Dear Warhammer, let’s talk about SEO, JavaScript, and website migrations .

Disclaimer

Before you start reading this post, please remember that all the analyses I present here are done from an external point of view.

I have no insights from or access to internal tools and documentation.

This article must not be considered as a j’accuse directed to any person or team working at Games Workshop, but simply a record of what an SEO like me can see and infer visiting the Warhammer.com website and investigating its new design and the migration itself, and use these analyses as a case study aimed to advise other SEOs, especially new ones, about errors and issues that must be considered when dealing with a re-platforming and a website migration.

I am 100% confident that Games Workshop is aware of all the problems Warhammer.com has, and I am equally sure they are already working on resolving them.

An Introduction to Games Workshop and Warhammer

Around November 1st, 2023, Games Workshop rolled out an important migration.

After almost 20 years, the Games Workshop domain name (games-workshop.com) has been migrated to Warhammer.com. Until then, the redirection was the other way around (warhammer.com >> 301 >> games-workshop.com).

The reasons for this migration are many, being what seems to be a reorganization of the company one of the causes (for instance, forgeworld.co.uk, a website dedicated to selling models for professional mini painters and pro-gamers, had been migrated to warhammer.com almost at the same time).

Another reason can be pointing firmly to the brand “Warhammer”, which is the most widely known of Games Workshop thanks to its tabletop games “Warhammer 40000” and “Warhammer: Age of Sigmar”, even if on Warhammer.com you can also buy other games and minis as the Middle Earth Strategy Battle Game.

On the other hand, the centrality of the brand “Warhammer” is nothing new, because during the years Games Workshop developed many websites using it:

We also have microsites dedicated to the many Warhammer games like, for instance, Warhammer 40000, Warhammer: The Old War or Warhammer: The Horus Heresy.

The only three websites of Games Workshop that are not branded “Warhammer” are:

Citadelcolour.com, being Citadel the brand for the painting and painting accessories of GW.
Blacklibrary.com, which is the website of the editorial company by GW that publishes, sells, and distributes fiction books and audiobooks based on the Warhammer lore.
Golden-demon.com, the (apparently abandoned) WordPress website about the painting competitions Games-Workshop holds every year in the UK and the US.

To these three, finally, we should add the microsite dedicated to the Middle-Earth strategy battle game.

As we can see, Games Workshop has developed an extensive online presence, and it is easy to understand how difficult can be to orchestrate them all during a migration like the one from Games-workshop.com to Warhammer.com.

Before & after the migration Search Visibility and estimated traffic: a look.

Here above we see the estimated traffic per channel presented by Semrush in its Traffic Analytics section.

As we can see from the screenshot, Games-workshop.com still obtains about 435.000 visits from Organic Search after about 2 months from the migration. This means that its URLs are visible in the SERPs. We must suppose that most of the still visible GW website search results redirect searchers to Warhammer.com.

In October 2023, its last month being online, Organic Search generated an estimated number of 9.300.000 visits to Games-workshop.com, being its second source of traffic after Direct (about 13.700.000 visits) and by far before Referral (914.000).

In an ideal world, the traffic before going to Games-workshop.com now should be going to Warhammer.com.

Let’s see what the estimation of Semrush tells us:

The numbers we see here above are referred to December 2023.

As we can see, organic traffic seems to have generated about 7.100.000 visits to Warhammer.com, which is 2.200.000 visits less than the organic traffic obtained by Games-workshop.com in October (and considering that December means Christmas), which is quite a big difference even if we consider that some subdomains are still alive under the Games-workshop.com domain name (i.e.: https://jobs.games-workshop.com/ and https://trade.games-workshop.com/).

We could say that these are just estimations, and I am the first to think that any traffic data not based on first-party sources like Search Console and Google Analytics must be taken with a grain of salt, even if they can be useful for having a macro view of the traffic trends for a website.

So, let’s use another type of metric: the Search Visibility by SISTRIX.

