Posted by Modestos
What is a site migration?
A site migration is a term broadly used by SEO professionals to describe any event whereby a website undergoes substantial changes in areas that can significantly affect search engine visibility — typically substantial changes to the site structure, content, coding, site performance, or UX.
Google’s documentation on site migrations doesn’t cover them in great depth and downplays the fact that so often they result in significant traffic and revenue loss, which can last from a few weeks to several months — depending on the extent search engine ranking signals have been affected, as well as how long it may take the affected business to rollout a successful recovery plan.
Quick access links
Site migration examplesSite migration typesCommon site migration pitfallsSite migration process1. Scope & planning2. Pre-launch preparation3. Pre-launch testing4. Launch day actions5. Post-launch testing6. Performance reviewSite migration checklistAppendix: Useful tools
Site migration examples
The following section discusses how both successful and unsuccessful site migrations look and explains why it is 100% possible to come out of a site migration without suffering significant losses.
Debunking the “expected traffic drop” myth
Anyone who has been involved with a site migration has probably heard the widespread theory that it will result in de facto traffic and revenue loss. Even though this assertion holds some truth for some very specific cases (i.e. moving from an established domain to a brand new one) it shouldn’t be treated as gospel. It is entirely possible to migrate without losing any traffic or revenue; you can even enjoy significant growth right after launching a revamped website. However, this can only be achieved if every single step has been well-planned and executed.
Examples of unsuccessful site migrations
The following graph illustrates a big UK retailer’s botched site migration where the website lost 35% of its visibility two weeks after switching from HTTP to HTTPS. It took them about six months to fully recover, which must have had a significant impact on revenue from organic search. This is a typical example of a poor site migration, possibly caused by poor planning or implementation.
But recovery may not always be possible. The below visibility graph is from another big UK retailer, where the HTTP to HTTPS switchover resulted in a permanent 20% visibility loss.
In fact, it’s is entirely possible to migrate from HTTP to HTTPS without losing so much traffic and for so such a long period, aside from the first few weeks where there is high volatility as Google discovers the new URLs and updates search results.
Examples of successful site migrations
What does a successful site migration look like? This largely depends on the site migration type, the objectives, and the KPIs (more details later). But in most cases, a successful site migration shows at least one of the following characteristics:
Minimal visibility loss during the first few weeks (short-term goal)
Visibility growth thereafter — depending on the type of migration (long-term goal)
The following visibility report is taken from an HTTP to HTTPS site migration, which was also accompanied by significant improvements to the site’s page loading times.
The following visibility report is from a complete site overhaul, which I was fortunate to be involved with several months in advance and supported during the strategy, planning, and testing phases, all of which were equally important.
As commonly occurs on site migration projects, the launch date had to be pushed back a few times due to the risks of launching the new site prematurely and before major technical obstacles were fully addressed. But as you can see on the below visibility graph, the wait was well worth it. Organic visibility not only didn’t drop (as most would normally expect) but in fact started growing from the first week.
Visibility growth one month after the migration reached 60%, whilst organic traffic growth two months post-launch exceeded 80%.
This was a rather complex migration as the new website was re-designed and built from scratch on a new platform with an improved site taxonomy that included new landing pages, an updated URL structure, lots of redirects to preserve link equity, plus a switchover from HTTP to HTTPS.
In general, introducing too many changes at the same time can be tricky because if something goes wrong, you’ll struggle to figure out what exactly is at fault. But at the same time, leaving major changes for a later time isn’t ideal either as it will require more resources. If you know what you’re doing, making multiple positive changes at once can be very cost-effective.
Before getting into the nitty-gritty of how you can turn a complex site migration project into a success, it’s important to run through the main site migration types as well as explain the main reason so many site migrations fail.
Site migration types
There are many site migration types. It all depends on the nature of the changes that take place to the legacy website.
Google’s documentation mostly covers migrations with site location changes, which are categorised as follows:
Site moves with URL changes
Site moves without URL changes
Site move migrations
These typically occur when a site moves to a different URL due to any of the below:
A classic example is when migrating from HTTP to HTTPS.
Subdomain or subfolder change
Very common in international SEO where a business decides to move one or more ccTLDs into subdomains or subfolders. Another common example is where a mobile site that sits on a separate subdomain or subfolder becomes responsive and both desktop and mobile URLs are uniformed.
Domain name change
Commonly occurs when a business is rebranding and must move from one domain to another.
Top-level domain change
This is common when a business decides to launch international websites and needs to move from a ccTLD (country code top-level domain) to a gTLD (generic top-level domain) or vice versa, e.g. moving from .co.uk to .com, or moving from .com to .co.uk and so on.
Site structure changes
These are changes to the site architecture that usually affect the site’s internal linking and URL structure.
Other types of migrations
There are other types of migration which are triggered by changes to the site’s content, structure, design, or platform.
This is the case when a website is moved from one platform/CMS to another, e.g. migrating from WordPress to Magento or just upgrading to the latest platform version. Replatforming can, in some cases, also result in design and URL changes because of technical limitations that often occur when changing platforms. This is why replatforming migrations rarely result in a website that looks exactly the same as the previous one.
