Wikimedia Portal clickthrough rate

A lot of people visit the Wikipedia portal - but what proportion actually go anywhere within our sites from it?

This dashboard tracks the proportion of visits to the portal that end in a click through to one of our projects - via search or via one of the links. This is expressed as a percentage - so a value of “39.4” means that 39.4% of visits end in a clickthrough.

Notes

Specifically, the clickthrough rate presented here is the overall rate without any filtering. It is the proportion of click events (by section) over the total number of landing events. By design, a single session is at least 15 minutes but can last indefinitely. For example: a single session can last for hours if the “user” (e.g. a computer in a public library) keeps returning to the page before the 15 minute expiration time, thus resetting the timer; and if that single session has 1000 page visits and 500 clicks, then all 1500 of those events will be used in the calculation of the clickthrough rate. We are considering surfacing the “clickthrough rate on first visit” that some of our A/B test reports use.

Outages and inaccuracies

  • From 7 December (marked “A”) the sampling changed to exclude a broader range of browsers, resulting in alterations to things like clickthrough rate and dwell time. We expect this to resolve itself on 4 January when a new schema version is launched.
  • B (13 September 2016): Added event logging of language-switching, causing some events to flow into old table and some events to flow into the new table. See T143149 for more details.

Broadly-speaking, it's worth noting that (as with all data based on JavaScript logging) the code that gathers this information requires a certain amount of browser capabilities to function. It's probably not going to work on 10 year old Nokia brick phones, and so the data will be biased against users using those kinds of devices.

Questions, bug reports, and feature suggestions

For technical, non-bug questions, email Mikhail. If you experience a bug or notice something wrong or have a suggestion, open a ticket in Phabricator in the Discovery board or email Deb.


Link to this dashboard: http://discovery.wmflabs.org/portal/#clickthrough_rate

Breakdown of user actions on the Wikipedia Portal

In the clickthroughs tab we look at the actual rate of clickthroughs. This graph, on the other hand, breaks user actions down into sub-categories so we can see what features, in particular, people are using, and how it varies over time. The options are:

  1. Language search: the “find a language” box.
  2. No action: Not taking any action on the page.
  3. Other projects: clicking the links to Wiktionary, Wikisource, etc. on the bottom of the page.
  4. Primary links: the prominent links to languages around the globe.
  5. Search: using the Search box and button.
  6. Other languages: using the “Other languages” links near the bottom of the page which take you to a list of Wikipedias.
  7. Secondary links: the less-prominent plaintext links to languages.

The values are expressed as percentages - so if a feature has the value “61.3” it is being used 61.3% of the time.

Outages and inaccuracies

  • From 7 December (marked “A”) the sampling changed to exclude a broader range of browsers, resulting in alterations to things like clickthrough rate and dwell time. We expect this to resolve itself on 4 January when a new schema version is launched.
  • B (13 September 2016): Added event logging of language-switching, causing some events to flow into old table and some events to flow into the new table. See T143149 for more details.

Broadly-speaking, it's worth noting that (as with all data based on JavaScript logging) the code that gathers this information requires a certain amount of browser capabilities to function. It's probably not going to work on 10 year old Nokia brick phones, and so the data will be biased against users using those kinds of devices.

Questions, bug reports, and feature suggestions

For technical, non-bug questions, email Mikhail. If you experience a bug or notice something wrong or have a suggestion, open a ticket in Phabricator in the Discovery board or email Deb.


Link to this dashboard: http://discovery.wmflabs.org/portal/#action_breakdown

Breakdown of most common section clicked on per visit on the Wikipedia Portal

In the clickthroughs tab we look at the actual rate of clickthroughs and in the action breakdown tab we can see that clickthrough broken down into the “last action taken”. This graph shows the most common clicked section per visit. Our analyses showed that any one session can have multiple visits and each visit can have multiple clicks associated with it. This shows us (for each section) the proportion of visits in which that section was the most commonly clicked on. The options are:

  1. No action: Not taking any action on the page.
  2. Other projects: clicking the links to Wiktionary, Wikisource, etc. on the bottom of the page.
  3. Primary links: the prominent links to languages around the globe.
  4. Search: using the Search box and button.
  5. Secondary links: the less-prominent plaintext links to languages.

