Wikilivres uses various sources for analytics information, including Google, Bing, and Yandex Webmaster Tools, Piwik Analytics, and the internal data from Wikilivres. Together these help provide insight into how the site is being used.
- Vnstats shows throughput of 600gb p/m @ ~3 Mbit/s average load.
- Pingdom reports 96.73% uptime from 2018-12-03 to 2018-12-09, based on 6 outages with a combined 5 hours duration.
- Without the compression historically used, the database and image storage takes around 120gb of space. Server capacity has been expanded to cope.
- Pingdom reports 98.71% uptime from 2018-12-17 to 2018-12-23, based on 4 outages with a combined 2h 10m duration.
- Christmas period fairly stable at approximately 250 unique visitors daily.
November 2018 Piwik chart
4,400 total visits. (5 - 11 Nov 2018)
|Country||Percentage of total visitor||Unique visitors|
Does anyone know if Turkey blocks Wikisource?
08-30 April 2017
Analytics data from search (webmaster tools) and from site visits (Piwik) came online about the second week of April.
Site Traffic from Search vs. Referral
Initial enthusiasm by search engines peaked and then waned. Tens of thousands of pages and files were indexed and were being served by search engines. However, the relatively poor performance cut especially Google's enthusiasm. By the end of the month, the average unique visitors per day are about 500, with a little more than half coming from Google.
Site Referral Traffic
Most site referral traffic is from other wikis, principally wikipedia (various languages), wikisource, and wikibooks sites.
Search and Page Traffic
While Yandex and other search engines provide some traffic, it Google who provides the principal amount. There are two main issues to consider:
- Search terms used (for which Wikilivres provides results)
- Pages that are popular (through various search terms)
This spreadsheet for April 2017 has two tabs with both kinds of information. This data comes from Google Webmaster Tools (and is only available to those who can validate control over the site). The spreadsheet uses some Google functions to url-decode, detect language, and translate the titles of pages and searches (into English).
The spreadsheet linked above shows a lot of diversity of documents being served, and a standard power-law regarding what is popular. The most popular content:
- Chronicles of Narnia
- Shakespearean Sonnets (in Russian Translation)
- Zinaida Gippius (Russian poet)
- Various other poetry, much of it in Russian Translation (Catallus, Rimbaud, Verlaine, Basho)
For English, besides Narnia is another C.S. Lewis book, Mere Christianity; Richard Wright's Native Son; Aldous Huxley's Brave New World. For French, Boris Vian's The Foam of Days is the most popular, followed by Antoine de Saint-Exupéry's The Little Prince.
General Sense of What is Found via Search
For Google, Wikilivres is clearly a cache of Russian poetry, Poetry in Russian Translation, and smaller collection of English and French classic books. Some of these classic works (in their original language) have additional translations available. It seems there might be a useful opportunity to build on what is here, and have sources in different languages. For example: Shakespeare in Russian, why not add the English, German, French? Anything in a foreign language will most likely have an original language also available in the public domain. As well, for works here that search is providing readership for, more from the same author might be a nice idea. --Jeffmcneill (talk) 08:53, 1 May 2017 (UTC)
20 April 2017 - Unique visitors to Wikilivres hit 1,000+ Day
Hi everyone, just wanted to let you know as of Sunday we are over 1,000 unique visitors per day, with more than 50% coming from Google. Google just started giving us traffic (and they haven't indexed the entire site). I had created a new sitemap and submitted to Google, Bing, Yandex. Bing has indexed 35,000+ pages, Yandex has indexed 21,000+ (not all are in live search yet), and Google is now providing 500+ visitors/day. There are still lots of errors coming out of these search engines, and not everything submitted is a page with content (Yandex is chugging through all the image files on the site, and Bing did also). However, we have started to get noticed. There is a lot more to do, on my side to clean up the sitemaps, and get better search-engine-readable markup on the pages, but we have begun.
I'll provide monthly stats, including popular works, geographical traffic patterns. FYI, as of the data I have for the last week or so, Russia has the most visitors, followed by the US and France, but France has the most page-views. Ukraine, India, Canada, Poland, Czech, UK, and Germany round out the top ten in terms of visits. These patterns will change as search comes to the fore. Lots of Russian, French, English, and Czech content coming through search at this point. Truly an international project. --Jeffmcneill (talk) 16:25, 20 April 2017 (UTC)
- note: -- main page needs a(n improved?) live/auto updater for info-stats, & links to a project:about page, & a stats/newest stuff page; as per the style on wm projects(?). Lx 121 (talk) 20:09, 24 April 2017 (UTC)
- that or there seems to be some serious discrepancy btwn the page count numbers from the search engines, & the internal mw page count numbers our stats are showing? Lx 121 (talk) 20:18, 24 April 2017 (UTC)
- What counts as a page or a file or a work are definitely not necessarily intuitive. Just include talk pages and things like MediaWiki pages, and you can see how numbers are inflated. Also, for a given image there can be: the image itself, the image "page" and the image "talk_page". For DJVU files, these can have inside of them hundreds of png files. Since these are all indexable, there is an explosion when it comes to what is countable (vs. what counts). --Jeffmcneill (talk) 07:37, 1 May 2017 (UTC)