Major content changes such as content rewrites, content consolidation, or content pruning can have a big impact on a site’s organic search visibility, depending on the scale. These changes can often affect the site’s taxonomy, navigation, and internal linking.
Mobile setup changes
With so many options available for a site’s mobile setup moving, enabling app indexing, building an AMP site, or building a PWA website can also be considered as partial site migrations, especially when an existing mobile site is being replaced by an app, AMP, or PWA.
These are often caused by major changes to the site’s taxonomy that impact on the site navigation, internal linking and user journeys.
These can vary from major design changes in the look and feel to a complete website revamp that may also include significant media, code, and copy changes.
In addition to the above, there are several hybrid migration types that can be combined in practically any way possible. The more changes that get introduced at the same time the higher the complexity and the risks. Even though making too many changes at the same time increases the risks of something going wrong, it can be more cost-effective from a resources perspective if the migration is very well-planned and executed.
Common site migration pitfalls
Even though every site migration is different there are a few common themes behind the most typical site migration disasters, with the biggest being the following:
Some site migrations are doomed to failure way before the new site is launched. A strategy that is built upon unclear and unrealistic objectives is much less likely to bring success.
Establishing measurable objectives is essential in order to measure the impact of the migration post-launch. For most site migrations, the primary objective should be the retention of the site’s current traffic and revenue levels. In certain cases the bar could be raised higher, but in general anticipating or forecasting growth should be a secondary objective. This will help avoid creating unrealistic expectations.
Coming up with a detailed project plan as early as possible will help avoid delays along the way. Factor in additional time and resources to cope with any unforeseen circumstances that may arise. No matter how well thought out and detailed your plan is, it’s highly unlikely everything will go as expected. Be flexible with your plan and accept the fact that there will almost certainly be delays. Map out all dependencies and make all stakeholders aware of them.
Avoid planning to launch the site near your seasonal peaks, because if anything goes wrong you won’t have enough time to rectify the issues. For instance, retailers should avoid launching a site close to September/October to avoid putting the busy pre-Christmas period at risk. In this case, it would be much wiser launching during the quieter summer months.
Lack of resources
Before committing to a site migration project, estimate the time and effort required to make it a success. If your budget is limited, make a call as to whether it is worth going ahead with a migration that is likely to fail in meeting its established objectives and cause revenue loss.
As a rule of thumb, try to include a buffer of at least 20% in additional resource than you initially think the project will require. This additional buffer will later allow you to quickly address any issues as soon as they arise, without jeopardizing success. If your resources are too tight or you start cutting corners at this early stage, the site migration will be at risk.
Lack of SEO/UX consultation
When changes are taking place on a website, every single decision needs to be weighted from both a UX and SEO standpoint. For instance, removing great amounts of content or links to improve UX may damage the site’s ability to target business-critical keywords or result in crawling and indexing issues. In either case, such changes could damage the site’s organic search visibility. On the other hand, having too much text copy and few images may have a negative impact on user engagement and damage the site’s conversions.
To avoid risks, appoint experienced SEO and UX consultants so they can discuss the potential consequences of every single change with key business stakeholders who understand the business intricacies better than anyone else. The pros and cons of each option need to be weighed before making any decision.
Site migrations can span several months, require great planning and enough time for testing. Seeking professional support late is very risky because crucial steps may have been missed.
Lack of testing
In addition to a great strategy and thoughtful plan, dedicate some time and effort for thorough testing before launching the site. It’s much more preferable to delay the launch if testing has identified critical issues rather than rushing a sketchy implementation into production. It goes without saying that you should not launch a website if it hasn’t been tested by both expert SEO and UX teams.
Attention to detail is also very important. Make sure that the developers are fully aware of the risks associated with poor implementation. Educating the developers about the direct impact of their work on a site’s traffic (and therefore revenue) can make a big difference.
Slow response to bug fixing
There will always be bugs to fix once the new site goes live. However, some bugs are more important than others and may need immediate attention. For instance, launching a new site only to find that search engine spiders have trouble crawling and indexing the site’s content would require an immediate fix. A slow response to major technical obstacles can sometimes be catastrophic and take a long time to recover from.
Business stakeholders often do not anticipate site migrations to be so time-consuming and resource-heavy. It’s not uncommon for senior stakeholders to demand that the new site launch on the planned-for day, regardless of whether it’s 100% ready or not. The motto “let’s launch ASAP and fix later” is a classic mistake. What most stakeholders are unaware of is that it can take just a few days for organic search visibility to tank, but recovery can take several months.
It is the responsibility of the consultant and project manager to educate clients, run them through all the different phases and scenarios, and explain what each one entails. Business stakeholders are then able to make more informed decisions and their expectations should be easier to manage.
Site migration process
The site migration process can be split into six main essential phases. They are all equally important and skipping any of the below tasks could hinder the migration’s success to varying extents.
