In this research paper, we present the results of the analysis conducted on Freeyork.org, a design blog from Poland that has been using WordLift, a plugin for WordPress, for a period of 6 months to automatically add the semantic markup for improving its online visibility.
WordLift 1 analyses articles using Named Entity Recognition (NER) and Named Entity Disambiguation (NED). The entities are ex-tracted from different knowledge graphs including but not limited to DBpedia, GeoNames and Wikidata. WordLift provides UIs for creating and curating custom vocabularies. The plugin implements a semi-automated annotation workflow and publishes metadata by asynchronously injecting a JSON-LD 2 on-page and by publishing linked (open) data in the cloud using Apache Marmotta: an open implementation of a Linked Data Platform 3 . Metrics used to measure the impact of structured data are based on organic search results as opposed to paid links (i.e. advertisement), that is to measure the position of a web site in the search results solely based on the ranking of the web site related to the search terms entered by the end user. So that a user entering a web site from an organic search result is considered an organic session and the traffic generated by this user is defined as organic traffic (vs. paid traffic, i.e. traffic generated by a sponsored or advertised link).
In this analysis, we used Google Analytics to collect and analyze traffic data, over the course of 6 months and we could see that linked structured data helps semantic search engines like Google provide better results to their users and indirectly improves the traffic of a website. This has been measured by comparing both quantitative metrics like pageviews and sessions and qualitative metrics like time spent on page and session duration.
Our goal was to measure organic traffic (page views and sessions) and engagement (time spent on the pages and number of page- views per session) considered as key performance indicators for the website. Since Freeyork.org has an advertising based business model, organic traffic is crucial to drive revenues.
Here follow the highlights:
After the first three months WordLift improved the number of organic sessions and the number of new users with a double digit growth (+ 12.13% of new users).
Google seems to be faster, compared to Bing and Yandex, in indexing and promoting articles that use WordLift’s semantic markup . In the first three months we have seen a 18.47% increase of sessions from Google, comparing with the previous period. Other search engines like Bing and Yandex have also provided more visits to the articles but only after the second and third month of activity.
On average, pages enriched with WordLift (featuring the schema.org markup that the software provides), compared with all the other pages, are performing 2.4 times better in terms of page views and in terms of sessions.
The average time spent on enriched articles is also improved by 17.3% in comparison to the rest of the website that is not annotated.
The average session duration is improved by 13.75% when articles are enriched.
A glossary of terms, like the one that WordLift automatically creates, has a positive impact in extending the length of users’ sessions but has little or no impact on search traffic unless content is original and curated.
Copyright © 2018 by the paper's authors |valid HTML5|