The values are expressed as percentages - so if a feature has the value “61.3” it is being used 61.3% of the time.

Outages and notes

  • A (13 September 2016): Added event logging of language-switching, causing some events to flow into old table and some events to flow into the new table. See T143149 for more details.

Questions, bug reports, and feature suggestions

For technical, non-bug questions, email Mikhail. If you experience a bug or notice something wrong or have a suggestion, open a ticket in Phabricator in the Discovery board or email Deb.


Link to this dashboard: http://discovery.wmflabs.org/portal/#most_common

Breakdown of Wikipedia Portal visitors' first visit

In the clickthroughs tab we look at the actual rate of clickthroughs. In the actions breakdown tab we look at the proportions of sessions performing various types of actions. This graph shows the actions breakdown at first visit and ignores any subsequent visits the user may have made. The options are:

  1. Language search: the “find a language” box.
  2. No action: Not taking any action on the page.
  3. Other projects: clicking the links to Wiktionary, Wikisource, etc. on the bottom of the page.
  4. Primary links: the prominent links to languages around the globe.
  5. Search: using the Search box and button.
  6. Other languages: using the “Other languages” links near the bottom of the page which take you to a list of Wikipedias.
  7. Secondary links: the less-prominent plaintext links to languages.

The values are expressed as percentages - so if a feature has the value “61.3” it is being used 61.3% of the time.

In the future, we might include data such as the average time to first clickthrough from initial page landing.

Outages and notes

  • A (13 September 2016): Added event logging of language-switching, causing some events to flow into old table and some events to flow into the new table. See T143149 for more details.

Questions, bug reports, and feature suggestions

For technical, non-bug questions, email Mikhail. If you experience a bug or notice something wrong or have a suggestion, open a ticket in Phabricator in the Discovery board or email Deb.


Link to this dashboard: http://discovery.wmflabs.org/portal/#first_visit

Dwell-time on the Wikipedia portal

This measures “dwell-time” on the Wikipedia portal: the amount of time people spend between arriving at the page and leaving. As “leaving” is defined as clicking on something for data collection purposes, this only covers the population that actually clicks through to pages or using search, not the population that simply closes their tab or browser, or goes to a different website altogether.

Outages and inaccuracies

  • From 7 December (marked “A”) the sampling changed to exclude a broader range of browsers, resulting in alterations to things like clickthrough rate and dwell time. We expect this to resolve itself on 4 January when a new schema version is launched.
  • B (13 September 2016): Added event logging of language-switching, causing some events to flow into old table and some events to flow into the new table. See T143149 for more details.

Broadly-speaking, it's worth noting that (as with all data based on JavaScript logging) the code that gathers this information requires a certain amount of browser capabilities to function. It's probably not going to work on 10 year old Nokia brick phones, and so the data will be biased against users using those kinds of devices.

Questions, bug reports, and feature suggestions

For technical, non-bug questions, email Mikhail. If you experience a bug or notice something wrong or have a suggestion, open a ticket in Phabricator in the Discovery board or email Deb.


Link to this dashboard: http://discovery.wmflabs.org/portal/#dwell_data

Where users come from to the Wikipedia portal

This measures where portal users come from each day, displaying the proportion of users from each top-10 country and the proportion from other countries, as a percentage.