Phase 1: Scope & PlanningWork out the project scope
Regardless of the reasons behind a site migration project, you need to be crystal clear about the objectives right from the beginning because these will help to set and manage expectations. Moving a site from HTTP to HTTPS is very different from going through a complete site overhaul, hence the two should have different objectives. In the first instance, the objective should be to retain the site’s traffic levels, whereas in the second you could potentially aim for growth.
A site migration is a great opportunity to address legacy issues. Including as many of these as possible in the project scope should be very cost-effective because addressing these issues post-launch will require significantly more resources.
However, in every case, identify the most critical aspects for the project to be successful. Identify all risks that could have a negative impact on the site’s visibility and consider which precautions to take. Ideally, prepare a few forecasting scenarios based on the different risks and growth opportunities. It goes without saying that the forecasting scenarios should be prepared by experienced site migration consultants.
Including as many stakeholders as possible at this early stage will help you acquire a deeper understanding of the biggest challenges and opportunities across divisions. Ask for feedback from your content, SEO, UX, and Analytics teams and put together a list of the biggest issues and opportunities. You then need to work out what the potential ROI of addressing each one of these would be. Finally, choose one of the available options based on your objectives and available resources, which will form your site migration strategy.
You should now be left with a prioritized list of activities which are expected to have a positive ROI, if implemented. These should then be communicated and discussed with all stakeholders, so you set realistic targets, agree on the project, scope and set the right expectations from the outset.
Prepare the project plan
Planning is equally important because site migrations can often be very complex projects that can easily span several months. During the planning phase, each task needs an owner (i.e. SEO consultant, UX consultant, content editor, web developer) and an expected delivery date. Any dependencies should be identified and included in the project plan so everyone is aware of any activities that cannot be fulfilled due to being dependent on others. For instance, the redirects cannot be tested unless the redirect mapping has been completed and the redirects have been implemented on staging.
The project plan should be shared with everyone involved as early as possible so there is enough time for discussions and clarifications. Each activity needs to be described in great detail, so that stakeholders are aware of what each task would entail. It goes without saying that flawless project management is necessary in order to organize and carry out the required activities according to the schedule.
A crucial part of the project plan is getting the anticipated launch date right. Ideally, the new site should be launched during a time when traffic is low. Again, avoid launching ahead of or during a peak period because the consequences could be devastating if things don’t go as expected. One thing to bear in mind is that as site migrations never go entirely to plan, a certain degree of flexibility will be required.
Phase 2: Pre-launch preparation
These include any activities that need to be carried out while the new site is still under development. By this point, the new site’s SEO requirements should have been captured already. You should be liaising with the designers and information architects, providing feedback on prototypes and wireframes well before the new site becomes available on a staging environment.
Review the new site’s prototypes or wireframes before development commences. Reviewing the new site’s main templates can help identify both SEO and UX issues at an early stage. For example, you may find that large portions of content have been removed from the category pages, which should be instantly flagged. Or you may discover that some high traffic-driving pages no longer appear in the main navigation. Any radical changes in the design or copy of the pages should be thoroughly reviewed for potential SEO issues.
Preparing the technical SEO specifications
Once the prototypes and wireframes have been reviewed, prepare a detailed technical SEO specification. The objective of this vital document is to capture all the essential SEO requirements developers need to be aware of before working out the project’s scope in terms of work and costs. It’s during this stage that budgets are signed off on; if the SEO requirements aren’t included, it may be impossible to include them later down the line.
The technical SEO specification needs to be very detailed, yet written in such a way that developers can easily turn the requirements into actions. This isn’t a document to explain why something needs to be implemented, but how it should be implemented.
Make sure to include specific requirements that cover at least the following areas:
Meta data (including dynamically generated default values)
Canonicals and meta robots directives
Copy & headings
Main & secondary navigation
Internal linking (in any form)
Hreflang (if there are international sites)
Mobile setup (including the app, AMP, or PWA site)
Custom 404 page
Page loading times (for desktop & mobile)
The specification should also include areas of the CMS functionality that allows users to:
Specify custom URLs and override default ones
Update page titles
Update meta descriptions
Update any h1–h6 headings
Add or amend the default canonical tag
Set the meta robots attributes to index/noindex/follow/nofollow
Add or edit the alt text of each image
Include Open Graph fields for description, URL, image, type, sitename
Include Twitter Open Graph fields for card, URL, title, description, image
Bulk upload or amend redirects
Update the robots.txt file
It is also important to make sure that when updating a particular attribute (e.g. an h1), other elements are not affected (i.e. the page title or any navigation menus).
Identifying priority pages
One of the biggest challenges with site migrations is that the success will largely depend on the quantity and quality of pages that have been migrated. Therefore, it’s very important to make sure that you focus on the pages that really matter. These are the pages that have been driving traffic to the legacy site, pages that have accrued links, pages that convert well, etc.
