for months afterwards until the new site has genuinely settled rather than merely stopped alarming anyone.
Most of the work happens before launch, which is why the time to commission it is while the new site is still being designed rather than after it ships. Expect around three months from kickoff to go-live on a site of any real size, then three to six months of monitoring.
Why migrations lose traffic, and it is never the redesign
Google does not rank a website. It ranks individual URLs, each carrying its own accumulated history: backlinks, crawl record, engagement, structured signals. Change a URL without telling anyone where it went and that history resets to nothing. The new page starts from zero regardless of how good it is.
Almost every failed migration comes down to the same short list:
The single largest cause, every time. Blanket-redirecting everything to the homepage is the version that looks like it worked and is not.
PDFs, image files that receive traffic, old campaign landing pages, paginated category pages, parameters, uppercase variants. The inventory is always longer than the sitemap.
Old to intermediate to final wastes authority and slows crawling. Every old URL should land on its destination in one hop.
A redesign that trims copy to fit a nicer template removes the reason the page ranked.
Rarer than it used to be and catastrophic when it happens.
Schema is usually attached to templates, so new templates mean it has to be rebuilt rather than inherited.
Analytics, tag manager containers and conversion tags live in the old templates. When they do not survive the rebuild, the site keeps working and the measurement does not, and nobody notices for a fortnight.
Without a record of what normal looked like, you cannot prove the migration worked or diagnose it when it did not.
None of that is difficult. All of it is boring, and boring work is what gets dropped when a launch date moves.
THE RISK NOBODY BUDGETS FOR
Your Google Ads account breaks on the same day, and it costs money by the hour
Organic damage is slow and visible. Paid damage is immediate and quiet, and the spend continues throughout.
Every final URL in the account points at a page that is about to change. Redirects will catch most of them, but an ad landing on a redirect loads slower and converts worse, and a destination that resolves to a 404 gets the ad disapproved rather than merely underperforming. Sitelinks, callouts, promotion extensions and dynamic ad targets each carry their own URLs and each need their own pass.
Then there is everything sitting behind the ads:
Conversion tracking
The tags live in the old templates. If they are not rebuilt in the new ones, the account keeps bidding with no feedback. This is worse than the account going dark, because Smart Bidding will confidently keep spending against a signal that no longer exists.
Quality Score
Landing page experience is a scored component with its own accumulated history. As far as the account is concerned, a new URL is a new landing page, and a new landing page has no history.
Remarketing and audience lists
A changed tag or a changed domain rebuilds these from zero. On a long sales cycle that is months of audience gone.
Merchant Center
Product feeds carry URLs. A stale feed on an e-commerce migration means disapprovals within days and Shopping revenue off a cliff.
Domain verification and account-level setup
On a domain change, verification, dynamic ad page feeds and brand safety settings all need redoing before the ads will serve properly.
Tracking templates and parameters
If the new site strips query parameters, every UTM and every auto-tagged click stops being attributable, and the reporting you use to judge the migration stops being trustworthy at exactly the moment you need it.
The fix is sequencing, not cleverness. Paid campaigns get repointed or paused as part of the launch runbook, in the same hour as the redirects go live, and verified within minutes rather than at the next weekly check-in.
NEW IN 2026
What a migration risks in 2026 that it did not three years ago
Rankings are no longer the only thing on the table. If your brand is being cited in AI answers, a migration can remove you from those too, and it happens quietly: there is no ranking chart showing you disappeared.
Three things cause it. Losing the structured data that let an engine understand what the page was. Breaking the topical clusters that established you as an authority on a subject. And, on a rebrand, changing the name a model had learned to associate with the answer, which resets the entity rather than moving it.
We benchmark AI citations before the move and check them afterwards, alongside rankings. Most migration suppliers are not looking at this yet, which is a reason to ask whoever you hire whether they are.
OUR PROCESS
How we run a migration
Benchmark before anything moves
A full URL inventory, current rankings, twelve months of traffic, the backlink profile, top-linked pages, Core Web Vitals, the full paid media inventory including every final URL and conversion action, and where you are currently cited by AI engines. If nobody records what normal looked like, nobody can tell whether the migration worked.