In this case, I will only consider the subdomains www. for Games-workshop.com and Warhammer.com:

Here above we see the Search Visibility evolution of Warhammer.com in Google.co.uk according to SISTRIX.

We can see when the migration of the website started.

It seems a satisfactory evolution, doesn’t it?

Now let’s compare the Warhammer.com Search Visibility with the one of Games-workshop.com for the same period:

The sentiment about the domain migration changes from “positive” to “worried” looking at the graph here above.

It clearly shows us that Warhammer.com has not been able yet to recover the visibility in the SERPs that Games-workshop.com had.

On the other hand, though, we also see that the subdomain www.games-workshop.com has practically disappeared.

However, a negative gap of almost two points in search visibility, and a growth that seems quite flat right now are signals that maybe the migration did not give the desired results because, I think we all agree here, in SEO a migration should represent the opportunity to improve the organic visibility of a website.

A question we could ask ourselves is: “Is the difference caused by the visibility owned by the Games Workshop subdomains that are still alive?”. The answer is: “It does not seem so”:

This is in Google.co.uk, and the British version is the most powerful of all the versions of Warhammer.com, but the website is multi-country.

So, let’s see what happened to the Italian version in Google.it:

The Italian version of www.games-workshop.com did not have great search visibility (this is important to remember for later), but the migration to www.warhammer.com has killed that small presence in the SERPs it had before.

Only slightly better is the case for the Spanish (Spain) version:

And this behaviour is similar to all the other versions of Warhammer.com.

What may have happened?

With an external analysis, we can find several reasons that can explain why the migration did not go as well as it was probably thought.

The Dark Lord Sauron: JavaScript

The first thing we must consider is that Warhammer.com is developed with Next.js and the Contentful Headless CMS.

As SEOs, we know that Google treats differently websites based on JavaScript frameworks than the others.

JavaScript is very resource-demanding for a search engine and for this reason, Google follows this process when it comes to indexing content generated via JavaScript:

This means that Google quite probably discovered all the new Warhammer.com URLs thanks to the redirect mapping, but the indexing of the rendered content of these pages started later.

How much time a URL can wait in the Render Queue?

The official documentation of Google says that a page “may stay on this queue for a few seconds, but it can take longer than that. Once Google’s resources allow, a headless Chromium renders the page and executes the JavaScript. Googlebot parses the rendered HTML for links again and queues the URLs it finds for crawling. Google also uses the rendered HTML to index the page.”

However, there are studies such as, for instance, Rendering Queue: Google Needs 9X More Time To Crawl JS Than HTML by Onely that tells us that the time spent in the Render Queue is longer depending on the depth of the URL in the architecture of the website:

These times do surely not favour a fast migration.

Moreover, we must also consider another question: how does Google prioritize the URLs in the Render Queue?

It is probable the existence of some sort of algorithm that says something like this: “The pages that can be deemed more relevant because of a series of factors, will be put higher in the queue”.

This means that pages that are considered less relevant because of freshness, external factors (aka backlinks), and all the other possible reasons why they were crawled less frequently than others already before a possible migration/change, these pages, then, are put in the back of the queue. Finally, we should also consider that every second new URLs are added in the Render Queue so that a page already back in the line can see its wait even increased.

The fact that the content of some pages is rendered only after a (long) delay from their publication (and with content, I also include internal links that are injected by JavaScript) may explain why the migration from Games-workshop.com to Warhammer.com takes/took so long, and why weaker versions like the Italian one are practically invisible after the migration.

For this reason, it is a classic mantra recommending choosing server-side rendering instead of client-side rendering, to present Google the rendered HTML page from the beginning as a way to skip the Render Queue.

And this is the big problem of Warhammer.com, as we are going to see.

Blood for the Blood God: the JS-based navigation menu

When a website has been developed using JavaScript, the first thing I check is the main menu.