In order to do this, you need to:
Crawl the legacy site
Identify all indexable pages
Identify top performing pages
How to crawl the legacy site
Crawl the old website so that you have a copy of all URLs, page titles, meta data, headers, redirects, broken links etc. Regardless of the crawler application of choice (see Appendix), make sure that the crawl isn’t too restrictive. Pay close attention to the crawler’s settings before crawling the legacy site and consider whether you should:
Ignore robots.txt (in case any vital parts are accidentally blocked)
Follow internal “nofollow” links (so the crawler reaches more pages)
Crawl all subdomains (depending on scope)
Crawl outside start folder (depending on scope)
Change the user agent to Googlebot (desktop)
Change the user agent to Googlebot (smartphone)
Pro tip: Keep a copy of the old site’s crawl data (in a file or on the cloud) for several months after the migration has been completed, just in case you ever need any of the old site’s data once the new site has gone live.
How to identify the indexable pages
Once the crawl is complete, work on identifying the legacy site’s indexed pages. These are any HTML pages with the following characteristics:
Return a 200 server response
Either do not have a canonical tag or have a self-referring canonical URL
Do not have a meta robots noindex
Aren’t excluded from the robots.txt file
Are internally linked from other pages (non-orphan pages)
The indexable pages are the only pages that have the potential to drive traffic to the site and therefore need to be prioritized for the purposes of your site migration. These are the pages worth optimizing (if they will exist on the new site) or redirecting (if they won’t exist on the new site).
How to identify the top performing pages
Once you’ve identified all indexable pages, you may have to carry out more work, especially if the legacy site consists of a large number of pages and optimizing or redirecting all of them is impossible due to time, resource, or technical constraints.
If this is the case, you should identify the legacy site’s top performing pages. This will help with the prioritization of the pages to focus on during the later stages.
It’s recommended to prepare a spreadsheet that includes the below fields:
Legacy URL (include only the indexable ones from the craw data)
Organic visits during the last 12 months (Analytics)
Revenue, conversions, and conversion rate during the last 12 months (Analytics)
Pageviews during the last 12 months (Analytics)
Number of clicks from the last 90 days (Search Console)
Top linked pages (Majestic SEO/Ahrefs)
With the above information in one place, it’s now much easier to identify your most important pages: the ones that generate organic visits, convert well, contribute to revenue, have a good number of referring domains linking to them, etc. These are the pages that you must focus on for a successful site migration.
The top performing pages should ideally also exist on the new site. If for any reason they don’t, they should be redirected to the most relevant page so that users requesting them do not land on 404 pages and the link equity they previously had remains on the site. If any of these pages cease to exist and aren’t properly redirected, your site’s rankings and traffic will negatively be affected.
Once the launch of the new website is getting close, you should benchmark the legacy site’s performance. Benchmarking is essential, not only to compare the new site’s performance with the previous one but also to help diagnose which areas underperform on the new site and to quickly address them.
Keywords rank tracking
If you don’t track the site’s rankings frequently, you should do so just before the new site goes live. Otherwise, you will later struggle figuring out whether the migration has gone smoothly or where exactly things went wrong. Don’t leave this to the last minute in case something goes awry — a week in advance would be the ideal time.
Spend some time working out which keywords are most representative of the site’s organic search visibility and track them across desktop and mobile. Because monitoring thousands of head, mid-, and long-tail keyword combinations is usually unrealistic, the bare minimum you should monitor are keywords that are driving traffic to the site (keywords ranking on page one) and have decent search volume (head/mid-tail focus)
If you do get traffic from both brand and non-brand keywords, you should also decide which type of keywords to focus on more from a tracking POV. In general, non-brand keywords tend to be more competitive and volatile. For most sites it would make sense to focus mostly on these.
Don’t forget to track rankings across desktop and mobile. This will make it much easier to diagnose problems post-launch should there be performance issues on one device type. If you receive a high volume of traffic from more than one country, consider rank tracking keywords in other markets, too, because visibility and rankings can vary significantly from country to country.
The new site’s page loading times can have a big impact on both traffic and sales. Several studies have shown that the longer a page takes to load, the higher the bounce rate. Unless the old site’s page loading times and site performance scores have been recorded, it will be very difficult to attribute any traffic or revenue loss to site performance related issues once the new site has gone live.
It’s recommended that you review all major page types using Google’s PageSpeed Insights and Lighthouse tools. You could use summary tables like the ones below to benchmark some of the most important performance metrics, which will be useful for comparisons once the new site goes live.
Old site crawl data
A few days before the new site replaces the old one, run a final crawl of the old site. Doing so could later prove invaluable, should there be any optimization issues on the new site. A final crawl will allow you to save vital information about the old site’s page titles, meta descriptions, h1–h6 headings, server status, canonical tags, noindex/nofollow pages, inlinks/outlinks, level, etc. Having all this information available could save you a lot of trouble if, say, the new site isn’t well optimized or suffers from technical misconfiguration issues. Try also to save a copy of the old site’s robots.txt and XML sitemaps in case you need these later.
Search Console data
Also consider exporting as much of the old site’s Search Console data as possible. These are only available for 90 days, and chances are that once the new site goes live the old site’s Search Console data will disappear sooner or later. Data worth exporting includes:
Search analytics queries & pages
Mobile usability issues
Structured data errors
Links to your site
The redirects implementation is one of the most crucial activities during a site migration. If the legacy site’s URLs cease to exist and aren’t correctly redirected, the website’s rankings and visibility will simply tank.