Map every URL, one to one
Old to new, each to its closest topical equivalent, in a single hop, checked against the live inventory rather than the sitemap. This is the task that decides the outcome and it is the one most often done in a rush on the Friday. It takes weeks on a site of any size because the inventory is always larger than anyone expects.
Sit in the build, not next to it
Template structure, heading hierarchy, schema, internal links, tracking implementation and content parity get reviewed while the new site is being made, because after it is built these are change requests rather than decisions.
Run the launch sequence
A pre-launch checklist executed in order: noindex removed, redirects live, analytics and conversion tracking verified firing, paid campaigns repointed, SSL confirmed, sitemaps submitted, Merchant Center feed updated, key pages tested within minutes of go-live rather than the following week.
Monitor until it has actually settled
Daily through the first month, then weekly. Google takes 30 to 90 days to fully process a migration, so the monitoring window has to outlast the processing window or you are guessing. A dip is normal and expected. Panic-editing on day three is how a normal dip becomes a real problem.
AFTER LAUNCH
Monitoring is a system, not someone remembering to check
Post-launch monitoring fails when it depends on a person opening dashboards. Attention fades from about week three, which is roughly when the interesting problems start appearing.
So we run it as a monitoring system with defined checks, thresholds and alerts, covering both content and commercial signals:
Redirect integrity
Every mapping in the redirect map retested on a schedule, not once. Redirects rot: chains appear as the development team adds rules over the following months, and a rule added in month three can quietly undo work from month one.
Content parity
Word count, headings, internal links and canonical tags per URL, compared against the pre-launch snapshot. This catches the slow version of content loss, where pages are trimmed in releases after launch and nobody connects it to the traffic.
Indexation and crawl
Coverage checked against the full URL inventory rather than the sitemap, plus 404s pulled from live server logs.
Structured data
Validity per template, because one broken template breaks every page using it.
Rankings, traffic and AI citations
All compared against the baseline rather than against last week.
Paid media health
Final URL status, disapprovals, conversion tag firing, feed status and cost per acquisition against the pre-migration benchmark.
Anything crossing a threshold raises an alert the same day. Everything gets written into a log you keep, so the post-migration report is a record rather than a reconstruction.
What you get
A pre-migration benchmark: URL inventory, rankings, traffic, backlinks, Core Web Vitals, paid media inventory and AI citation baseline
A complete one-to-one redirect map, tested in staging before launch
A content parity check, naming any page that lost substance in the redesign
A structured data plan for the new templates
A paid media migration plan covering final URLs, conversion tracking, audiences and feeds
A launch-day runbook with the sequence and who does what
A monitoring system running for three to six months, daily for the first month and weekly after, with alerts and a written log
A post-migration report comparing everything against the baseline, including what did not recover and why
All of it is yours. If your developers execute rather than us, the redirect map and the runbook still work.
The kinds of migration this covers
Replatform
WordPress to headless, a legacy CMS to something maintainable, one e-commerce platform to another.
Redesign
Same domain, new templates. Deceptively risky, because everyone assumes the URLs are unchanged and they usually are not.
Domain change or rebrand
The highest-stakes version, because the entity moves as well as the URLs, and because the paid account needs reverifying from scratch.
Site consolidation
Several domains or subdomains merged into one, which is mostly a redirect and cannibalisation problem.
Restructure
Same site, new architecture, usually following a topical map.
Internationalisation
Adding languages or markets, where hreflang decides whether it works.
When to call us, and it is earlier than you think
URL structure, templates, tracking and content parity are decisions at that stage and change requests afterwards. Three months is not a padded timeline, it is what a full inventory, a one-to-one map and a build review actually take on a site of any size.
A redirect map built in the last fortnight is worth far more than a recovery project in the first quarter. We will tell you which parts of the process we are compressing and what that costs you.
That is a different piece of work, it is slower, and some of it is not recoverable. We will tell you honestly which parts are before you commit to anything.