Why? Because the main menu is sitewide, therefore, considering that Google can start crawling a website from any page it has in its index, it is always accessible for a crawler to start following other pages according to a defined architecture.

What is the menu of Warhammer.com?

As you can see, we can define it as minimalistic:

“Start here” links to a sort of shorter secondary home page.
If we click on “Shop”, instead, we have the real menu.

We can navigate it quite easily and finally end in the category or landing page that interests us the most, for instance, the one presenting the forces of Good for Middle Earth: Lord of the Rings strategy game.

If we inspect the page, we see that the links are good old “href=”, and respectful of the best practices Google offers us about JavaScript and links:

The same seems to be for the menu:

But… if we close the menu, the links in it disappear and what we see is a span class:

So… what is the problem?

The problem is that Google can see the links in the menu only when expanding the wrapper including them because they are a classic href:

<a data-testid="navigation-slug-1-0" href="/en-GB/shop/middle-earth/the-lord-of-the-rings-good" class="normal-case text-black inline-flex no-underline tracking-[.05em] leading-7 undefined text-base font-normal md:text-base">The Lord of the Rings™ - Good</a>

However, it cannot reach the links because all the clicks for opening the menu and the button before finally finding links in the menu navigation are dynamic, and Google doesn’t click or expand them. In other words, Google sees something (“Shop”) but does not navigate it.

Beast of Chaos: the version selector

Something similar can be discovered regarding the Change Location menu of Warhammer.com.

The country/language selector is an easy solution for helping Google discover and crawl the URLs of other versions, always if, for instance, the URL A of /en-GB/ links to the corresponding URL A of /it-IT/, /es-ES/, /fr-FR/ et al.

The first problem we can discover is that the version selector always sends us to the homepage of the version we want to visit, and not to the corresponding page we are visiting (this can also be considered an important problem in terms of UX).

However, the biggest issue is that no link is present in the Change Location menu. Every single version is presented via JavaScript without any link.

Sure, Google may discover the many versions of Warhammer.com through other means like:

Hreflang, which is present in the HTML (i.e.: <link rel=”alternate” hrefLang=”it-IT” href=”https://www.warhammer.com/it-IT/home”/>.
XML Sitemaps.
External links, at least those that correspond to URLs, which have been included in the 301 redirects mapping.

The problem is that apart from the backlinks, neither the URLs in the hreflangs nor the XML Sitemaps pass link equity, which means that one of the reasons for creating an international SEO architecture on subfolders is nuked by the Change Location menu, and the stronger versions will not pass any link equity to the weaker ones.

This issue, which must be added to the uncrawlable navigation menu and the client-side rendering, can partly explain why a version like the Italian is struggling so evidently to gain visibility or why the Polish version (in English) ranks very very poorly with /en-PL/ subfolder URLs in Google.pl.

However, the list of SEO problems of Warhammer.com is not finished yet.

Mordor Battlehost: the “weird” XML Sitemaps strategy, hreflangs and external links.

The Sitemaps

As we have seen, the XML Sitemaps are extremely important for a website like Warhammer.com, because they can supply the non-existent internal menu navigation due to the use of JavaScript. So, let’s see how they are.

We can find the Sitemaps.xml index here: https://www.warhammer.com/sitemap.xml.

Being an index, the URLs listed are of other XML Sitemaps, and we can notice as a pattern the presence of these two types:

Shop sitemaps, which list the URLs of the PLPs (some of them respond 404 as this one).
Landing pages sitemaps, which list a set of different types of pages:
- Lists (i.e.: https://www.warhammer.com/es-ES/plp?series=Siege%20of%20Terra)
- Internal Search Results pages (i.e.: https://www.warhammer.com/es-ES/plp?search=BT33), which formally is not something that should be done, if we want to follow the guidelines of Google, and wrong if we consider that the robots.txt file of Warhammer.com blocks bots from crawling these pages as we can see here below:

- Filters pages, which URLs cannot be reached as such via internal navigation (i.e.: the https://www.warhammer.com/es-ES/plp?paintType=Technical).