Notes

  • Broadly-speaking, it's worth noting that (as with all data based on JavaScript logging) the code that gathers this information requires a certain amount of browser capabilities to function. It's probably not going to work on 10 year old Nokia brick phones, and so the data will be biased against users using those kinds of devices.
  • A: On 28 June 2016 our Event Logging system started recording a finer view of U.S. traffic, breaking it down into 5 regions:
    • Northeast Region
      • New England Division: Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island and Vermont
      • Middle Atlantic Division: New Jersey, New York and Pennsylvania
    • Midwest Region
      • East North Central Division: Illinois, Indiana, Michigan, Ohio and Wisconsin
      • West North Central Division: Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota and South Dakota
    • South Region
      • South Atlantic Division: Delaware, District of Columbia, Florida, Georgia, Maryland, North Carolina, South Carolina, Virginia and West Virginia
      • East South Central Division: Alabama, Kentucky, Mississippi and Tennessee
      • West South Central Division: Arkansas, Louisiana, Oklahoma and Texas
    • West Region
      • Mountain Division: Arizona, Colorado, Idaho, Montana, Nevada, New Mexico, Utah and Wyoming
    • Pacific Region
      • Alaska, California, Hawaii, Oregon and Washington
  • B (13 September 2016): Added event logging of language-switching, causing some events to flow into old table and some events to flow into the new table. See T143149 for more details.

See T136257 for more details.

Outages and inaccuracies

  • A (13 September 2016): Added event logging of language-switching, causing some events to flow into old table and some events to flow into the new table. See T143149 for more details.

Questions, bug reports, and feature suggestions

For technical, non-bug questions, email Mikhail. If you experience a bug or notice something wrong or have a suggestion, open a ticket in Phabricator in the Discovery board or email Deb.


Link to this dashboard: http://discovery.wmflabs.org/portal/#country_breakdown

Case insensitive & accepts comma-separated input.

What users use to view the Wikipedia portal

This measures which Internet browsers portal users use from each day, displaying the proportion of traffic accounted for by the browser. For any given day, we only record browsers with at least 0.5% share of users.

Sorting options

  • Alphabetical sorting and sorting by last recorded percentage do what they say on the tin.
  • Sorting by popularity growth/decay uses an estimated rate (computed by fitting an exponential model).
  • Sorting by number of times can be useful because a browser is only included if it has at least 0.5% share, so the number of times can be regarded as a rough estimate of longevity.

Grouping options

  • Browser Family collapses the different versions of browsers. For example: “Chrome 40”, … , “Chrome 48” become just “Chrome”.
  • MediaWiki Support groups the browsers according to their support on MediaWiki. See the browser support matrix on Compatibility page for more details, such as why Chrome Mobile (iOS & Android) and Opera Mobile are included in “Unknown” group. See T130027 for the patch notes related to this grouping option.

Outages and inaccuracies

None so far!

Questions, bug reports, and feature suggestions

For technical, non-bug questions, email Mikhail. If you experience a bug or notice something wrong or have a suggestion, open a ticket in Phabricator in the Discovery board or email Deb.


Link to this dashboard: http://discovery.wmflabs.org/portal/#browser_breakdown

Pageviews (PVs) to Wikipedia.org Portal

This looks at, without sampling, the number of pageviews the Wikipedia Portal gets per day. This is expressed as raw values. See our Hive query for the technical definition of a Wikipedia.org pageview.

Notes

  • A: Started filtering out search-redirect.php requests. See T138411 for more information.
  • B: Rise could not be determined due to deleted data. See write-up and T143045 for more information.
  • C: The investigation of pageviews rise caused us to redefine how we count wikipedia.org pageviews. Pageviews from 2016-06-22 to 2016-08-22 were then recounted using the new definition. See T143064 for more information.
  • D: On 11 July 2016 we started to split pageview counts into pageviews from “low-volume” clients and “high-volume” clients. A “high-volume” client is a client whose wikipedia.org pageviews are equal to or greater than the 99.99th percentile for the whole population on any particular day. The rationale for this being that the low-volume clients' PV counts would be more stable and the high-volume clients' PV counts would soak up outliers and bots. See T143605 for more details.

Questions, bug reports, and feature suggestions

For technical, non-bug questions, email Mikhail. If you experience a bug or notice something wrong or have a suggestion, open a ticket in Phabricator in the Discovery board or email Deb.