Why are redirects important in site migrations?
Redirects are extremely important because they help both search engines and users find pages that may no longer exist, have been renamed, or moved to another location. From an SEO point of view, redirects help search engines discover and index a site’s new URLs quicker but also understand how the old site’s pages are associated with the new site’s pages. This association will allow for ranking signals to pass from the old pages to the new ones, so rankings are retained without being negatively affected.
What happens when redirects aren’t correctly implemented?
When redirects are poorly implemented, the consequences can be catastrophic. Users will either land on Not Found pages (404s) or irrelevant pages that do not meet the user intent. In either case, the site’s bounce and conversion rates will be negatively affected. The consequences for search engines can be equally catastrophic: they’ll be unable to associate the old site’s pages with those on the new site if the URLs aren’t identical. Ranking signals won’t be passed over from the old to the new site, which will result in ranking drops and organic search visibility loss. In addition, it will take search engines longer to discover and index the new site’s pages.
When the URLs between the old and new version of the site are different, use 301 (permanent) redirects. These will tell search engines to index the new URLs as well as forward any ranking signals from the old URLs to the new ones. Therefore, you must use 301 redirects if your site moves to/from another domain/subdomain, if you switch from HTTP to HTTPS, or if the site or parts of it have been restructured. Despite some of Google’s claims that 302 redirects pass PageRank, indexing the new URLs would be slower and ranking signals could take much longer to be passed on from the old to the new page.
302 (temporary) redirects should only be used in situations where a redirect does not need to live permanently and therefore indexing the new URL isn’t a priority. With 302 redirects, search engines will initially be reluctant to index the content of the redirect destination URL and pass any ranking signals to it. However, if the temporary redirects remain for a long period of time without being removed or updated, they could end up behaving similarly to permanent (301) redirects. Use 302 redirects when a redirect is likely to require updating or removal in the near future, as well as for any country-, language-, or device-specific redirects.
If you’d like to find out more about how Google deals with the different types of redirects, please refer to John Mueller’s post.
Redirect mapping process
If you are lucky enough to work on a migration that doesn’t involve URL changes, you could skip this section. Otherwise, read on to find out why any legacy pages that won’t be available on the same URL after the migration should be redirected.
The redirect mapping file is a spreadsheet that includes the following two columns:
Legacy site URL –> a page’s URL on the old site.
New site URL –> a page’s URL on the new site.
When mapping (redirecting) a page from the old to the new site, always try mapping it to the most relevant corresponding page. In cases where a relevant page doesn’t exist, avoid redirecting the page to the homepage. First and foremost, redirecting users to irrelevant pages results in a very poor user experience. Google has stated that redirecting pages “en masse” to irrelevant pages will be treated as soft 404s and because of this won’t be passing any SEO value. If you can’t find an equivalent page on the new site, try mapping it to its parent category page.
Once the mapping is complete, the file will need to be sent to the development team to create the redirects, so that these can be tested before launching the new site. The implementation of redirects is another part in the site migration cycle where things can often go wrong.
Increasing efficiencies during the redirect mapping process
Redirect mapping requires great attention to detail and needs to be carried out by experienced SEOs. The URL mapping on small sites could in theory be done by manually mapping each URL of the legacy site to a URL on the new site. But on large sites that consist of thousands or even hundreds of thousands of pages, manually mapping every single URL is practically impossible and automation needs to be introduced. Relying on certain common attributes between the legacy and new site can be a massive time-saver. Such attributes may include the page titles, H1 headings, or other unique page identifiers such as product codes, SKUs etc. Make sure the attributes you rely on for the redirect mapping are unique and not repeated across several pages; otherwise, you will end up with incorrect mapping.
Pro tip: Make sure the URL structure of the new site is 100% finalized on staging before you start working on the redirect mapping. There’s nothing riskier than mapping URLs that will be updated before the new site goes live. When URLs are updated after the redirect mapping is completed, you may have to deal with undesired situations upon launch, such as broken redirects, redirect chains, and redirect loops. A content-freeze should be placed on the old site well in advance of the migration date, so there is a cut-off point for new content being published on the old site. This will make sure that no pages will be missed from the redirect mapping and guarantee that all pages on the old site get redirected.
Don’t forget the legacy redirects!
You should get hold of the old site’s existing redirects to ensure they’re considered when preparing the redirect mapping for the new site. Unless you do this, it’s likely that the site’s current redirect file will get overwritten by the new one on the launch date. If this happens, all legacy redirects that were previously in place will cease to exist and the site may lose a decent amount of link equity, the extent of which will largely depend on the site’s volume of legacy redirects. For instance, a site that has undergone a few migrations in the past should have a good number of legacy redirects in place that you don’t want getting lost.
Ideally, preserve as many of the legacy redirects as possible, making sure these won’t cause any issues when combined with the new site’s redirects. It’s strongly recommended to eliminate any potential redirect chains at this early stage, which can easily be done by checking whether the same URL appears both as a “Legacy URL” and “New site URL” in the redirect mapping spreadsheet. If this is the case, you will need to update the “New site URL” accordingly.