Maybe it would have been better to have, at least, the breadcrumb implemented in the shop and product pages to allow the crawlers to discover, parse and index the pages of the website and understand the taxonomy of the shop.

Another important thing we can notice is how the Italian version has only the landing page XML Sitemap listed, and not the PLPs’ one; one more reason for it to be practically invisible in Google.it.

The hreflang implementation

Before we saw that another possible way Google has for discovering URLs of the several international versions of Warhammer.com are the hreflangs.

In general, we must say that the hreflang implementation is correct most of the time.

However, if you do a crawl (as I did), we can find several errors as:

2700 URLs present at least one or two no-return errors, which means that the alternative URL B indicated in URL A, for instance, doesn’t refer back to A as its alternate. In these cases, Google will not consider the incorrect hreflang annotation, which means that – quite probably – it won’t either consider the URL for discovering and this – from what we have seen before – means losing one of the few opportunities Warhammer.com has for having the URLs being crawled, queued and its content indexed.
14800 URLs do not have any hreflang. This is not always a mistake because a URL may exist only in one version. However, among the URLs without hreflangs, we can find landing pages like https://www.warhammer.com/en-GB/shop/other-games/necromunda, which has alternative versions for other countries (i.e.: https://www.warhammer.com/en-CA/shop/other-games/necromunda or https://www.warhammer.com/en-NZ/shop/other-games/necromunda). Apart from the misalignments in the local SERPs, this mistake represents, this is another missed opportunity to help Google discover faster the URLs of the website.
322 URLs have hreflangs pointing to 404 pages, which is another reason for Google to ignore the annotations.

Finally, there is another potential “error” that could cause the hreflang annotation to be considered invalid: the sitewide absence of the rel=”canonical” (quite strange, considering that Contentful has native SEO capabilities, including defining the canonical tag).

One of the rules of the hreflang is that it must always indicate a canonical URL as an alternative.

This means that even if it is not an obligation to have the rel=”canonical” defined, it is highly recommended to do so because, on the contrary, we are giving Google complete freedom to choose by itself what is the canonical version of a page.

Considering that Warhammer has 17 identical English versions, the risk of Google canonicalizing, for instance, the English for Poland version to one of the other English versions may mean that the hreflangs of /en-PL/ could not indicate the canonical URL as perceived by Google, hence they would be flagged as wrong and not considered.

Therefore, even if the hreflang annotations can be a way Googlebot may use to discover URLs, something crucial for how Warhammer.com has been developed, they alone cannot be considered a reliable alternative for that purpose if all the variables related to a correct hreflang implementation are not correct.

External links.

External links can be a way for Google to discover a redirect, hence a useful means for discovering the URLs of the new website faster, especially if you did not use a “trick” that Googlers themselves suggest: having the XML Sitemaps of the old website alive, so for Google to crawl them regularly, hence regularly see the 301s we mapped.

In the case of Games-workshop.com, however, we must consider its more than 30 years of existence and the giant fanbase of players, painters and collectors that have talked about it on the web in these two decades.

Many backlinks had been created to pages that had been deleted and hadn’t been correctly redirected.

In some cases, for more recently deleted product pages, we can see how rules have been created.

For instance, the URL https://www.games-workshop.com/GW-Warhammer-40k-Armies-Aeldari-GW-RL is redirected 301 to its main landing page (even if the category page of the Aeldari Armies would be a better choice).

In other cases, though, no rule has been created, which means that many natural backlinks are lost as, for instance, these pointing to older PDF pages:

Trying to recover older broken backlinks with:

Creating a process for discovering them.
Analyzing patterns in the older URL linked.
Setting up rules that will quit us the tedious and time-consuming need to singularly map the 301s

is something that can be very important especially for the weaker country versions of Warhammer.com, considering how they are not receiving any link equity from the stronger ones because of the uncrawlable Location Selector.