Link to this dashboard: http://discovery.wmflabs.org/portal/#pageview_tab

Traffic to Wikipedia Portal from external search engines

This dashboard simply looks at, very broadly, where our Portal pageviews are coming from. Direct traffic is users typing “wikipedia.org” directly into their browser or using a bookmark. It should also be noted that our request refinery only tracks top global search engines, so some search engines will make it into the “referred by something other than search engine” category despite being a search engine.

General trends

On average (or “average” because we use medians), traffic from search engines accounts for 1.42% of overall traffic to the Portal.

Outages and notes

  • A: We switched to a finalized version of the UDF that extracts internal traffic (see T130083)
  • B: Started filtering out search-redirect.php requests. See T138411 for more information.

Questions, bug reports, and feature suggestions

For technical, non-bug questions, email Mikhail. If you experience a bug or notice something wrong or have a suggestion, open a ticket in Phabricator in the Discovery board or email Deb.


Link to this dashboard: http://discovery.wmflabs.org/portal/#referrals_summary

Traffic to Wikipedia Portal from external search engines, broken down

This dashboard simply breaks down the summary data to investigate how much of search engine traffic is coming from each engine individually.

General trends

On average, Google (the largest engine in our dataset) accounts for 80-81% of the search-engine-referred traffic. Since search-engine-referred traffic accounts for 1.42% of the overall Portal traffic, Google referrals accounts for approximately 1.14% of the overall traffic.

Outages and inaccuracies

None so far!

Questions, bug reports, and feature suggestions

For technical, non-bug questions, email Mikhail. If you experience a bug or notice something wrong or have a suggestion, open a ticket in Phabricator in the Discovery board or email Deb.


Link to this dashboard: http://discovery.wmflabs.org/portal/#search_engines

Traffic to Wikipedia from Wikipedia.org Portal

This dashboard shows the number of clicks (and users) to Wikipedia (across all the languages it is in) as recorded by our Portal event logging, which users are randomly selected into at a sampling rate of 1 in 200 – 0.5%. When the user comes to wikipedia.org and are randomly selected to be anonymously tracked via event logging, we set a timer of 15 minutes. Every time the user comes back to the page (lands), the timer is reset. After 15 minutes of not returning, the user's session is no longer tracked. For this reason, a single session may have one or dozens (sometimes hundreds!) of visits, each with a click or multiple clicks. We have seen sessions with as many as hundred clicks following the first and only landing. In cases where almost every session only has a single click associated with it, the graphs for clicks and users will look VERY similar, if not exactly the same.

When looking at clicks, you can view counts or proportions to see how much of the traffic comes from the three sections:

  • search: wikipedia.org visitors can search Wikipedia's in different languages and end up on specific articles and we log the language of the Wikipedia they visited. If the visitor did not find a specific article during their initial query from the search metadata that is displayed, or by hitting 'Enter', they will be redirected to a default search results page in the language that they searched in (even if they changed the language in the small dropdown in the search box while on the Portal). However, at this time the search-box language selection change is not logged.
  • primary: the links around the Wikipedia globe logo, which are dynamically placed and sorted according to each visitor's browser's language preferences.
  • secondary: the links below the Wikipedia globe logo, which the wikipedia.org visitor can use to find a version of Wikipedia in any of the almost 300 languages.

Note: Sister project link clickthroughs are not tracked on this page, see this page for more info.

Until 27 June 2016, traffic to Wikipedias was almost evenly split between people searching and people using the primary links. Since then, search has been the more dominant section. However, this is due to a rise in Portal visitors searching EnWiki from the Portal. If we exclude EnWiki from the total, we see that clicks on primary links consistently account for 50-60% of the traffic to Wikipedias in languages other than English, searches account for ~35% of the traffic, and clicks on secondary links account for ~8% of the traffic.

