SEO Migration Checklist
A migration is where a site earns or loses a decade of link equity in a single afternoon, and the difference is almost always the redirect map. This checklist covers the fifteen checks that matter, from the inventory and the mapping through to the launch-day checks, the rollback plan, and what recovery actually looks like.
A site migration puts rankings, traffic, and years of accumulated link equity at risk. Across 1,052 domain migrations, the median time for traffic to return to its baseline was 304 days. This SEO migration checklist covers the golden 301 map, staging lockdown, launch sequence, and post-launch monitoring in 15 steps.
A Site Migration Is a Data Transfer
Google does not understand your new domain until you transfer the signals page by page: redirects, sitemaps, metadata, canonical tags. Every top URL that redirects to the homepage leaks its equity. Every URL mapped to its closest equivalent transfers cleanly.
The failures repeat: staging leaked to crawlers, metadata dropped by the new platform, internal links pointing at old URLs, robots.txt blocking the wrong paths. A pre-launch checklist and a post-launch audit prevent each of them.
Moz's migration guide notes Google's documentation on site moves is thin and understates how often migrations cost traffic and revenue. The two checkpoints that matter most are the redirect map and the first 48 hours after go-live.
Plan it like a launch, because it is one. Pick a low-traffic window, write the rollback plan, and treat the first 48 hours as the highest-attention period. This page is part of the complete technical SEO checklist.
Sources
- Google Search Central: keep redirects for at least a year, Google.
- Change of Address links old and new sites for 180 days, Search Console Help.
- 892 migrations, 523-day average, 17% not recovered after 1,000 days, Search Engine Journal, 2025.
- 1,052 migrations, median 304 days, mean 489 days, 22.8% by day 90, SALT.agency, 2026.
- Normal post-migration dip is 10 to 20%; a migration hangover starts at 30 to 50%, Search Engine Journal, 2026.
- The Website Migration Guide: SEO Strategy, Process, and Checklist, Moz, 2018.
Migration Types and Risk
Types of Website Migrations
Not every migration carries the same risk. Google treats them differently.
A website migration is one of nine distinct events. Each one moves a different set of signals, and the size of the risk follows directly from which signals move and which stay put. The table below is the decision matrix: what Google handles for you, what it cannot, and what you must do by hand.
| Migration type | What transfers on its own | What you must do manually | Change of Address | Measured recovery |
|---|---|---|---|---|
| Domain change | Nothing | Every URL must be mapped one to one, redirected, and verified. Submit Change of Address with verified properties on both domains. | Yes | Median 304 days, mean 489 days (1,052 migrations) |
| Merging two sites into one | Nothing | Map both sites against the destination. Keep both domains registered and redirecting. Decide which brand identity the destination carries. | Yes, for each old domain | Treated as domain migrations, so the same 304-day median applies |
| Subdomain to subdirectory | Nothing automatically | 301 every subdomain URL to its new path. Keep the subdomain registered and redirecting. Align canonicals, hreflang, and sitemaps. | Yes | No separate measurement, but it is a domain-class move, so expect the same recovery curve |
| Platform change, URLs unchanged | URL equity and most ranking signals | Metadata must be exported and imported. Confirm rendering, template defaults, canonical tags, and structured data survive the CMS switch. | No | No measured data. Usually weeks, because there is no recrawl of new URLs |
| URL structure restructure | Nothing automatically | Map old paths to new paths with one-to-one 301s. Keep slugs where you can. Resubmit the sitemap. | No | No measured data, but the same researcher's reasoning applies: no authority gap to close, so it tends to be faster than a domain move |
| HTTP to HTTPS, same host | Nearly everything | 301 the HTTP host to HTTPS in one hop. Serve HSTS once stable. Migrate internal links to HTTPS. Do not submit Change of Address. | No | No measured data. Usually days, as a protocol change does not move a URL path |
| www to non-www, or reverse | Nearly everything | Pick one host and 301 the other. Align canonicals, sitemaps, and internal links with the chosen host. Do not submit Change of Address. | No | No measured data. Usually days, as a host change does not move a URL path |
| Design refresh, URLs unchanged | Everything, if URLs stay stable | Content, headings, internal links, structured data, and media must all carry across. The risk is the platform stripping them, not the redirect. | No | No measured data. Usually weeks |
| Server change or CDN move, URLs unchanged | Everything | TTL, DNS, and security headers. Confirm the old service still answers while DNS propagates, then watch for stale responses from cache. | No | No measured data. Usually hours to days |
A domain change is the only migration where every signal has to be moved by hand, which is why the median recovery is 304 days rather than two weeks. Protocol and host changes are the smallest jobs of all, and submitting them to the Change of Address tool is worse than useless: the tool has nothing to move.
The SEO Migration Plan: Before, During, and After Go-Live
Three phases, one rule: nothing transfers itself.