Finally, what surely the people at GW should do is update the backlinks to www.games–workshop.com pages present in the many websites they own.

Missed opportunities

A migration usually should also be the occasion for improving the search visibility, optimizing the content and, for multi-country websites like Warhammer.com, improving the localization in other languages and local dialects in case of country versions sharing the same language.

Unfortunately, it seems this was not the case for Warhammer.com and we can see from these examples.

Title tags and on-page

I know I know, title tags sound very old SEO in the era of AI.

However, they still play a very important role, plus they serve the purpose of creating compelling search results, which can improve Click Through Rates, something that – as we have seen with the revelation of NavBoost – is one of the user engagement metrics that impact how a web page is valued by Google.

1600 indexable URLs, among them quite many listing pages as https://www.warhammer.com/en-SE/shop/age-of-sigmar/grand-alliance-order/daughters-of-khaine, do not have any Title tag.

In general, though, classic on-page SEO is very poorly handled; look at the /en-GB/ homepage:

Google, though, can eventually use the og:title and og: description for creating the search result snippet.

It would not have that much success either:

<meta property="og:title" content="Home - Warhammer"/><meta property="og:description" content="This is the home page"/><meta property="og:url" content="https://www.warhammer.com/en-GB/home"/>

Structured Data

Before we saw how Warhammer.com has no breadcrumb implemented, and how its absence contributes to the difficulty Google has in navigating the website and understanding its architecture.

However, the absence of the breadcrumb is just a little thing concerning the real issue: no structured data is implemented on Warhammer.com.

At least this is what seems to be looking at the crawl data.

I say “at least”, because if we search for “Rohan Battlehost”, we see this:

Then, if we look at the image search results, we see the classic “product” label, which is given to pages that have the structured data for products implemented:

Somehow this is a mystery, at least for me, also because it is impossible to do any test with the Rich Results testing tool of Google, probably because of the Queue-it implementation Warhammer.com has:

Moreover, I checked the JSON in the code of the page (view-source:https://www.warhammer.com/en-GB/shop/rohan-battlehost-2022), and I was not able to find any reference to Schema.org or to any distinctive object fields that Contentful use for adding structured data to a page (see https://www.contentful.com/seo-guide/schema-seo/ and https://www.contentful.com/blog/how-we-increased-our-serp-and-organic-traffic/).

The only alternative possible is considering that structured data is injected using Google Tag Manager, but also of this, there is no trace in the code.

If you are auditing a website, this is the classic situation that needs to be clarified with the developers and the in-house SEO, so for you – the consultant – to understand better the implementation and see if there are areas of improvement or things that can be done better.

Nevertheless, the rich result we saw in the example screenshot above (especially if compared with the Amazon one below) suggests that there is space for improvement in the Product markup of the Warhammer.com PLPs.

For instance, as we can see below, the product page presents a few positive highlights to consider by the potential buyer of a product:

If we talk of Product structured data, then we must talk also of Google Merchant (and Manufacturer), and it is not clear if Games Workshop is making use of it. From the many SERPs related to their products, I never saw a Merchant snippet by them.

Should the improvements of structured data be included in the roadmap of a website re-platforming and migration, or we could wait until after the migration, considering its complexity?

For an ecommerce website like Warhammer.com, I think that we should add structured data optimization to our migration to-do tasks, including, of course, testing them, if there are no compelling reasons to not do it.

International SEO. Structure and localization

Warhammer.com inherited the same international architecture of Games-workshop.com.

Maybe (once again if nothing else was suggesting the contrary) the migration could have been the opportunity to reconsider it and implement a better international SEO structure in Warhammer.com.

We saw before how Warhammer presents a long list of English versions also targeting countries where English is not the official language (i.e.: /en-PL/ or /en-BE), and this may mean that the risks of versions’ misalignments can be very high, as we can see here below for the visibility of Warhammer.com in Google.be:

As we can see, the version that is ranking is not the /en-BE/ but the x-default one.