Outages and notes

  • A: Languages visited data backfilled
    1. The data we used for a retrospective study of Portal deployments started on 16 November 2016, although there were filters applied to the data used in the analysis. Specifically, known spiders were excluded and only data from the first 10 visits per session was kept for data storage space reasons.
    2. When we began work on this part of the dashboard, we could only backfill data from 2016-05-10 due to the 90-day restriction our event logging system has. Therefore, we had to use the previously saved (slightly filtered) data to backfill visited language counts from November 16th to May 9th. We checked how the filtered data (post May 10th) compared to the unfiltered data and some counts were off by 1-8 clicks, hence why we are noting the difference here.
  • B (13 September 2016): Added event logging of language-switching, causing some events to flow into old table and some events to flow into the new table. See T143149 for more details.

Questions, bug reports, and feature suggestions

For technical, non-bug questions, email Mikhail. If you experience a bug or notice something wrong or have a suggestion, open a ticket in Phabricator in the Discovery board or email Deb.


An experimental dashboard for looking at patterns of Wikipedia.org Portal usage pertaining to users' languages. See T140816 for more information. Link to this dashboard: http://discovery.wmflabs.org/portal/#languages_summary

Traffic to Wikipedia in a particular language

This dashboard shows the number of clicks (and users) from Wikipedia.org Portal to Wikipedia in any language (that the user may be interested in) as recorded by our Portal event logging, which users are randomly selected into at a sampling rate of 1 in 200 – 0.5%. When the user comes to wikipedia.org and are randomly selected to be anonymously tracked via event logging, we set a timer of 15 minutes. Every time the user comes back to the page (lands), the timer is reset. After 15 minutes of not returning, the user's session is no longer tracked. For this reason, a single session may have one or dozens (sometimes hundreds!) of visits, each with a click or multiple clicks. We have seen sessions with as many as hundred clicks following the first and only landing. In cases where almost every session only has a single click associated with it, the graphs for clicks and users will look VERY similar, if not exactly the same.

While the summary panel shows the total traffic from the Portal to Wikipedia across all the available languages, this panel allows you to view traffic from the Portal to a specific language and compare several languages. There are several sorting options to choose from, which affects the order of the languages listed when you click to select additional languages. Previously selected languages can be removed with the backspace key. When switching from one sorting option to another, the selected language(s) may change because they might not be part of the new subset of languages (e.g. switching from “Top 10” to “Bottom 50”), in which case the first language in the list will be selected by default.

When looking at clicks, you can view counts or proportions to see how much of the traffic comes from the three sections:

  • search: wikipedia.org visitors can search Wikipedia's in different languages and end up on specific articles and we log the language of the Wikipedia they visited. If the visitor did not find a specific article during their initial query from the search metadata that is displayed, or by hitting 'Enter', they will be redirected to a default search results page in the language that they searched in (even if they changed the language in the small dropdown in the search box while on the Portal). However, at this time the search-box language selection change is not logged.
  • primary: the links around the Wikipedia globe logo, which are dynamically placed and sorted according to each visitor's browser's language preferences.
  • secondary: the links below the Wikipedia globe logo, which the wikipedia.org visitor can use to find a version of Wikipedia in any of the almost 300 languages.

Note: Sister project link clickthroughs are not tracked on this page, see this page for more info.

Viewing count/proportional breakdown of clicks is only available when a single language is selected. When multiple languages are selected, only the total number of clicks of the three sections is shown for each language. When multiple languages are selected (e.g. Top 10 or Bottom 50), the order they appear in inside the legend corresponds to the total number of clicks (or users), from highest to lowest. If several languages have the same number of historical clicks/users, they are then sorted further alphabetically by name.

By far, the English Wikipedia is the most visited one. Majority of the traffic to EnWiki is from Portal visitors searching (consistently >50% and as high as 70%), while primary link clicks account for 30-40% of the traffic to EnWiki from the Portal.