An SEO migration plan runs in three phases. Before go-live you capture a baseline, build the golden 301 map, and lock down staging. On launch day you switch DNS and verify every redirect, robots.txt, and canonical. After launch you submit the Change of Address and track recovery week by week.
Before: 4-6 Weeks Out
Preparation decides the outcome.
- Crawl the live site and export the Search Console URL inventory as your baseline
- Build the golden 301 map covering every old URL to its closest new equivalent
- Inventory titles, meta descriptions, headings, and schema for parity on the new build
- Lock staging behind authentication with a robots noindex
- Dry-run the redirect rules on staging and lower DNS TTL in advance
During: Launch Day
The highest-attention 48 hours.
- Switch DNS, then confirm the new site answers with 200s
- Sample old URLs and verify every one returns a single-hop 301, never a chain
- Check robots.txt is crawlable and blocks nothing it should not
- Remove the staging noindex and confirm canonical tags point to final URLs
- Publish the new XML sitemap and keep the old server live
After: First 8 Weeks
Monitoring turns a dip into a recovery.
- Submit the Change of Address tool for whole-domain moves
- Monitor 404s and redirect gaps daily in week 1, then weekly
- Check index coverage in Search Console every week
- Compare traffic against the baseline and expect the normal U-curve dip
- Keep old-domain redirects alive for at least 12 months
The Pre-Migration SEO Audit
Compare the old site to the new build before anything goes live.
A migration audit is two crawls and one diff. Crawl the old site and the new build with the same tool and the same settings, then join the exports on the URL path. Every difference is a decision: keep it, map it, merge it, or retire it. The table is what to compare.
| Field to compare | What to diff | What a mismatch means | Launch gate |
|---|---|---|---|
| URL set | Old crawl vs new crawl, joined on path | URLs in the old site with no counterpart are unmapped redirects waiting to happen. Every one needs a row in the golden map. | 100% of old crawl URLs have a mapping decision |
| Indexed pages | Search Console Page index report, indexed URLs on both properties | A new build that launches with fewer indexed pages than the baseline has dropped content, not just URLs. | New build indexes at least the baseline count, or the gap is fully mapped to 301s |
| Status codes | 200, 301, 410 on both sides | Soft 404s in the new build waste crawl budget and dilute the transfer. New 200 pages with no body content are the common failure. | No soft 404s. No unplanned 404s among pages that earned impressions |
| Titles | Title text, joined on path | The new platform truncating titles costs CTR and relevance on every affected page. Compare against the old export, not against a rule of thumb. | At least 95 percent of indexed pages carry their original title |
| Meta descriptions | Description text, joined on path | Template defaults overwrite the optimized copy silently. Absent descriptions on high-impression pages cost CTR directly. | Every page in the top 100 by impressions has its original description |
| H1s | H1 text, joined on path | A missing H1 breaks the heading hierarchy and hands relevance to whatever template heading appears instead. | Every template that rendered an H1 before renders one now |
| Canonicals | Canonical target on every new page | A canonical pointing at a redirected URL tells Google the new page is not the one to index. This is the single most common post-launch failure. | Every new page self-references, or points at its final destination |
| Internal links | Link targets across the site | Internal links still pointing at old URLs force crawlers through a redirect on every hop and split crawl budget. | All internal links target new URLs directly |
| Structured data | JSON-LD presence and type on both builds | New platforms often drop schema during import. Lost schema costs rich results and AI crawl confidence. | Every template that emitted schema still emits it |
| Analytics and conversion tags | Tag presence on a live page on the new host | A migration that loses tracking looks exactly like a traffic drop. The data you use to judge recovery is gone before the first day ends. | Test a form submission or a call on the new host before launch |
Run the diff in both directions. The first pass, old against new, finds the pages that did not survive the move. The second pass, new against old, finds the URLs the new build introduced that the old site never had. Both directions feed the redirect map.
A useful sanity threshold is simple. If more than two percent of the old URLs cannot be mapped, the migration is not ready. Finish the map before you argue about anything else.
The 15-Step SEO Site Migration Checklist
From baseline crawl to post-launch recovery monitoring.
| # | Check | Who Does It | Tool | Difficulty |
|---|---|---|---|---|
| 1 | Record a pre-migration baseline crawl | You | Screaming Frog | Easy |
| 2 | Build the golden 301 redirect map | You | Spreadsheet | Hard |
| 3 | Lock down staging with authentication | Developer | Auth / robots | Medium |
| 4 | Keep URLs stable where possible | You | URL planning | Medium |
| 5 | Preserve titles, meta descriptions, and H1s | You | CMS | Medium |
| 6 | Match status codes, avoid soft 404s | Developer | Server / crawl | Medium |
| 7 | Update internal links to new URLs | Developer | CMS / find-replace | Medium |
| 8 | Update canonicals, hreflang, and sitemaps | Developer | CMS | Medium |
| 9 | Launch in a low-traffic window | You | Analytics | Easy |
| 10 | Check robots.txt and noindex after launch | Developer | Test | Easy |
| 11 | Submit the Change of Address for domain moves | You | GSC | Easy |
| 12 | Verify old and new properties in Search Console | You | GSC | Easy |
| 13 | Monitor crawls, indexation, and 404s post-launch | You | GSC / logs | Medium |
| 14 | Track the U-curve and recovery timeline | You | Analytics | Easy |
| 15 | Keep the old site alive with redirects | You | DNS / server | Medium |
Every Migration Check Explained
How to execute each phase of the migration safely.