Why?

Because the x-default is meant to be shown to the users, which language is not targeted by any language and/or language-country version.
In Belgium, people speak French or Dutch, therefore also their Google Search is set up – normally – to French or Dutch.
Google, hence, presents the x-default URLs to them in Google.be.
Therefore, the /en-BE/ version is substantially invisible, and Belgians landing on Warhammer.com from Search are obliged to pass through the passage of choosing the English (!) for Belgium version in the Location Selector menu, which undoubtedly is a friction.

In other words, only if the Google Search settings are set up to English, Belgians will eventually see the /en-BE/ version, and this is why its visibility in Google.be is 0:

If a generic “global French” version existed, then Warhammer.com could have served it to people speaking French in Belgium (and other French-speaking countries).

These are the types of reasoning and considerations, which should have been made when the change of domain was decided together with the change of technology stack.

Are these changes too much to add to all the tasks that are involved in a domain migration? Maybe it can be an ambitious goal, but if it is well-defined and the steps finely explained and presented, not impossible.

Surely, what Games Workshop should have done is localise in Italian, Spanish, French and German their product pages, whose main content is in English.

This happens at least in two cases:

When the product exists only in English, as is the case of the Codex books (i.e.: https://www.warhammer.com/es-ES/shop/codex-space-marines-hb-2023-eng and https://www.warhammer.com/it-IT/shop/codex-space-marines-hb-2023-eng), which is a quite strange decision because even if books about factions are in English, that does not mean that we cannot show also its PLP in the local languages.
When the product existed before on the forgeworld.co.uk website, which too was redirected to Warhammer.com (i.e.: https://www.warhammer.com/fr-FR/shop/iron-hills-dwarf-warrior-warband-with-mattocks-2023 and the German, Spanish and Italian versions). The Forge World minis are thought for more expert players and painters, and to maintain this distinction and targeting now they are presented as “Expert Kit” and for users 15+ (normally the Games Workshop games and minis are rated 12+).

What can we learn from the Warhammer.com migration and replatforming?

We can surely learn that one does not simply walk into Mordor.

More seriously, we can learn that re-platforming and domain migrations are possibly two of the most complex events that an SEO can face in his career. And if they feature a multi-country site, then an SEO is walking towards Mount Doom.

These are events that require planning, strict organization, and technical knowledge but, above all, the ability of an SEO to put his recommendations at the centre of the decisions or at least at the same level as those of the developers and to know how to convince stakeholders of their validity.

If all this is absent, then “disasters” will have more chances to happen.

Ah! Of course… we also learned that client-side rendering is Evil.

Useful resources based on this analysis.

Here you can find articles, decks and guides that complement the links to resources I presented in the analysis of Warhammer.com, and they are a good complement to them.

Websites’ Migration:

A domain migration checklist and ultimate Google Data Studio dashboard by Kristina Azarenko.
Winning SEO when doing web migrations by Aleyda Solís.
The website migration guide: SEO strategy, process & checklist by Modestos Siotos on Moz.com.
Website redesign SEO checklist by Olga Zarr.

JavaScript SEO:

The ultimate guide to JavaScript SEO by Onely.
Rendering on the web by Addy Osmani and Jason Miller on Google’s Web.dev.
JavaScript and Structured Data (Google guide).
How JavaScript rendering affects Google indexing by Patrick Hathaway.

Headless SEO:

Headless & SEO by Contentful (and quite specific to Contentful).
Headless SEO 101: everything you need to get started by Lidia Infante.

SEO relation with organizations, and stakeholders:

How to convince execs to care about SEO by Tom Critchlow.
How to communicate the value of SEO to executives on Seer Interactive.

Thank you Games Workshop for all the hours of mini-painting.

Share if you care

Dear Warhammer, let’s talk about SEO, JavaScript, and website migrations