Website migration: seo implications

Table of Contents

Summary

This article authored by Nertis Tech SEO Agency discusses the SEO implications of website migration and the ways to mitigate them. We analyse 5 types of migration, discuss business reasoning behind site migrations, suggest 3 biggest SEO problems related to migrations. Discussion of the SEO impact of website migration offers a framework of scenario analysis as well as a template for assessing the potential SEO casualties. Examples include the best and the second best scenarios for migrating several geo-specific websites to one domain.

Also, we offer an methodology of estimation of the number of SEO losses when reporting to internal stakeholders. Additionally, author success things to do pre-migration, suggest a comprehensive migration SEO checklist as well as come up with a multi-step redirection plan that helps mitigate crawl-related risks.

What is website migration?

When we say migration we actually mean a "catch all" term. There are different types of migration, including:


  1. Server migration,
  2. Website redesign,
  3. URL migration, and also
  4. Tech platform migration, and finally
  5. The domain migration, which is to say a new domain. 


So a risk level attached to each type of migration is obviously different: 


  • from the lowest risk of server migration to 
  • the highest risk of domain migration. 


Often, businesses have several types of migrations at once. For example “weare thinking about a domain migration because we are considering moving to another domain and there also might be a server migration. Yet, we are also thinking about redesign or a change in CSS and layout, i.e. design migration, and finally URL migration is also in question”. So, here at least four types of migration are caught by a single term. For the sake of this discussion, we will focus on domain and URL migration, which has the highest impact on SEO.



Why even migrate?

From the business practice migration is a "top down" decision, i.e. this decision was taken by management for internal reasons that are not necessarily aligned with SEO. This may include: 

  • M&A and their aftermath,
  • rebranding / change of name, etc. 

In most of the cases SEOs actually employ a kind of "afterthought style" here, i.e. now that the decision is taken SEO people would only discuss the input of each individual action and put numbers against it, including: 

  1. What's the predicted SEO impact of website migration?
  2. How are the business OKRs going to be affected?
  3. How to mitigate the risk from SEO and business operations?



Why does website migration create a problem for SEO?

Redirection is not 1-to-1: when googlebot comes to a website by external link, the link quality signals (that would apply to the old URL) must be re-applied to the new one, and it’s not necessarily 1-to-1. When googlebot comes to visit and crawl the old URL and gets forwarded to the new one, the crawling capacities (also known as crawl budget) won’t necessarily be 1-to-1 as before. We call it two major risks of redirection and will discuss them (and how to mitigate them) later on.

During a website migration a number of individual factors come into play: the more people, the more parties involved, the more complex the problem gets and the risk of error is rising.

Bad timing risk. So bad timing is about when the migration comes in time with the Google algorithm update, which becomes more often these days. Why is it a problem? Because Google creates a classifier, or a website score, before making an update. And then test some hypothesis during the four weeks time to change this classifier. So, if you migrate in between, a whole new website shows up and Google must recalculate a classifier, test a hypothesis, etc.   


These three high-level considerations create a possible approach to website migration through the lens of SEO: website migration is never a winning bet SEO-wise, so the role of SEO is to help mitigate the negative impacts of it.


Website migration seo impact

SEO impact may be evaluated coming from the scenario analysis of the future migration. Let’s take an example.


Suppose we have several geo-specific websites that we want to migrate to one. ccTLD is a country-code top level domain, for example, .fr. So, we named 5 scenarios of migration depending on their impact on SEO. This includes:


  • What has to be changed,
  • The resulting URL mask (or structure)
  • Coincidence with other types of migration
  • Risk level as well as 
  • The nature of risk involved.

Template for assessing site migration SEO impact

This template ranges the scenarios from Best to Worst depending on the level of perceived SEO casualties. Apparently, the best scenarios will be when only one thing changes at a time and the maximum control over the process is retained.


The best case-scenario

The best case scenario is when we:


  1. Migrate websites to a .com site. 
  2. Turn the ccTLDs or geography domains into folders, i.e. PT become /pt/.
  3. Keep the same URL structure as we used to have and it, for example, after /pt/. 
  4. Make no redesign, no change in template, we just take the old page and put it on a slightly changed URL where the a ccTLD becomes a folder. 
  5. Create a plan of migration 12 weeks before and show it to every interested party,  then interested parties would agree that once signed and sealed, we make no changes whatsoever. 
  6. Take specific care of using analytics: decide measurement shall we consider most important is what is a "vanity" measuring.  
  7. Migrating Google Analytics because we must need the data to be looking in the same direction.


Pros: 

  • maximum control of the whole process: migrating smaller geos in the first case, then taking a pause and going to the larger ones one by one.
  • maximum "granularity": URL migrates 1-to-1 with only a minor change in the path,
  • purity of experiment: only 1 thing changes within the duration of experiment.


The drawbacks of this scenario:

  • higher exposure to a "bad timing" risk, since we use a migration sequence (a "reverse side" of maximum control).
  • relatively low amplitude of change in traffic (low impact) in the first place, which means that we potentially can’t see a single point where we sent wrong (if it’s the case).



Second-best scenario

Here the Geo-folders become geo+language pairs, not geo only as above. For example, not PT for Portugal, but PT-PT for Portugal AND Portuguese language or PT-BR for Portugal and Brazilian language. So, we'd eventually have several sub-sites from one geography. Yet, from the very beginning we do only one: we migrate everything into one geography. PT-PT. 