1 Record a Pre-Migration Baseline Crawl
Before touching anything, crawl the entire site and export every URL, status code, title, and meta description. Record traffic and rankings per page. This baseline is the ground truth you compare against after launch to know exactly what was lost, gained, or missing.
2 Build the Golden 301 Redirect Map
Create a one-to-one mapping of every old URL to its closest new equivalent. Redirect the homepage to the homepage, category to category, product to product. Every old URL with links or traffic needs a destination. Top pages dumped on the homepage leak most of their equity. Avoid redirect chains and loops.
3 Lock Down Staging With Authentication
A staging site indexed by Google is a classic migration disaster: duplicate content, split signals, and diluted rankings. Protect staging with HTTP authentication and a robots.txt disallow, and verify with the URL Inspection tool that Google cannot see it. The staging site must never be publicly crawlable.
4 Keep URLs Stable Where Possible
The best redirect is no redirect. If the platform allows it, preserve existing URL slugs during migration. Every URL you keep unchanged is link equity that never has to travel. Change URLs only when the platform forces it or the new structure is genuinely better.
5 Preserve Titles, Meta Descriptions, and H1s
New platforms frequently drop or truncate metadata. Export the old titles, meta descriptions, and H1s and import them into the new site. Missing metadata means lost CTR and weaker relevance signals. Do not let a template default overwrite years of optimized copy.
6 Match Status Codes, Avoid Soft 404s
Every migrated page must return a real status: 200 for live pages, 301 for redirected pages, and 404 or 410 for genuinely removed content. Soft 404s, pages returning 200 with no content, confuse crawlers and waste crawl budget. Validate the status code matrix after launch.
7 Update Internal Links to New URLs
Internal links still pointing at old URLs force crawlers through redirects on every hop. Find and replace old internal links with new ones sitewide. This preserves internal link equity and keeps the crawl path clean. Redirects for legacy external links stay, but your own navigation should be direct.
8 Update Canonicals, Hreflang, and Sitemaps
Every canonical tag, hreflang annotation, and sitemap entry must reference the new URLs. An old canonical on a new page points Google back to a redirected URL and undermines the transfer. Regenerate the XML sitemap from the new site and submit it fresh.
9 Launch in a Low-Traffic Window
Migrate during the quietest hours for your audience, typically overnight or a weekend for most businesses. A low-traffic launch window limits the number of users hitting errors during the highest-risk period and gives you hours of crawl data before peak traffic.
10 Check Robots.txt and Noindex After Launch
The new platform often ships a restrictive robots.txt or accidentally inherits noindex rules from staging. Test the live site with the URL Inspection tool immediately after launch. A sitewide noindex left in place is a silent killer that looks like a normal migration dip for weeks.
11 Submit the Change of Address for Domain Moves
For whole-domain moves, use Search Console's Change of Address tool. It tells Google the old domain has permanently moved to the new one and only works with verified properties on both domains plus 301 redirects in place. It cannot be undone, so submit it only after the redirects are verified live.
12 Verify Old and New Properties in Search Console
Keep both old and new Search Console properties verified for months. The old property shows redirect traffic and legacy 404s; the new property shows the indexation build. Comparing the two tells you whether signals are transferring or stalling.
13 Monitor Crawls, Indexation, and 404s Post-Launch
Watch the new property for crawl activity, indexed pages, and 404 reports. A spike in 404s means a redirect gap. Check server logs to confirm Googlebot is following redirects and not hitting chains. Fix gaps daily during the first two weeks.
14 Track the Recovery Timeline
Expect the dip and do not panic. A normal post-migration dip is 10 to 20 percent and stabilizes within two to six weeks, according to practitioner reporting in Search Engine Journal. A drop of 30 percent or more with fresh 404s and a falling indexed count is a different problem, and it needs a redirect and indexation audit rather than patience.
15 Keep the Old Site Alive With Redirects
Maintain the old domain with 301 redirects for at least a year. Google's own site move guidance says at least one year, and suggests indefinitely for users, because bookmarks, external links, and direct traffic all depend on it. Letting the old domain lapse or serve dead content destroys residual equity you did not need to lose.
What to Expect After an SEO Site Migration
Two clocks run at once, and they disagree with each other.
