The SEO Migrations Case Study covers an American multinational (listed on NASDAQ) we joined to support the European market (17 countries) in an operational-model restructuring and a push to revitalize sales through the digital channel.
The situation we faced in September 2018 was a Drupal-based framework for all countries, which each country had adapted to their acquisition and conversion needs. All the Tag Manager configurations, Google Analytics goals, CTAs and other elements had been put together by each country without setting common guidelines to serve a global strategy. For example, the main goal existed in some markets and not in others. Or that main goal was in Estonian in one Google account and in Polish in another, among other inconsistencies.
Mass migration and resource centralization
As always, when we start a project, we do two things:
We joined a month before a mass migration from version 2.0 to 3.0, and what we detected was that there were no migration plans (neither for SEO nor web analytics), plus redirect problems, 404 pages, etc. that had been carried over from the 1.0-to-2.0 migration.
We'd like to highlight the bad practice of some agencies when it comes to analytics during a migration, creating a new Analytics property and not trying to preserve historical data. Please, stop the legacy approach and put more effort into preserving the logic of historical data.
So, after several alignment meetings, we kicked off the full study and migration plan:
On the SEO front
One of the client's main goals in this migration was to homogenize information across its European branches: at the level of data capture and analysis; in internal CMS and CRM management; and in corporate image, including how content was displayed and organized.
Search intent
While on the analytics and management side we saw their homogenization goal as essential for the project as a whole, we didn't see it as feasible to apply the same architecture and content across all sites/countries, even if each one was correctly translated into its corresponding language. The reason is that today it's not only crucial to properly answer user search intent to rank well, you also have to do it better than your competitors. And it's well known that there are cultural differences between European countries that influence the behaviour of their inhabitants. For example, a Swede's consumption habits aren't the same as a Portuguese person's, nor is the way they search for information.
We could confirm this empirically when we ran a thorough keyword research for each market, alongside a search-intent research. That's why, from the beginning, we advised against applying content homogenization because we foresaw it wasn't going to deliver good SEO results. Despite our recommendations, the decision was made to push ahead with the initial plan due to the urgency of launching version 3.0.
When you work with a client, you don't always have decision-making power, and there are factors or decisions involving parties with different weightings. That's why we always have to adapt to situations as flexibly, quickly and productively as possible. Based on that, we adapted the content to each market's search intent once the entire migration process had been completed, even though ideally we would have launched the adapted content from the start. Still, with all the data we had from our keyword researches, we made sure everything was optimized to the maximum as far as possible in terms of metas, headings, content, alts, etc.
On page
When it comes to migrations we've managed to be meticulous thanks to our action protocols (migrations, content, audits, link building, etc.).
In this project, since we were constrained on content, it was essential for the on-page aspects to be perfect:
On the CRO front
We worked along 3 lines:
1. Clean up, standardize and protocolize the information needed in
Google Tag Manager:
At this point we had to understand what should be kept in the new Tag
Manager containers, while we added the cross-domain integration; reused variables for different
tags and triggers; and did so through documentation to standardize naming conventions across
markets:
We also implemented all the third-party software tags that over the following months we'd use to improve site conversion.
2. See what information should be sent to Google Analytics and under what conditions:
In
this section, while documenting, we did all the IP exclusion mappings; parity between old goals
and the new versions; we built a global account to understand how the company was performing at
group level; and prepared a staging system to validate tags before promoting to production.
3. We prepared a two-tier KPI system:
The first tier was focused on each market and allowed
a high level of KPI customization:
The other tier was an executive at-a-glance status view that let us detect inefficiencies, business opportunities and a state comparison against the competition.
Here we'd like to highlight that data visualization worked on 3 levels:
The visual goal of the data was to standardize work along with a defined methodology and joint analysis, while allowing markets to work autonomously, since they're the ones day-to-day with clients and who know market trend shifts well.
Actions after the migration
On the SEO front
After the migration, we took traffic, keyword and performance monitoring of the most critical URLs in rankings very seriously, because we needed to see how the sites were performing in the SERPs with the new version 3.0.
From the beginning we were clear that content had to be improved to better match search intent, so we prioritized this aspect, but in parallel we also made on-page adjustments, as well as a link-building action plan.
Search intent
With a tracking tool, we closely analysed the behaviour of the most important keywords after the migration. At the same time, we kept gathering data from the SERPs and from competitors for those keywords, which, together with the keyword researches we'd done, allowed us to design a content strategy adapted to each market.
These strategies were presented to the marketing leads of each country so that, with their insights, we could define all the final details.
With all of this, the different strategies were approved in priority order, and we launched the content generation plan, which involved writing and translation tasks. At the same time, we had to coordinate to upload the content to the CMS and apply noindex and index directives as the situation required.
With the CRO implementations we made, we were able to test which version was better for both the user and Google.
On page
On the on-page front we made the necessary adjustments to the site architecture and internal linking to adapt to the new content strategies. The idea was to boost the sections most relevant to the business and its objectives.
Data from the DinoRank SEO Suite
Also, content changes imply URL variations, so redirects had to be revisited, server errors checked, sitemaps updated, canonicals reviewed, content analysed, and so on.
In a second phase, different meta titles were to be tested to grow CTR.
Link building
On the link-building side, we found we had sites for very large markets (like UK) with high domain authorities, while sites for smaller markets, such as Lithuania, had low domain authority from receiving lower-quality and fewer links.
However, we also verified that the markets with the highest authority were also competing against large, high-authority sites, while low-authority markets weren't in the same situation.
For that reason, we decided to design a strategy at the global level, intelligently passing authority from one group domain to another (from the big ones to the smaller ones), as well as at the local level, earning links from important media for all domains.
On the CRO front
Throughout 2019, we focused on continuing to improve data quality and visualization. To do this, we standardized a set of micro-conversions that would be useful across all markets, and other in-site events that let us understand user behaviour. If a particular market needed a specific event and it could be replicated across the other countries, we did so.
Also, one of the problems centralization has in terms of conversion is that a product isn't always understood the same way in one country as in another. For example:
This situation made it hard to scale SEO, since Google (and even less so the user) didn't always understand the search intent of that specific landing page. With this in mind, what we did was develop a test pipeline (obviously considering the effects of interactions between tests) so that markets could prepare their commercial proposals and the outcome could be standardized for company leadership.
It's important to mention that, depending on market and sector, it's hard to reach statistical significance on certain tests (i.e. the tire sector in a country like Lithuania). For that reason, what we did was define a system of positive trends, implementation and rollback if we ran into false positives. Here's an example of a test with a good trend:
These tests were superiority tests, with binomial data input and a statistical significance level of 95% for a statistical power of 80%.
After several cases, improvements were positive in several markets, but they must always be monitored over time to avoid, as already stated, false positives:
SEO Migrations Case Study conclusions
With all of this, the results after nearly a year and a half of work are:
+26.56% increase in organic users and an 8.3% increase in qualified leads.
With this case study we hope to show that the challenges multinationals face when aligning a cross-market strategy are at times incompatible with maximizing per-country adaptation, but that protocols and systems must be implemented to maximize value for users and Google in each market.