URL A redirects to URL B (legacy redirect)
URL B redirects to URL C (new redirect)
Which results in the following redirect chain:
URL A –> URL B –> URL C
To eliminate this, amend the existing legacy redirect and create a new one so that:
URL A redirects to URL C (amended legacy redirect)
URL B redirects to URL C (new redirect)
Pro tip: Check your redirect mapping spreadsheet for redirect loops. These occur when the “Legacy URL” is identical to the “new site URL.” Redirect loops need to be removed because they result in infinitely loading pages that are inaccessible to users and search engines. Redirect loops must be eliminated because they are instant traffic, conversion, and ranking killers!
Implement blanket redirect rules to avoid duplicate content
It’s strongly recommended to try working out redirect rules that cover as many URL requests as possible. Implementing redirect rules on a web server is much more efficient than relying on numerous one-to-one redirects. If your redirect mapping document consists of a very large number of redirects that need to be implemented as one-to-one redirect rules, site performance could be negatively affected. In any case, double check with the development team the maximum number of redirects the web server can handle without issues.
In any case, there are some standard redirect rules that should be in place to avoid generating duplicate content issues:
URL case: All URLs containing upper-case characters should be 301 redirected to all lower-case URLs, e.g. https://www.website.com/Page/ should be automatically redirecting to https://www.website.com/page/
Protocol: On a secure website, requests for HTTP URLs should be redirected to the equivalent HTTPS URL, e.g. http://www.website.com/page/ should automatically redirect to https://www.website.com/page/
Trailing slash: For instance, any URLs not containing a trailing slash should redirect to a version with a trailing slash, e.g. http://www.website.com/page should redirect to http://www.website.com/page/
Even if some of these standard redirect rules exist on the legacy website, do not assume they’ll necessarily exist on the new site unless they’re explicitly requested.
Avoid internal redirects
Try updating the site’s internal links so they don’t trigger internal redirects. Even though search engines can follow internal redirects, these are not recommended because they add additional latency to page loading times and could also have a negative impact on search engine crawl time.
Don’t forget your image files
If the site’s images have moved to a new location, Google recommends redirecting the old image URLs to the new image URLs to help Google discover and index the new images quicker. If it’s not easy to redirect all images, aim to redirect at least those image URLs that have accrued backlinks.
Phase 3: Pre-launch testing
The earlier you can start testing, the better. Certain things need to be fully implemented to be tested, but others don’t. For example, user journey issues could be identified from as early as the prototypes or wireframes design. Content-related issues between the old and new site or content inconsistencies (e.g. between the desktop and mobile site) could also be identified at an early stage. But the more technical components should only be tested once fully implemented — things like redirects, canonical tags, or XML sitemaps. The earlier issues get identified, the more likely it is that they’ll be addressed before launching the new site. Identifying certain types of issues at a later stage isn’t cost effective, would require more resources, and cause significant delays. Poor testing and not allowing the time required to thoroughly test all building blocks that can affect SEO and UX performance can have disastrous consequences soon after the new site has gone live.
Making sure search engines cannot access the staging/test site
Before making the new site available on a staging/testing environment, take some precautions that search engines do not index it. There are a few different ways to do this, each with different pros and cons.
Site available to specific IPs (most recommended)
Making the test site available only to specific (whitelisted) IP addresses is a very effective way to prevent search engines from crawling it. Anyone trying to access the test site’s URL won’t be able to see any content unless their IP has been whitelisted. The main advantage is that whitelisted users could easily access and crawl the site without any issues. The only downside is that third-party web-based tools (such as Google’s tools) cannot be used because of the IP restrictions.
Password protecting the staging/test site is another way to keep search engine crawlers away, but this solution has two main downsides. Depending on the implementation, it may not be possible to crawl and test a password-protected website if the crawler application doesn’t make it past the login screen. The other downside: password-protected websites that use forms for authentication can be crawled using third-party applications, but there is a risk of causing severe and unexpected issues. This is because the crawler clicks on every link on a page (when you’re logged in) and could easily end up clicking on links that create or remove pages, install/uninstall plugins, etc.
Adding the following lines of code to the test site’s robots.txt file will prevent search engines from crawling the test site’s pages.
One downside of this method is that even though the content that appears on the test server won’t get indexed, the disallowed URLs may appear on Google’s search results. Another downside is that if the above robots.txt file moves into the live site, it will cause severe de-indexing issues. This is something I’ve encountered numerous times and for this reason I wouldn’t recommend using this method to block search engines.
User journey review
If the site has been redesigned or restructured, chances are that the user journeys will be affected to some extent. Reviewing the user journeys as early as possible and well before the new site launches is difficult due to the lack of user data. However, an experienced UX professional will be able to flag any concerns that could have a negative impact on the site’s conversion rate. Because A/B testing at this stage is hardly ever possible, it might be worth carrying out some user testing and try to get some feedback from real users. Unfortunately, user experience issues can be some of the harder ones to address because they may require sitewide changes that take a lot of time and effort.