Indexing and traffic recovery run on different clocks. Google moves most pages to the new URLs within weeks, but traffic trails because every old URL must be recrawled, every redirect processed, and every pointing link reassigned.
| Recovery window | Share of migrations | Cumulative recovered |
|---|---|---|
| 0 to 30 days | 5 percent | 5.1 percent |
| 31 to 90 days | 18 percent | 22.8 percent by day 90 |
| 91 to 180 days | 13 percent | 35.4 percent by day 180 |
| 181 to 365 days | 24 percent | 59.5 percent by day 365 |
| 1 to 2 years | 24 percent | 83.5 percent by day 730 |
| 2 years and beyond | 16 percent | 86.1 percent by day 1,095 |
The figures cover domain migrations, where the whole site moves. Same-domain moves (URL restructuring only) tend to recover faster: the research attributes the longest recoveries to authority gaps between domains, and a same-domain move has none to close.
Both studies are by the same researcher. The 1,052-migration study found a median of 304 days and a mean of 489 days, with 58 percent recovered within a year and 42 percent still climbing. The earlier 892-migration study found a 523-day average and 17 percent never returned to baseline after 1,000 days. Treat the range as the honest answer, not a single number.
What matters for your launch is the shape of the curve, not the total. A 10 to 20 percent dip that stabilizes inside six weeks with no new 404s is normal volatility. A 30 percent or larger drop alongside a falling indexed count means something is broken, and the fix is a redirect and indexation audit, not more waiting.
Read the curve weekly, not daily. Google reprocesses a migration in waves, so a single bad day tells you nothing. Export the baseline, compare the same week against the same week before the migration, and look at the trend line over four weeks before you change anything.
The First 90 Days After a Migration
A monitoring schedule measured against the recovery curve.
Each checkpoint below is tied to the measured recovery data, not a generic schedule. Four moments in the calendar where a specific measurement tells you whether you are inside the normal band and what to do about it.
| When | What to check | Normal looks like | Time to act |
|---|---|---|---|
| 24 hours | URL Inspection on your 20 highest-value pages | Status shows the new URL as fetched and indexed, canonical resolves, page renders | If Inspection says the live URL is blocked, fix it the same day |
| Fetch robots.txt on the new host | Reachable, and none of your money paths disallowed | Unblock and confirm with Inspection | |
| Search for the noindex flag on live URLs | Zero noindex tags in the served HTML | Remove every noindex the same hour. This silences the site completely | |
| 7 days | Coverage report, new property: 404s on old URLs | Zero spikes. Any single 404 on a previously indexed URL is a gap | Map each 404 to its closest new page and add the 301 |
| Coverage report, index count vs pre-migration baseline | Level with baseline or climbing | If falling, your redirect map has holes. Audit the unmapped URLs | |
| Internal links to old URLs, sitewide | Near zero | Replace with new URLs. Redirects hold, but direct links keep crawl budgets low | |
| Change of Address status, if used | "In progress" with no errors | Errors mean something failed. Read the message and fix what it names | |
| 30 days | Compare traffic against your pre-migration baseline | Down 10 to 20 percent, plateauing rather than falling | A 30 percent or larger drop that is still falling is a broken transfer, not normal volatility |
| Did expected pages index in the new build | Coverage at least as high as the baseline, adjusted for retired pages | Falling index count after 30 days means the site is being de-indexed | |
| Query-level movement in Search Console | Volatility on non-brand queries, stability on brand | Brand queries dropping usually means homepage or sitelinks have broken | |
| Start backlink reclamation on top-linked pages | First reclaimed links arriving | No replies means the outreach copy is wrong, not the link value | |
| 90 days | Traffic versus the pre-migration baseline | Down less than at day 30. The curve should be bending up by now | Only 22.8 percent of domain migrations have fully recovered by day 90. Still down more than 20 percent does not mean failure, but falling indexed count does |
| Backlink reclamation, full pass | Prioritized list from most important URLs, working downward | If nobody answered, test the email and the pitch before resending | |
| Compare against the 90-day expected position | Share recovered above the 22.8 percent benchmark | If still falling, run the diagnostic below |
The clearest signal is whether the indexed count is falling. Rankings falling with the index count holding is normal volatility. Rankings falling and the index count falling is a technical break.
Day 180 and day 365 are worth a calendar entry as well. If day 365 passes and traffic is still below baseline, the cause is almost always one of three things: the redirects leak, internal links bind the old and new domains together, or the new domain's authority is genuinely weaker.
How to Build the Golden 301 Redirect Map
The one artifact every later migration question comes back to.
The redirect map is a spreadsheet with one row per old URL and exactly one destination. Building it from a single crawl is the most common mistake, because a crawl only finds URLs that are linked. Orphan pages, old campaign URLs, and files that earned backlinks all carry value a crawler will never discover on its own.