Outages and notes

  • A: Languages visited data backfilled
    1. The data we used for a retrospective study of Portal deployments started on 16 November 2016, although there were filters applied to the data used in the analysis. Specifically, known spiders were excluded and only data from the first 10 visits per session was kept for data storage space reasons.
    2. When we began work on this part of the dashboard, we could only backfill data from 2016-05-10 due to the 90-day restriction our event logging system has. Therefore, we had to use the previously saved (slightly filtered) data to backfill visited language counts from November 16th to May 9th. We checked how the filtered data (post May 10th) compared to the unfiltered data and some counts were off by 1-8 clicks, hence why we are noting the difference here.
  • B (13 September 2016): Added event logging of language-switching, causing some events to flow into old table and some events to flow into the new table. See T143149 for more details.

Questions, bug reports, and feature suggestions

For technical, non-bug questions, email Mikhail. If you experience a bug or notice something wrong or have a suggestion, open a ticket in Phabricator in the Discovery board or email Deb.


Link to this dashboard: http://discovery.wmflabs.org/portal/#languages_visited

to

Wikimedia Portal Traffic and Clickthrough Rate by Geography

This dashboard shows where portal users come from each day, displaying the number of events, number of visits and the number of sessions sampled according to our WikipediaPortal schema at a 1:200 rate. This dashboard also tracks the proportion of events/visits/sessions to the portal that end in a click through to one of our projects - via search or via one of the links. This is expressed as a percentage - so a value of “39.4” means that 39.4% of visits end in a clickthrough.

Notes

  • This dashboard is aim to expand the “other” countries display in the current Portal Dashboard to reflect more data that we're interested in seeing. See T138107 for more details.
  • Broadly-speaking, it's worth noting that (as with all data based on JavaScript logging) the code that gathers this information requires a certain amount of browser capabilities to function. It's probably not going to work on 10 year old Nokia brick phones, and so the data will be biased against users using those kinds of devices.
  • On 28 June 2016 our Event Logging system started recording a finer view of U.S. traffic, breaking it down into 5 regions:
    • Northeast Region
      • New England Division: Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island and Vermont
      • Middle Atlantic Division: New Jersey, New York and Pennsylvania
    • Midwest Region
      • East North Central Division: Illinois, Indiana, Michigan, Ohio and Wisconsin
      • West North Central Division: Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota and South Dakota
    • South Region
      • South Atlantic Division: Delaware, District of Columbia, Florida, Georgia, Maryland, North Carolina, South Carolina, Virginia and West Virginia
      • East South Central Division: Alabama, Kentucky, Mississippi and Tennessee
      • West South Central Division: Arkansas, Louisiana, Oklahoma and Texas
    • West Region
      • Mountain Division: Arizona, Colorado, Idaho, Montana, Nevada, New Mexico, Utah and Wyoming
    • Pacific Region
      • Alaska, California, Hawaii, Oregon and Washington
    • Other
      • American Samoa, Guam, Northern Mariana Islands, Puerto Rico and Virgin Islands, U.S.
  • The overall clickthrough rate presented here is the proportion of click events over the total number of landing events. The clickthrough rate per visit is the proportion of landing events that end in at least one click through. The clickthrough rate per session is the proportion of sessions that end in at least one click through. By design, a single session is at least 15 minutes but can last indefinitely. For example: a single session can last for hours if the “user” (e.g. a computer in a public library) keeps returning to the page before the 15 minute expiration time, thus resetting the timer; and if that single session has 1000 page visits and 500 clicks, then all 1500 of those events will be used in the calculation of the overall clickthrough rate.

Outages and inaccuracies

  • 13 September 2016: Added event logging of language-switching, causing some events to flow into old table and some events to flow into the new table. See T143149 for more details.

Questions, bug reports, and feature suggestions

For technical, non-bug questions, email Mikhail. If you experience a bug or notice something wrong or have a suggestion, open a ticket in Phabricator in the Discovery board or email Deb.


Link to this dashboard: http://discovery.wmflabs.org/portal/#all_country

to

Breakdown of Wikipedia Portal Visitors' First Visit by Geography

This dashboard shows the geographical breakdown at first visit action and ignores any subsequent visits the user may have made. The actions are:

  1. No action: Not taking any action on the page.
  2. Other projects: clicking the links to Wiktionary, Wikisource, etc. on the bottom of the page.
  3. Primary links: the prominent links to languages around the globe.
  4. Search: using the Search box and button.
  5. Other languages: using the “Other languages” links near the bottom of the page which take you to a list of Wikipedias.
  6. Secondary links: the less-prominent plaintext links to languages.

