In October 2016 we received a request for support from wizaz.pl. They needed assistance with the implementation of a new mobile version of one of the site’s most popular sections, Kosmetyk Wszech Czasów (KWC), as well as optimization for the desktop version.
KWC is the largest cosmetics directory in Poland. This giant resource of reviews has brought together a huge community of very committed female users. When it comes to traffic, KWC takes second place in the entire service.
Unfortunately, wizaz.pl’s previous implementation turned out to be unsuccessful and caused a 33% drop in visibility.
Search engine results pages visibility
Drop in KWC’s keywords in SERPs
KWC’s visibility had been decreasing since August 2013 and finally reached its lowest point in January 2017:
The launch of the previous version of KWC took place in March 2016. Unfortunately, as it is shown below, it wasn’t successful. This caused a 33% visibility drop in the whole directory, which we will describe further in this article.
Entire wizaz.pl domain
Likewise, the visibility of the domain had been progressively decreasing since July 2013 (with a few increases and the previously mentioned large decrease in March 2016) before reaching its minimum point in February 2017:
In the case of the forum, after the visibility dropped in 2014, it stabilized and remained at a low level throughout 2015 and 2016:
What happened in March 2016?
After wizaz.pl asked us for help with their visibility, we started to analyze the previous implementation to identify the main problems. We realized that there was more than one problem which impacted the drastic visibility drop.
1. Redirect Chains
During the crawl analysis, we realized that some of the old URLs weren’t being redirected directly to the new URLs. They were instead redirected to the indirect addresses and after that to the target URLs, meaning the old URLs were forgotten and weren’t redirected to the newest ones.
Unfriendly links which were inaccessible to Googlebot were found in the main navigation, such as internal searching results or sorting URLs, e.g. http://wizaz.pl/kosmetyki/akcesoria?search_full[sort]=reviewsLast.
At the same time, SEO friendly URLs existed, e.g. http://wizaz.pl/kosmetyki/akcesoria
We found the same problem in the facet navigation, placed on every category and subcategory page.
Redirections are something obvious and inevitable in SEO, and can be very effective when done properly. However, during the website’s redesign, which included a URLs scheme, the best approach would have been to internally map the URLs to the new scheme. It was this lack of action during the implementation of the previous KWC version that negatively affected several aspects, such as:
- the crawl budget for the entire website was being wasted – Googlebot had to follow redirects instead of going directly to the target URLs.
- the website was slower and more heavy, which isn’t easy to fix in such big websites.
- the redirections didn’t pass along the link juice, meaning it was broken down between two URLs.
4. Domain duplication
The rule which should definitely be observed in SEO is to avoid internal duplication. This means that the same content is available on different URLs. The Wizaz.pl domain was available under both the HTTP and HTTPS versions, which led to the duplication of the entire website. Furthermore, Googlebot indexed URLs with a randomly chosen protocol or sometimes even both of them.
5. Rendering problems
The cosmetics directory also had problems rendering the parts of the website that were important for internal linking. The main and faceted menus were unavailable for Googlebot, causing the removal of the internal links in the main categories. Moreover, clicking on the sidebar menu didn’t send requests to the server – the content was changing without moving to another URL.
Those were only some of the problems we identified at the beginning; however, they had, in our opinion, the biggest impact on the visibility drop in the search results pages.
Technical SEO project scope
We started working on this project in November 2016. From the very beginning, we focused our efforts on the technical SEO of the current version of the KWC directory and, among other things, on repairing the problems mentioned above. Additionally, we made corrections to the rest of the website. Then we made recommendations for a beta version of KWC before it was launched in the first half of 2017. We have been analyzing the beta version of the KWC directory since early April, and since then we have taken intense actions concentrated mainly on this part of the service. Thanks to the effective cooperation with a team of developers on the wizaz.pl side, our recommendations have been efficiently implemented, without threatening the launch date of the production version.
STAGE I: BEFORE THE LAUNCH
1. Full crawl
Thanks to a complete directory crawl by Ryte.com, we identified URLs errors, e.g. 404 errors, 302 redirections, wrong configured canonicals, and unfriendly links placed in the main menu.
2. Preventing internal duplication
Wizaz.pl struggled with internal duplication within the cosmetics vendors and brands pages. Many of these names were cannibalized. We made these pages either redirections or noindex, depending on whether the vendor was the only brand at the time or whether it had multiple cosmetics brands in its portfolio.
3. Indexation strategy
A huge index bloat existed within the domain before the roll out. Every paginated page, filter and sorting URL could be indexed. We provided a large clean up of the Google index for the forum part of the website. Every page there was indexable – the parameters related to the design for the user, sorting and filtering URLs. Furthermore, we checked to see if other types of pages should be indexed. This helped us find the ones which should be removed from the Google index:
- Users profiles and activities related to them
- Internal search result pages
- Single products reviews
- Other URLs of the same recipe (e.g. printing version)
In order for Google to understand the website better and also to display microformats in the result pages, we used schema.org tags in certain elements of the website. We recommended using breadcrumbs, products, and ratings tags.
5. XML sitemaps analysis and errors fixing
In the first stage, the XML sitemaps contained a lot of errors like 404 errors or redirected URLs. A lot of the XML sitemaps and sitemap indexes existed as well. It took a lot of time to crawl all the sitemaps, but we finally managed to clean them, remove those deemed unnecessary, and fix the errors.
6. Initial and final version
The initial version of the mobile KWC directory had been modified several times before it was launched in its present form. We tried to save as much content and functionalities on the mobile version as possible, compared to the desktop version. On the mobile version, there was a lack of main content, and what was there was very poor. We wanted this website to be extremely easy and user-friendly. Finally, we managed to add those elements which improved the website for both users and Googlebot.
Launch of the new version of the KWC directory
The launch of the new version KWC took place on 24th May 2017. We launched the next website’s crawl immediately after that to eliminate as many errors which might negatively affect SEO as possible. Fortunately, there were no critical problems which could have caused visibility drops. We eliminated unnecessary redirections, single page duplications, mapped several URLs, and removed links leading to 404 errors. We also changed some basic SEO elements like headings or meta tags.
As of today, we have achieved the following:
- A visibility recovery after the March 2016 drop.
- +43% visibility increase compared to the month before starting the cooperation.
- + 33% sessions increase comparing November 2017 to November 2016.
STAGE II: AFTER THE LAUNCH
1. Crawling strategy
As a result of the server logs analysis, it turned out that Googlebot had been crawling around many sections of the website where it shouldn’t, or was not advisable. This wasted the crawling budget dedicated to wizaz.pl. We not only analyzed the whole service, but also the forum and editorial parts. In all of these sections of the website, we found elements that should be blocked in the robots.txt file. These included, for example: forum post attachment pages, KWC user profile functionality pages, and comment pages. Of course, we deindexed some types of pages, when necessary. If the order of deindexing and blocking were opposite, it would be impossible to deindex those pages by Googlebot. At the same time, we managed to completely remove some low-value pages from the website’s structure.
2. Index bloat cleaning up
In such a large service, Google index cleaning is almost cyclical. Therefore, we are trying to remove as many of the low-value pages as possible. Especially in the case of the remaining sections of wizaz.pl – with such a wide structure and the existing various sections, it is sometimes difficult to detect immediately.
3. Unblock JS and CSS scripts
Some subdomains were blocked in the robots.txt file. We recommended removing these directives for such subdomains or only for specified paths to let Googlebot crawl the entire website.
In the case of a large website like wizaz.pl, the optimization possibilities are seemingly unlimited. Despite that, we are very satisfied with our cooperation with the wizaz.pl team. The huge challenges such a project provides is something we really appreciate.