Six sources for the URL list
- A full crawl of the live site
- Every URL in your XML sitemaps
- Search Console pages with clicks or impressions
- Analytics landing pages with sessions
- URLs with external backlinks
- Your existing redirect rules
Four mapping rules
- One to one wherever a matching page exists
- Merged pages redirect to the merged page
- Retired pages go to the nearest real category, or return 404 or 410
- Never redirect in bulk to the homepage
Template columns
- Old URL
- New URL
- Status
- Traffic (30d)
- Backlinks
- Decision rule
- Source
- Verified
| Old URL | New URL | Status | Decision rule |
|---|---|---|---|
| /blog/site-migration-seo/ | /blog/website-migration-seo/ | 301 | Closest equivalent, one to one |
| /old-product-category/ | /collections/all/ | 301 | Merged into one consolidated page |
| /sale-2019/ | no destination | 410 | Retired with no replacement, so it is removed rather than redirected |
| /index.php?p=142 | /blog/example-post/ | 301 | Parameter duplicate of an existing page |
| /old-blog/post-1/ | /blog/post-1/ | 301 | Path change only, same page and same topic |
Google's guidance is explicit on the last rule. Redirecting many old URLs to one destination, such as a homepage, can be treated as a soft 404, which means Google discards the redirect signal instead of transferring it.
Free Redirect Mapping Template
The exact spreadsheet used to build the golden map, with dropdowns for status and source, the decision rules, and illustrative example rows. Ready to fill in.
Special Cases That Break Most Migrations
The standard checklist does not cover these, and they cause most failures.
| Case | What breaks | What to do instead |
|---|---|---|
| HTTP to HTTPS | A duplicate host serving both protocols splits signals | 301 the HTTP host to HTTPS in one hop. Skip Change of Address, Google handles it |
| www to non-www | Two live hosts competing for the same pages | Pick one canonical host, 301 the other, and align canonicals, sitemaps, and internal links |
| Subdomain to subdirectory | The new path looks like a fresh site with no history | 301 every subdomain URL and keep the old subdomain live and redirecting |
| Merging two sites into one | Duplicate pages compete inside the destination site | Map each old site's pages to its best home, and keep every old domain registered and redirecting |
| Moving to a client-rendered framework | Content that only appears in the browser is invisible to crawlers and AI crawlers | Confirm the main content sits in the server-rendered HTML before launch |
| Changing platform and URLs together | Two variables change at once and nobody can tell which caused the drop | Move the URLs first or freeze the URL structure, so one change can be diagnosed |
| Ecommerce catalog with filter URLs | Faceted URLs multiply into thousands of thin pages | Map only the canonical category and product URLs, and let filter combinations return 404 |
Google's Change of Address tool does not apply to every move. It is for domain and subdomain moves only. HTTP to HTTPS, www to non-www, and path changes within the same domain are handled by canonical tagging and redirects alone, and submitting them through the tool is at best unnecessary.
For a domain move, submit Change of Address for every verified variant of the old domain, including subdomains and both www forms, even the ones you no longer use.
Migrating an International or Multilingual Site
Hreflang and Change of Address behave differently per locale.
An international site is not one migration, it is one per locale, and the rules that govern them are stricter than the rules for a single-site move. Getting the hreflang and canonical setup wrong after a move is the most common way an international site loses its rankings without a single 404 appearing in the report.
Check the bidirectional hreflang before launch. Every page in every locale must link to every other locale, including itself, and the alternate URLs must be live. If one locale page points at an alternate that does not resolve or that returns the wrong locale, Google disregards the whole hreflang set on that page. Run the check across every template in every locale before you flip DNS.
| Locale setup | What moves | Change of Address applies? |
|---|---|---|
| Country or language subdomains (en.example.com) | Subdomain by subdomain, or all at once per domain-level move | Yes, submit one per old subdomain |
| Country or language subfolders (example.com/en/) | Not a domain move at all | No, canonicals and redirects only |
| Separate country domains (example.co.uk) | Domain by domain | Yes, one per old domain |
| Currency, tax, or language plus country split | Same as whichever host holds the canonical | Per the canonical holder above |
Verify every alternate from both sides. Language A pointing at Language B is not enough. Language B must point back, and every parent alternate must resolve in the served HTML on every template. Inspect ten pages per locale before launch. If even one fails, Google treats the whole hreflang declaration as unreliable for those URLs.
After launch, resubmit the sitemap for the new search-console property of each locale. Do not point the old locale's sitemap at the new one until the migration is verified stable, or Google will try to recrawl a thousand pages that do not exist yet.
Migrating PDFs, Images, and Video
Non-HTML assets need redirects too, and they break silently.