The values are expressed as numbers (sampled at a 1:200 rate) and as percentages - so if a country has the value “61.3” for search action, then in first visits, 61.3% of all searches are done by users from this country.

Notes

  • This dashboard is aim to expand the “other” countries display in the current Portal Dashboard to reflect more data that we're interested in seeing. See T138107 for more details.
  • Broadly-speaking, it's worth noting that (as with all data based on JavaScript logging) the code that gathers this information requires a certain amount of browser capabilities to function. It's probably not going to work on 10 year old Nokia brick phones, and so the data will be biased against users using those kinds of devices.
  • On 28 June 2016 our Event Logging system started recording a finer view of U.S. traffic, breaking it down into 5 regions:
    • Northeast Region
      • New England Division: Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island and Vermont
      • Middle Atlantic Division: New Jersey, New York and Pennsylvania
    • Midwest Region
      • East North Central Division: Illinois, Indiana, Michigan, Ohio and Wisconsin
      • West North Central Division: Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota and South Dakota
    • South Region
      • South Atlantic Division: Delaware, District of Columbia, Florida, Georgia, Maryland, North Carolina, South Carolina, Virginia and West Virginia
      • East South Central Division: Alabama, Kentucky, Mississippi and Tennessee
      • West South Central Division: Arkansas, Louisiana, Oklahoma and Texas
    • West Region
      • Mountain Division: Arizona, Colorado, Idaho, Montana, Nevada, New Mexico, Utah and Wyoming
    • Pacific Region
      • Alaska, California, Hawaii, Oregon and Washington
    • Other
      • American Samoa, Guam, Northern Mariana Islands, Puerto Rico and Virgin Islands, U.S.

Outages and inaccuracies

  • 13 September 2016: Added event logging of language-switching, causing some events to flow into old table and some events to flow into the new table. See T143149 for more details.

Questions, bug reports, and feature suggestions

For technical, non-bug questions, email Mikhail. If you experience a bug or notice something wrong or have a suggestion, open a ticket in Phabricator in the Discovery board or email Deb.


Link to this dashboard: http://discovery.wmflabs.org/portal/#first_visits_by_country

to

Breakdown of Users' Last Actions on the Wikipedia Portal by Geography

This dashboard breaks users' last actions at each session down by geography. The actions are:

  1. No action: Not taking any action on the page.
  2. Other projects: clicking the links to Wiktionary, Wikisource, etc. on the bottom of the page.
  3. Primary links: the prominent links to languages around the globe.
  4. Search: using the Search box and button.
  5. Other languages: using the “Other languages” links near the bottom of the page which take you to a list of Wikipedias.
  6. Secondary links: the less-prominent plaintext links to languages.

The values are expressed as numbers (sampled at a 1:200 rate) and as percentages - so if a country has the value “61.3” for search action, then 61.3% of all searches are done as last actions by users from this country.

Notes

  • This dashboard is aim to expand the “other” countries display in the current Portal Dashboard to reflect more data that we're interested in seeing. See T138107 for more details.
  • Broadly-speaking, it's worth noting that (as with all data based on JavaScript logging) the code that gathers this information requires a certain amount of browser capabilities to function. It's probably not going to work on 10 year old Nokia brick phones, and so the data will be biased against users using those kinds of devices.
  • On 28 June 2016 our Event Logging system started recording a finer view of U.S. traffic, breaking it down into 5 regions:
    • Northeast Region
      • New England Division: Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island and Vermont
      • Middle Atlantic Division: New Jersey, New York and Pennsylvania
    • Midwest Region
      • East North Central Division: Illinois, Indiana, Michigan, Ohio and Wisconsin
      • West North Central Division: Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota and South Dakota
    • South Region
      • South Atlantic Division: Delaware, District of Columbia, Florida, Georgia, Maryland, North Carolina, South Carolina, Virginia and West Virginia
      • East South Central Division: Alabama, Kentucky, Mississippi and Tennessee
      • West South Central Division: Arkansas, Louisiana, Oklahoma and Texas
    • West Region
      • Mountain Division: Arizona, Colorado, Idaho, Montana, Nevada, New Mexico, Utah and Wyoming
    • Pacific Region
      • Alaska, California, Hawaii, Oregon and Washington
    • Other
      • American Samoa, Guam, Northern Mariana Islands, Puerto Rico and Virgin Islands, U.S.