On full site overhauls, not all UX decisions can always be backed up by data and many decisions will have to be based on best practice, past experience, and “gut feeling,” hence getting UX/CRO experts involved as early as possible could pay dividends later.
Site architecture review
A site migration is often a great opportunity to improve the site architecture. In other words, you have a great chance to reorganize your keyword targeted content and maximize its search traffic potential. Carrying out extensive keyword research will help identify the best possible category and subcategory pages so that users and search engines can get to any page on the site within a few clicks — the fewer the better, so you don’t end up with a very deep taxonomy.
Identifying new keywords with decent traffic potential and mapping them into new landing pages can make a big difference to the site’s organic traffic levels. On the other hand, enhancing the site architecture needs to be done thoughtfully. Itt could cause problems if, say, important pages move deeper into the new site architecture or there are too many similar pages optimized for the same keywords. Some of the most successful site migrations are the ones that allocate significant resources to enhance the site architecture.
Meta data & copy review
Make sure that the site’s page titles, meta descriptions, headings, and copy have been transferred from the old to the new site without issues. If you’ve created any new pages, make sure these are optimized and don’t target keywords that have already been targeted by other pages. If you’re re-platforming, be aware that the new platform may have different default values when new pages are being created. Launching the new site without properly optimized page titles or any kind of missing copy will have an immediate negative impact on your site’s rankings and traffic. Do not forget to review whether any user-generated content (i.e. user reviews, comments) has also been uploaded.
Internal linking review
Internal links are the backbone of a website. No matter how well optimized and structured the site’s copy is, it won’t be sufficient to succeed unless it’s supported by a flawless internal linking scheme. Internal links must be reviewed throughout the entire site, including links found in:
Main & secondary navigation
Header & footer links
Body content links
Horizontal links (related articles, similar products, etc)
Vertical links (e.g. breadcrumb navigation)
Cross-site links (e.g. links across international sites)
A series of technical checks must be carried out to make sure the new site’s technical setup is sound and to avoid coming across major technical glitches after the new site has gone live.
Robots.txt file review
Prepare the new site’s robots.txt file on the staging environment. This way you can test it for errors or omissions and avoid experiencing search engine crawl issues when the new site goes live. A classic mistake in site migrations is when the robots.txt file prevents search engine access using the following directive:
If this gets accidentally carried over into the live site (and it often does), it will prevent search engines from crawling the site. And when search engines cannot crawl an indexed page, the keywords associated with the page will get demoted in the search results and eventually the page will get de-indexed.
But if the robots.txt file on staging is populated with the new site’s robots.txt directives, this mishap could be avoided.
When preparing the new site’s robots.txt file, make sure that:
It doesn’t block search engine access to pages that are intended to get indexed.
The legacy site’s robots.txt file content has been reviewed and carried over if necessary.
It references the new XML sitemaps(s) rather than any legacy ones that no longer exist.
Canonical tags review
Review the site’s canonical tags. Look for pages that either do not have a canonical tag or have a canonical tag that is pointing to another URL and question whether this is intended. Don’t forget to crawl the canonical tags to find out whether they return a 200 server response. If they don’t you will need to update them to eliminate any 3xx, 4xx, or 5xx server responses. You should also look for pages that have a canonical tag pointing to another URL combined with a noindex directive, because these two are conflicting signals and you;’ll need to eliminate one of them.
Meta robots review
Once you’ve crawled the staging site, look for pages with the meta robots properties set to “noindex” or “nofollow.” If this is the case, review each one of them to make sure this is intentional and remove the “noindex” or “nofollow” directive if it isn’t.
XML sitemaps review
Prepare two different types of sitemaps: one that contains all the new site’s indexable pages, and another that includes all the old site’s indexable pages. The former will help make Google aware of the new site’s indexable URLs. The latter will help Google become aware of the redirects that are in place and the fact that some of the indexed URLs have moved to new locations, so that it can discover them and update search results quicker.
You should check each XML sitemap to make sure that:
It validates without issues
It is encoded as UTF-8
It does not contain more than 50,000 rows
Its size does not exceed 50MBs when uncompressed
If there are more than 50K rows or the file size exceeds 50MB, you must break the sitemap down into smaller ones. This prevents the server from becoming overloaded if Google requests the sitemap too frequently.
In addition, you must crawl each XML sitemap to make sure it only includes indexable URLs. Any non-indexable URLs should be excluded from the XML sitemaps, such as:
3xx, 4xx, and 5xx pages (e.g. redirected, not found pages, bad requests, etc)
Soft 404s. These are pages with no content that return a 200 server response, instead of a 404.
Canonicalized pages (apart from self-referring canonical URLs)
Pages with a meta robots noindex directive
Pages with a noindex X-Robots-Tag in the HTTP header
HTTP/1.1 200 OK
Date: Tue, 10 Nov 2017 17:12:43 GMT
Pages blocked from the robots.txt file
Building clean XML sitemaps can help monitor the true indexing levels of the new site once it goes live. If you don’t, it will be very difficult to spot any indexing issues.