A redirect map built from a crawl covers HTML pages. It almost never covers the PDFs, images, and video files that also earned links and rankings. A file dropped from the map is a 404 on a URL that external sites are still sending traffic and link equity to, and nobody notices for months because the analytics dashboard does not show file URLs by default.
| Asset type | What breaks in a migration | The fix |
|---|---|---|
| PDFs and downloads | Path rewrites. Old file URLs return 404 while the new build serves the file at a different path | Map every file URL from the crawl and sitemap, and add 301s for each one to the new path |
| Images | The media CDN host changes, or the path restructure orphan images referenced from old pages | Crawl the new build with image source URLs, confirm every src target resolves, and 301 any image URL with external references |
| Video embed URLs and oEmbed assets | Thumbnail and player URLs regenerate on the new platform, breaking previews in RSS and search results | Regenerate the video sitemap, check thumbnails, and test the embed URLs on both the old and new builds |
| Attachment sitemaps | Attachment sitemap index no longer includes the moved files | Rebuild the attachment sitemap and resubmit it with the rest |
The diagnostic that catches a broken file URL fastest is a spike in 404s filtered by file extension: .pdf, .png, .jpg, .mp4. If your monitoring only watches page URLs, file 404s will sit unnoticed below the fold of every report for months.
Score Your Migration Before You Start
Six questions that determine how much protection the launch needs.
Most migration failures are predictable from six inputs. This migration risk assessment scores each factor, and the total tells you what the launch has to guarantee before anyone presses the button.
| Factor | Low risk (1 point) | Elevated (2 points) | High risk (3 points) |
|---|---|---|---|
| Indexed URLs | Under 500 | 500 to 5,000 | Over 5,000 |
| Monthly organic sessions | Under 1,000 | 1,000 to 10,000 | Over 10,000 |
| URLs with external backlinks | Under 50 | 50 to 500 | Over 500 |
| What changes at launch | Design only, URLs stable | URL structure, platform, or host | Domain change plus any second variable |
| Redirect implementation | One hop, fully tested on staging | Some chains, partial testing | Server-level rules, untested or unmapped |
| Rollback readiness | Written, tested, and someone on call | Written, not rehearsed | None |
Standard guard
A smaller site where a bad outcome costs days, not quarters. The full 15-step checklist is your protection. Keep redirects live for at least a year, re-crawl at day 30, and expect a 10 to 20 percent dip.
Elevated guard
Real revenue at stake. Rehearse the full migration on staging, verify the redirect map against the live build, negotiate a low-traffic window, and get the rollback command confirmed with whoever is on call that weekend. Expect a 10 to 20 percent dip, and plan for four weeks of daily checks.
Business-critical guard
This is the profile of the 13.9 percent of migrations that never fully recovered in the SALT study. Decide the rollback trigger before launch, put a developer on call for 72 hours, and do not launch on a Friday. Budget for the 304-day median recovery and tell the business about it in writing.
Two of these six factors are under your control at the moment you are reading this. Redirect implementation and rollback readiness are free. A staging rehearsal with a full redirect test costs a day. If your score is below 9 because of those two factors, improving them costs nothing and changes your outcome more than any other decision you will make in the project.
Launch Day Hour by Hour
The order below catches the failures in the sequence they would otherwise surprise you.
| When | Action | How you verify it |
|---|---|---|
| 24 hours before | Lower DNS TTL to a few minutes so the switch propagates fast | DNS panel shows the new TTL |
| 24 hours before | Issue the TLS certificate on the new host | Certificate validates on the new hostname |
| 24 hours before | Freeze publishing and write the rollback command down | Rollback command saved in the runbook |
| T0, switch | Switch DNS, keep the old server running | New host answers with 200 |
| T0 plus 15 minutes | Remove the staging block: auth, noindex, robots disallow | URL Inspection shows the page as crawlable |
| T0 plus 30 minutes | Purge the server cache and the CDN, then test redirects | Scripted test returns one hop and a final 200 |
| T0 plus 45 minutes | Run the whole old URL list against the live site | Every row resolves to its intended destination in one hop |
| T0 plus 60 minutes | Check canonicals, hreflang, robots.txt, and the new sitemap | Canonicals self-reference, sitemap loads and lists new URLs only |
| T0 plus 90 minutes | Verify analytics and forms, and test a message from a phone | Test conversion recorded, message arrives in the inbox |
| T0 plus 2 hours | Submit the new sitemap and file Change of Address for domain moves | Search Console confirms both submissions |
Two of these steps catch mistakes that hide for weeks. Purging the cache before testing redirects is the one that fools teams most often: a cached 200 makes every check pass while every visitor still sees the old site. And leaving the staging noindex in place looks exactly like a normal migration dip for weeks, so check it in the first hour, not the first month.
If you changed DNS, confirm the mail records were left untouched and that email still arrives. A migration that silently breaks contact forms costs more in lost leads than it costs in lost rankings.
The Rollback Plan
The step that turns a failed migration into a delay.
A rollback plan that you never use costs an afternoon of preparation. A rollback plan that you need and do not have costs the business. Write it before cutover, not after the first alarm, and store it somewhere the person on call can reach without asking anyone.