What it costs
A migration engagement is scoped by the size of the URL inventory and the type of move rather than by hours. It runs for roughly three months before launch and three to six months of monitoring afterwards, so budget it as a month engagement rather than a project with a delivery date.
If the whole project is with us, design, build and SEO, the migration is part of the build rather than a separate engagement, which is how Ski.com was done.
Worth the comparison: a botched migration typically costs 30 to 60 per cent of organic traffic and takes anywhere from two to five months to recover, depending on the sector, and that figure does not include the paid budget spent on broken destinations and untracked conversions in the meantime. This is one of the few services where the cost of not buying it is straightforwardly calculable from your own analytics.
Questions, answered straight
No fluff, as promised.
How long does a migration take?
Around three months of work from kickoff to launch, then three to six months of monitoring. The pre-launch work is the bulk of it: inventory, one-to-one redirect mapping and reviewing the build while it is still being built. The monitoring runs long because Google takes 30 to 90 days just to process the change, so a shorter window ends before the outcome is known.
Why do you monitor for six months when everyone else says one?
Because most of what goes wrong after launch does not go wrong on launch day. Redirect rules get edited in later releases, pages get trimmed in the next content update, a template change breaks schema across a section. A month of watching catches the launch-day faults and misses all of that.
Will we lose traffic?
A short dip in the first weeks is normal even on a clean migration. A sustained drop is not normal and means something was missed. Ski.com replatformed with zero rankings lost, so it is achievable, but nobody honest will guarantee it in advance.
How long before it settles?
Most of the recovery happens in the first four to eight weeks, with full processing taking up to three months and some sectors taking longer. That is why the monitoring runs to six months rather than stopping while the picture is still moving.
What happens to our Google Ads?
It breaks unless someone plans for it. Final URLs, sitelinks, conversion tags, remarketing audiences, tracking templates and Merchant Center feeds all reference the old site. The account will keep spending regardless, which is the part that catches people out. We inventory the paid setup during the benchmark and repoint it as part of the launch sequence rather than afterwards.
Our developers say they will handle the redirects. Is that enough?
Sometimes, and the question to ask them is whether the map is one-to-one against a full crawl of the live site, or a set of rules against the sitemap. Those are very different things, and the second one is the most common way a competent development team loses a client’s traffic. Worth asking the same question about the conversion tracking.
We have already launched and traffic has dropped. Can you fix it?
Usually a good share of it. Missing and chained redirects are recoverable. Content that was cut in the redesign can be restored. Some link equity lost through months of 404s does not fully come back. We will audit it first and tell you what is realistic before you commit.
Does a redesign on the same domain count as a migration?
Yes, and it is the version that catches people out. New templates change structure, headings, schema and tracking even when the URLs look the same, and "looks the same" and "is the same" are rarely both true after a rebuild.
What about changing domain in a rebrand?
The highest-risk version, because you are moving the entity as well as the URLs. Redirects handle the pages. The brand itself needs the new name stated consistently across your site, your profiles and any third-party source that describes you, or models and engines keep resolving to the old one. The ad account needs the new domain verified before anything serves properly.
Can we keep the old domain live for a while?
The redirects should stay in place for years rather than months, because backlinks pointing at the old URLs keep working through them. Letting the old domain lapse after six months is a slow way to undo the work.
Do you work with our existing agency or developers?
Routinely. We run the SEO side and hand the redirect map and runbook to whoever is executing. It works better when we are in the build conversations rather than sent the URL list afterwards, but it works either way.
Will this affect our AI visibility?
It can, and it is easy to miss because nothing shows on a ranking chart. Lost structured data, broken topical clusters and a changed brand name all reset signals models had built up. We benchmark citations before the move and check them after, alongside rankings.
What if we cannot move the launch date?
Then we work to it and tell you which risks we are accepting. A launch date that cannot move is a normal constraint, not a reason to skip the redirect map. What we will not do is agree to a date and stay quiet about what it costs.
Migrating soon? Talk to us before the new site exists.
Tell us what is moving and when. You will get an honest read on the risk and what it would take to protect, whether or not you use us to do it. If your launch is less than three months away, that conversation is more useful today than next week.