We do virtually the same, but we reserve the option for us to eventually (as a second step) make a PT-BR version for this website. Why is it better to do this later on? Or why is it a second best scenario?  


Because It actually involves the bigger crawl budget: we will need a bot to index all Portuguese pages (except for those taken to Brazil) and then go to another sub folder, PT-BR and index some Brazilian ones too. Google may be reluctant to index in the first place without backlinks coming from Brazil.


Pros:

  • maximum flexibility (the "future proof" option): may add more country-language pairs and implement the "country of destination" idea, which might be good for monetisation,
  • good granularity IF we migrate everything into one geo-language folder in the beginning,


Cons:

  • It's not the end: additional migration may be needed in the future
  • Not so many countries have "natural" options for Geo-Language sub-sites (thinking of EN, PT, ES, etc).


How to calculate the possible SEO losses?

SEO casualties can be calculated on the basis of data you already have in case of apparent losses.

URL mask change for a segment of pages

Our experience with redirect suggests that, for example, if you redirect example.com/category to example.com/main-category/category, then allow for at least 30% traffic loss in this segment of pages (it can be used in communication with your stakeholders).

Timeframe for recovery: 2 months.

Permanent losses: 10-15% of clicks may be lost for good.  


Things to do before migration

Here’s the minimal Pre-migration SEO Master Checklist:


  1. Recrawl the whole website.
  2. Create a small list of pages to be crawled (you don't need them all to detect problems).
  3. Have your SEO crawler software, i.e. Botidfy, Screaming Frog or Sitebulb, etc filled with credits.
  4. No-index the new domain from Google! NOT TO FORGET!
  5. Create and fill in the checklists as per each point of the plan of migration.


Below the migration checklist organised by main stages and areas.

Website Migration SEO Checklist

How to mitigate the negative impact of website migration of SEO

Main risks for SEO coming from website migration are:


  1. Potential PageRank loss that occurs when not 100% of link signals are being transferred from old URLs to new ones through redirection
  2. Potential change in crawl activity occurring when bots that hit the old URLs are being redirected to new ones without 


Arguably, the crawl budget decline risk is the most profound one. It springs from a "deferred crawling", i.e. when Google visits pages and does not crawl it right away, instead creates a list of URL to return and crawl with regards to the:


  • "crawl budget" capacity available and
  • internal "importance" factors, e.g. Internal PageRank, number of incoming links, page depth, etc
  • urgency, e.g. frequency of updates of the content
  • external "user" factors: whether page Has Impressions / Clicks.


Consider what happens when a bot has created a list of URLs to crawl and the website migration is performed (in the meatime).


So, when a robot comes to hit the pages from his (already) created list, it will be redirected and after several attempts will need to create a whole new list (abandon the old). This, in turn, has two implications:


  1. Volume of Visits and Crawls will drastically increase, until all the pages are being (discovery) visited and (discovery) crawled
  2. Internal "importance" factors are "force" re-calculated based on - first - discovery crawls, and - second - "refresh" crawls


In the meantime, while the pages are "blank tablet" vis a vis the ranking signals, the site-wide ranking signals are being applied, which is what we normally call a SEO loss due to a migration, i.e. certain well performing URLs will get (temporary) demoted vs before. 


Now, a force re-calculation of ranking signals is rarely 1-to-1 as before. Some pages (and even page segments) might have less attention from Google vs before. Therefore, it's critical to avoid several types of migration at once and, most importantly, to avoid change of internal linking scheme, i.e. internal "importance" factors, which may result in:


  • more lengthy standby period, i.e. more traffic losses,
  • wrong pages being promoted (vs the right ones). 


PageRank loss mitigation

Loss of Pagerank is normally mitigated through active link building in the migration phase (PR, etc campaigns) that drive external links to new URLs. Yet, most probably, at least a "soft" setback in SEO traffic is inevitable following a migration.

Crawl-related risks mitigation

Standby period and ranking signals recalculation risk is mitigated by: 

  1. staged release to allow for estimation, etc
  2. multi-step redirection plan that makes up for deferred crawl in advance
  3. splitting in time all (other) types of migration, i.e. keeping design stable at least for some time before (and after)


Multi-step plan how to Migrate website without losing seo traffic

If you want to practically fence from crawl-related SEO risk, below is a multi-step migration plan, which provides a solution for “deferred crawl” issues. Here, destination domain refers to old URLs, targeted domain - new URLs.


  1. Make sure the Destination domain is indexable. 
  2. Create a new Targeted Domain. 
  3. Load a portion of content to the Targeted Domain, e.g. one or several categories, blog, etc. If not possible, load all content. 
  4. Put canonical links from Targeted domain to Old domain: indicate that content belongs to the Old domain. It's counter-intuitive, but it's so: we keep the same content on 2 domains and must indicate the original one.
  5. Reverse canonicals for a portion of content (!): from Old to the Targeted domain.
  6. Add a crypto-canonical where the canonicals have been reversed, i.e. specify the source of content and submit a link to the Old domain.
  7. Point the links from Old domains Header and Footer to a new Targeted Domain.
  8. Reverse canonicals for all of content: from Old to the Targeted domain.
  9. Put Targeted Domain in hreflangs (if you use them).
  10. Use "Change of address" in GSC
  11. Make 310 redirect from Old to the Targeted Domain and keep it:
  12. it has to be a page to page redirect (write rules)
  13. where not possible, redirect to Home.


Article written by: Bohdan Lytvyn, a founder of Nertis SEO Agency