- Define the trigger in advance, for example the indexed count falling below half the baseline or a redirect map error rate above a set threshold.
- Keep the old server, the old DNS records, and a pre-migration database backup for at least 30 days after launch.
- Write the exact rollback command, not the intention to roll back, and confirm the person on call has the credentials to run it.
- Do not cancel old hosting on cutover day. Old hosting is the rollback.
- Keep the old domain renewed and paid for at least a year so nobody buys it and points it somewhere else.
Friday evening launches with nobody on call over the weekend are how a small fault becomes a large one. Launch when your developer is available for the following days.
SEO Migration Mistakes That Cost Rankings
The failures behind most post-migration traffic drops.
Redirecting everything to the homepage
Dumping every old URL on the homepage leaks most link equity and breaks user intent. Google's guidance warns this can be read as a soft 404. Map each URL to its closest equivalent.
Leaking the staging site
A crawlable staging environment creates duplicate content and splits signals. Lock it down with auth and robots.txt, then verify with the URL Inspection tool.
Sitewide noindex after launch
Inherited noindex rules from staging look like a normal dip for weeks while the whole site stays out of the index. Check it in the first hour.
Redirect chains and loops
Googlebot can follow up to 10 hops, and Google advises redirecting straight to the final destination instead. Chains add latency for users and lose signal at every step.
Dropped metadata
New platforms truncate titles and descriptions. Without an import, CTR and relevance signals disappear with them. Export before you rebuild.
Letting the old domain die
Lapsing the old domain breaks legacy links and direct traffic. Keep it alive with 301s for at least a year, and keep paying for it.
Changing everything at once
New URLs, new content, new design, and a new domain in one launch leaves no way to diagnose the drop. Move one variable at a time.
Forgetting to purge the cache
A cached 200 makes every redirect test pass while every visitor still sees the old site. Purge the server and the CDN, then test.
Backlink Reclamation After Launch
The step most teams skip, and one of the reasons recovery takes a year.
A 301 redirect tells Google to transfer a link. It does not update the page that carries the link, so every referring domain keeps pointing at a URL that now redirects. Google has to recrawl that page, read the redirect, and reassign the signal. That crawl is the reason recovery takes months rather than days.
Asking the linking sites to update their URLs directly removes the recrawl from the critical path. It is a short email to a small list, and the effect on recovery time is disproportionate to the effort.
- Export your top linking pages from a backlink tool and sort by the number of links each page sends.
- Work down the list, starting with the pages that hold the most links to your most valuable URLs.
- Send one email per site, naming the old URL and the new URL so the editor does not have to look anything up.
- Update your own internal links and your own profiles first, because those are the links you control outright.
- Log the responses. A reclaimed link is worth more than a redirect, and it is measurable.
Google's own site move guidance makes the same point, recommending you update your links and any high-volume links from other websites to point at the new URLs, and to keep redirects in place for the long term regardless.
The Post-Migration Recovery Diagnostic
Match the symptom to the cause before you change anything.
Every post-migration problem presents as one of seven symptoms. Each one has a test that distinguishes it from normal volatility. Run the test that matches what you are seeing, fix that cause, and wait one crawl cycle before touching the next thing. Changing two variables at once means you will never know which one worked.
| Symptom | Most likely cause | The test | The fix |
|---|---|---|---|
| Traffic down 10 to 20 percent, indexed count flat, no new 404s | Normal migration volatility | Compare against your pre-migration baseline, not the week before launch | Change nothing. Check again in one week |
| Traffic down 30 percent or more, indexed count falling | Redirect gap or de-indexation after launch | Search Console coverage report, then URL Inspection on the ten most valuable pages | Map the missing URLs, resubmit the sitemap, confirm with Inspection |
| Traffic stable, conversions or calls down | Analytics or forms broken on the new build | Submit a test enquiry on a phone, check it arrives, check the test conversion records | Fix tracking, fix form delivery, then watch for one full week |
| Some pages recovered, others never came back | Missed or soft-404 redirects for those URLs | Export the indexed set from Search Console, compare against your old crawl | Map and 301 each missing URL to its closest equivalent |
| Old pages show up in the index instead of new ones | Canonicals pointing at redirected URLs | Inspect ten new pages, confirm canonical self-references the new URL | Replace every old canonical reference on the new build |
| Rankings moved but impressions held | Titles or meta descriptions not imported | Diff titles and descriptions against the pre-migration export | Import the original metadata into the new build |
| Traffic still flat after six months | Authority gap or a migration hangover | Check whether the domain changed, and how strong the new domain is | Start backlink reclamation, and revisit the recovery curve expectation |
Three of these seven are detectable before launch with the parity audit above. The rest are detectable within a week of launch if you look. The last one, the six-month flat line, is the one people abandon without doing the reclamation work, and it is the one where a short outreach campaign is most likely to move the number.
