Jump to content

Research:Article Creation Workflow/Clicktracking

From Meta, a Wikimedia project coordination wiki

ACW workflow diagram and events

[edit]
Article Creation Worfklow diagram, as of May 2012.


Event naming scheme

[edit]

The general naming scheme used in Article Creation Workflow is the following:

'ext.articleCreationWorkflow@' + version + '-' + key

where version is the version number as set in the configuration file and key is the event label, which in turn is based on the following scheme (using a hyphen as a separator)

[referral-id]-[bucket-id]-[event]

Identifiers

[edit]

Referral IDs

[edit]
search
user followed a redlink from a search result page
browse
user followed a redlink while browsing a regular article
url
user followed a redlink from the interstitial displayed upon typing in the URL the name of a non-existing page
unknown
the referral is not available

Bucket IDs

[edit]
anon
Anonymous users hitting the landing page
reg
Logged-in users (we intentionally dropped the distinction between already logged-in and people returning to the landing page after logging in)
new
New logged-in users returning to the landing page upon successful signup
control
Users redirected to the default edit screen

Event IDs

[edit]
impression
impression of the landing page
draft_button_click
click on the "create article draft"
login_button_click
click on the "login" button
signup_button_click
click on the "create account" button
article_wizard_button_click
click on the "article wizard" button
create_button_click
click on the "create article" button, which opens the flyover
exit_button_click
click on the "go back" button
create_submit
click on "let's go" submit button, from the flyover
create_skip_submit
click on "let's go" submit button with the "skip" option selected, from the flyover
create_direct_button_click
click on "create" button, directly opening the edit screen and skipping the flyover
create_edit_attempt
attempt to save edit, coming from the flyover
create_skip_edit_attempt
attempt to save edit, coming from the flyover, with the skip option selected
create_direct_edit_attempt
attempt to save edit coming from the ACW landing page (skip option previously selected)
create_edit_success
successfully saved edit, coming from the flyover
create_skip_edit_success
successfully saved edit, coming from the flyover, with skip option selected
create_direct_edit_success
successfully saved edit coming from the ACW landing page (skip option previously selected)

Additional data

[edit]

The log should include as additional data (as the last field of a log record) a pipe-separated value: page_title|rev_id.

rev_id will be available only upon successful completion of an edit.

Full list of events

[edit]

All events will be prefixed with a referral flag {search, browse, url, unknown}. Unless otherwise specified all events are tracked at 100%.

We're temporarily making these referral flags optional
anon-impression
anon-login_button_click
anon-signup_button_click
anon-article_wizard_button_click

reg-impression
reg-draft_button_click
reg-article_wizard_button_click
reg-create_button_click
reg-create_submit
reg-create_skip_submit
reg-create_direct_button_click
reg-create_edit_attempt
reg-create_skip_edit_attempt
reg-create_direct_edit_attempt
reg-create_edit_success
reg-create_skip_edit_success
reg-create_direct_edit_success

new-impression
new-draft_button_click
new-article_wizard_button_click
new-create_button_click
new-create_submit
new-create_skip_submit
new-create_direct_button_click
new-create_edit_attempt
new-create_skip_edit_attempt
new-create_direct_edit_attempt
new-create_edit_success
new-create_skip_edit_success
new-create_direct_edit_success

control-create_edit_impression
control-create_edit_attempt
control-create_edit_success

Log format

[edit]

Clicktracking data is stored via a filter on emery using the following, tab-separated format:

enwiki ext.articleCreationWorkflow@0-new-create_direct_edit_success 20120113013257 0 w4dtyfWSw714KsgmMAAkdC7isy8CmHvP9 0 0 0 0 Pokémon|274278

Log format specification

[edit]
  1. project name (e.g. enwiki or ptwikisource)
  2. event name (as documented above)
  3. timestamp (in YYYYMMDDhhmmss format)
  4. user category: 1 for logged in users, NULL for anons
  5. user token: an anonymous token replacing user names or IP addresses
  6. namespace
  7. lifetime edit count (logged in only)
  8. 6-month edit count (logged in only)
  9. 3-month edit count (logged in only)
  10. last month edit count (logged in only)
  11. additional data (page title, rev_id)

Data collection

[edit]

Events and deployments that affected the collection of this data are documented here

See also

[edit]