Pro tip: Download and open each XML sitemap in Excel to get a detailed overview of any additional attributes, such as hreflang or image attributes.
HTML sitemap review
Depending on the size and type of site that is being migrated, having an HTML sitemap can in certain cases be beneficial. An HTML sitemap that consists of URLs that aren’t linked from the site’s main navigation can significantly boost page discovery and indexing. However, avoid generating an HTML sitemap that includes too many URLs. If you do need to include thousands of URLs, consider building a segmented HTML sitemap.
The number of nested sitemaps as well as the maximum number of URLs you should include in each sitemap depends on the site’s authority. The more authoritative a website, the higher the number of nested sitemaps and URLs it could get away with.
For example, the NYTimes.com HTML sitemap consists of three levels, where each one includes over 1,000 URLs per sitemap. These nested HTML sitemaps aid search engine crawlers in discovering articles published since 1851 that otherwise would be difficult to discover and index, as not all of them would have been internally linked.
Structured data review
Errors in the structured data markup need to be identified early so there’s time to fix them before the new site goes live. Ideally, you should test every single page template (rather than every single page) using Google’s Structured Data Testing tool.
Be sure to check the markup on both the desktop and mobile pages, especially if the mobile website isn’t responsive.
The tool will only report any existing errors but not omissions. For example, if your product page template does not include the Product structured data schema, the tool won’t report any errors. So, in addition to checking for errors you should also make sure that each page template includes the appropriate structured data markup for its content type.
Again, the Fetch and Render tool can help diagnose this type of issue that, if left unresolved, could have a significant negative impact.
Mobile site SEO reviewAssets blocking review
Mobile-first index review
In order to avoid any issues associated with Google’s mobile-first index, thoroughly review the mobile website and make there aren’t any inconsistencies between the desktop and mobile sites in the following areas:
Meta robots attributes (i.e. noindex, nofollow)
A responsive website should serve the same content, links, and markup across devices, and the above SEO attributes should be identical across the desktop and mobile websites.
In addition to the above, you must carry out a few further technical checks depending on the mobile site’s set up.
Responsive site review
A responsive website must serve all devices the same HTML code, which is adjusted (via the use of CSS) depending on the screen size.
To signal browsers that a page is responsive, a meta=”viewport” tag should be in place within theof each HTML page.
If the meta viewport tag is missing, font sizes may appear in an inconsistent manner, which may cause Google to treat the page as not mobile-friendly.
Separate mobile URLs review
If the mobile website uses separate URLs from desktop, make sure that:
Each desktop page has a tag pointing to the corresponding mobile URL.
Each mobile page has a rel=”canonical” tag pointing to the corresponding desktop URL.
When desktop URLs are requested on mobile devices, they’re redirected to the respective mobile URL.
Redirects work across all mobile devices, including Android, iPhone, and Windows phones.
There aren’t any irrelevant cross-links between the desktop and mobile pages. This means that internal links on found on a desktop page should only link to desktop pages and those found on a mobile page should only link to other mobile pages.
The mobile URLs return a 200 server response.
Dynamic serving review
Dynamic serving websites serve different code to each device, but on the same URL.
On dynamic serving websites, review whether the vary HTTP header has been correctly set up. This is necessary because dynamic serving websites alter the HTML for mobile user agents and the vary HTTP header helps Googlebot discover the mobile content.
Regardless of the mobile site set-up (responsive, separate URLs or dynamic serving), review the pages using a mobile user-agent and make sure that:
The viewport has been set correctly. Using a fixed width viewport across devices will cause mobile usability issues.
The font size isn’t too small.
Touch elements (i.e. buttons, links) aren’t too close.
There aren’t any intrusive interstitials, such as Ads, mailing list sign-up forms, App Download pop-ups etc. To avoid any issues, you should use either use a small HTML or image banner.
Mobile pages aren’t too slow to load (see next section).
Google’s mobile-friendly test tool can help diagnose most of the above issues:
AMP site review
If there is an AMP website and a desktop version of the site is available, make sure that:
Each non-AMP page (i.e. desktop, mobile) has a tag pointing to the corresponding AMP URL.
Each AMP page has a rel=”canonical” tag pointing to the corresponding desktop page.
Any AMP page that does not have a corresponding desktop URL has a self-referring canonical tag.
You should also make sure that the AMPs are valid. This can be tested using Google’s AMP Test Tool.
Mixed content errors
Mixed content occurs when a page that’s loaded over a secure HTTPS connection requests assets over insecure HTTP connections. Most browsers either block dangerous HTTP requests or just display warnings that hinder the user experience.
There are many ways to identify mixed content errors, including the use of crawler applications, Google’s Lighthouse, etc.
Image assets review
Google crawls images less frequently than HTML pages. If migrating a site’s images from one location to another (e.g. from your domain to a CDN), there are ways to aid Google in discovering the migrated images quicker. Building an image XML sitemap will help, but you also need to make sure that Googlebot can reach the site’s images when crawling the site. The tricky part with image indexing is that both the web page where an image appears on as well as the image file itself have to get indexed.
Site performance review
Read more about this at moz.com.