Migration Tools and What They Cost
Audit redirects, crawls, and indexation through the transition.
| Tool | Use it for | Cost |
|---|---|---|
| Screaming Frog | Crawling both sites and exporting the URL list that becomes your 301 map | Free to 500 URLs |
| Google Search Console | Change of Address, URL Inspection, index coverage, and 404 reports | Free |
| Google Analytics | The pre-migration baseline and the recovery curve you compare it against | Free |
| Clienvora Redirect Chain Checker | Validating every redirect in the map: one hop, no loops, and a 200 at the end | Free |
| Clienvora Log File Analyzer | Confirming Googlebot follows the redirects and is not hitting errors or chains | Free |
| Backlink tool | Building the reclamation list of pages that still point at old URLs | Paid, subscription |
| Site crawler suite | Continuous post-launch audits for indexation gaps and metadata loss | Paid, subscription |
Related Technical SEO Checklists
Keep exploring the technical SEO series. Every checklist follows the same structure.
Redirects and Status Codes
The 301 and 410 decisions that make or break a migration.
Indexing and Canonicalization
Getting the new site indexed after Google has processed the redirects.
Site Architecture
Plan the new URL structure and internal linking before you move.
Crawl Budget and Log Analysis
Use logs to confirm Googlebot is following the golden map.
Ecommerce Faceted Navigation
Extra care when migrating a catalog with filter URL structures.
Full Technical SEO Audit
Audit before and after launch to confirm nothing was dropped.
When to Bring in an SEO Migration Specialist
The work splits into what a developer can do and what needs an SEO.
Not every move needs outside help. A 50-page brochure redesign that keeps its URLs is a hosting problem with a checklist. A 4,000-URL ecommerce replatform across two domains is a different job, and the failure mode is not cosmetic: it is a quarter of organic revenue.
| Decision point | Handle it in the house | Bring in a migration specialist |
|---|---|---|
| Site size | Under 500 indexed URLs | Over 1,000 indexed URLs, or over 10,000 monthly organic sessions |
| What changes | Design only, URLs stable, same CMS | Domain change, subdomain to subdirectory, or a merge of two sites |
| Redirect complexity | One-to-one map, no chains, all URLs equivalent | Faceted URLs, parameters, or retired pages you must decide on |
| Your developer's experience | Has shipped a migration before and owns the rollback command | Has not shipped one and the business cannot afford a quarter of missing traffic |
| Rollback plan | Written, tested on staging, someone on call | The site is live right now with no rehearsal environment |
| Budget and timeline | A developer with an afternoon spare | Launch this month, revenue depends on the migration going right |
An engagement covers: the golden 301 map built from every crawl, sitemap, analytics landing page, Search Console URL, and backlink profile; the launch runbook and rollback command, written before cutover; a verification pass on the live host covering every hop, status code, canonical, and a form on a phone; and a monitoring plan for the first ninety days.
It cannot cover a guarantee. Even a well-run migration has a median recovery of 304 days. A specialist removes the preventable failure, not the recrawl. If your migration is two variables moving at once, no checklist will fix that for you.
Get Help With Your Site Migration
A migration needs two roles: a developer who owns the server and an SEO who owns the map. Clienvora writes the redirect map, the launch runbook, and the post-launch monitoring plan, then works with your developer to verify every hop and every status code before and after go-live. Start with a free audit of your current URL inventory.
We reply within 24 hours with the redirect risks we can see from your current URL structure.
Site Migration FAQs
Direct answers on redirects, recovery, and risk.
Longer than most teams plan for. Google's own documentation says a small to medium site takes a few weeks for most pages to move, and that redirects should be kept for at least a year. Independent measurements of 1,052 domain migrations put the median recovery at 304 days and the mean at 489 days, with 42 percent of migrations taking more than twelve months. Plan for months, not weeks.
Export every URL from the old site with its traffic, links, and rankings, then decide for each one where it goes. Every URL that had a link or an impression needs a destination, and the default of sending everything to the homepage is what destroys the equity. One old URL, one new URL, one 301, and no chains.
No. Every URL change adds a layer of risk to an event that already carries plenty. Keep the existing paths if you can, and if the platform makes that impossible, change as few as possible and redirect every one of them. A migration with no URL changes is the version where the SEO outcome is predictable.
That the old URLs redirect in one hop to the right new URLs, that the new pages return 200, that the canonical points at themselves, that the sitemap lists only the new URLs, and that the analytics are recording. Then submit the sitemap and watch the coverage report for a week. The first day is for finding the obvious breakages.
Roll back to the previous version, which is why you take a full backup and a database snapshot before you start rather than after. A rolled-back migration costs a day. A half-migrated site with 4,000 broken redirects costs a quarter. The rollback plan is not a precaution, it is the cheapest thing on this list.