Outages and inaccuracies

  • 13 September 2016: Added event logging of language-switching, causing some events to flow into old table and some events to flow into the new table. See T143149 for more details.

Questions, bug reports, and feature suggestions

For technical, non-bug questions, email Mikhail. If you experience a bug or notice something wrong or have a suggestion, open a ticket in Phabricator in the Discovery board or email Deb.


Link to this dashboard: http://discovery.wmflabs.org/portal/#last_action_by_country

to

Breakdown of Most Common Section Clicked on per Visit on the Wikipedia Portal by Geography

This dashboard shows the most common clicked section per visit. Our analyses showed that any one session can have multiple visits and each visit can have multiple clicks associated with it. This shows us (for each section) the number/proportion of visits in which that section was the most commonly clicked on. The sections are:

  1. No action: Not taking any action on the page.
  2. Other projects: clicking the links to Wiktionary, Wikisource, etc. on the bottom of the page.
  3. Primary links: the prominent links to languages around the globe.
  4. Search: using the Search box and button.
  5. Other languages: using the “Other languages” links near the bottom of the page which take you to a list of Wikipedias.
  6. Secondary links: the less-prominent plaintext links to languages.

The values are expressed as numbers (sampled at a 1:200 rate) and as percentages - so if a country has the value “61.3” for search section, then 61.3% of all visits (in which search section was the most commonly clicked on) are from users in this country.

Notes

  • This dashboard is aim to expand the “other” countries display in the current Portal Dashboard to reflect more data that we're interested in seeing. See T138107 for more details.
  • Broadly-speaking, it's worth noting that (as with all data based on JavaScript logging) the code that gathers this information requires a certain amount of browser capabilities to function. It's probably not going to work on 10 year old Nokia brick phones, and so the data will be biased against users using those kinds of devices.
  • On 28 June 2016 our Event Logging system started recording a finer view of U.S. traffic, breaking it down into 5 regions:
    • Northeast Region
      • New England Division: Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island and Vermont
      • Middle Atlantic Division: New Jersey, New York and Pennsylvania
    • Midwest Region
      • East North Central Division: Illinois, Indiana, Michigan, Ohio and Wisconsin
      • West North Central Division: Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota and South Dakota
    • South Region
      • South Atlantic Division: Delaware, District of Columbia, Florida, Georgia, Maryland, North Carolina, South Carolina, Virginia and West Virginia
      • East South Central Division: Alabama, Kentucky, Mississippi and Tennessee
      • West South Central Division: Arkansas, Louisiana, Oklahoma and Texas
    • West Region
      • Mountain Division: Arizona, Colorado, Idaho, Montana, Nevada, New Mexico, Utah and Wyoming
    • Pacific Region
      • Alaska, California, Hawaii, Oregon and Washington
    • Other
      • American Samoa, Guam, Northern Mariana Islands, Puerto Rico and Virgin Islands, U.S.

Outages and inaccuracies

  • 13 September 2016: Added event logging of language-switching, causing some events to flow into old table and some events to flow into the new table. See T143149 for more details.

Questions, bug reports, and feature suggestions

For technical, non-bug questions, email Mikhail. If you experience a bug or notice something wrong or have a suggestion, open a ticket in Phabricator in the Discovery board or email Deb.


Link to this dashboard: http://discovery.wmflabs.org/portal/#most_common_by_country