Reading/Search Engine Optimization

Background

The number of users and visits to Wikipedia are directly related to our global impact, pool of potential editors, and fundraising. The majority of Wikipedia users and visits come to us through Google search results - determined by a constantly changing algorithm and presentation scheme. This brings more people to our projects - readers, editors, donors.

Despite our current high rankings, we still have space in which to grow. Our high-quality results on English Wikipedia may not transfer to smaller projects. This creates an inequality of content access. We are also not investing in the most optimal ways to structure our content and potentially missing opportunities for improvement.

Research and implementation

Over the 2018-2019 fiscal year, we invested in the following:

Identifying weaknesses and opportunities for improvement:

Contracting GoFish digital to review and recommend improvements
Compiling and prioritizing SEO improvement recommendations
Researching the viability and effects of suggested improvements

Moving towards rigorous analysis:

Looking at current patterns in search engine optimization
Measuring effects of our changes through intervention analysis and A/B tests
Prioritizing for impact, starting with English Wikipedia before moving to other languages

Implementing changes to make sites easier to crawl:

Creating sitemaps to tell crawlers how pages relate to one another
Adding annotations that make our sites easier to understand by crawlers

Enabling Schema.org article linked data for main namespace pages

We began exploring schema.org markup within our articles as a means of providing search engines with more confidence within the context of ambiguous queries as well as with the general goal of increasing the amount of structure data across our sites overall. We began by adding the sameAs meta property to Wikipedia articles which pointed to the corresponding Wikidata entity. To determine the success of the changes, we ran an A/B test on 50% of articles across most of our projects (phab:T206868).

A/B test results

Using hierarchical regression modeling of search engine-referred daily traffic to 269 editions of Wikipedia, we estimated the effect of adding the sameAs property to be a 1.4% increase (95% CI: 0.7-2.1) in average page views per day. Based on this increase, we decided to roll out the feature to 100% of main namespace pages (where applicable) on all Wikipedias.

The full report of our A/B test is available here.

Deployment Timelines

The sameAs schema property will be added to 100% of main namespace pages in April, 2019

Enabling $wgMFNoindexPages

To properly index mobile sites, we looked into adding a rel = "alternate" tag to mobile versions of the page. In T206497, this was enabled on Italian, Dutch, Korean, Arabic, Chinese, and Hindi Wikipedias on October 2, 2019 to determine the effects of this fix on search engine-referred traffic from mobile search engines.

Analysis Results

We reviewed trends in the average pageviews per day pre and post deployment to both the desktop and mobile versions of test wikis, focusing on external search-related traffic to determine if there were any significant changes (phab: T234807). These trends were compared to previous years to account for any seasonal fluctuations. We also reviewed google search console data to determine impressions and clicks to the test wikis before and after the deployment date.

Results indicated that there was no negative impact on search-referred traffic to the test wikis. There were no visible changes in traffic to either the desktop or mobile versions of the site that could be clearly attributed to the deployment of this change. As there was no negative impact, we will enable $wgMFNoindexPages to all projects