Skip to content
This repository has been archived by the owner on Mar 2, 2021. It is now read-only.

SQL data model for working with Snowplow web data. Supports Redshift and Looker. Snowflake and BigQuery coming soon

Notifications You must be signed in to change notification settings

snowplow/snowplow-web-data-model

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

34 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Web data model

Deprecation notice

This version of the web model is now deprecated.

The latest version can be found in the new data-models repo.

It is incremental, more extensible, and fully supported - unlike the one here which was originally intended as an example.

Table of contents

1. Introduction

The most common tracker for Snowplow users to get started with is the JavaScript Tracker. Like all our trackers, you can use it to track the self-describing events and entities that you define yourself. In addition, Snowplow provides built-in support for the web-native events that most users will want to track. This includes events such as page views, page pings, and link clicks.

The Snowplow SQL data model makes it easier to get started with web data. It aggregates the page view and page ping events to create a set of derived tables that contain a lot of detail, including: time engaged, scroll depth, and page performance. The model comes in three variants:

  1. A straightforward set of SQL queries Use this version to model your web events data in Amazon Redshift.
  2. A variant optimized for SQL Runner SQL Runner is an open source app that makes it easy to execute SQL statements programmatically as part of a Snowplow data pipeline. This version of the model includes all the SQL queries that make up the model, as well as a playbook you can use to update the tables whenever there's new data.
  3. A variant optimized for Looker Looker is a modern BI tool that can be used to model and explore Snowplow data. Use this version to quickly set up a Looker Block with the Snowplow web data model.

1.1 Model structure

The web model aggregates Snowplow atomic data to to a page view level, then aggregates that data to a session and user level, to produce three derived tables (page_views, sessions, users):

model-structure

2. Requirements

To use this model you must have already set up Snowplow data tracking on your website (using the JavaScript tracker v2.6.0+), to track standard page views, with activity tracking enabled to set to send page pings - the model presumes 10 second intervals between page pings, so if your setup is different you will need to amend the events-time step. You will also need to have enabled the web page context.

To amend the SQL to match your page ping settings, replace the first two 10s in line 34 of the events-time step with the heartBeat setting of your JavaScript tracker, and the last one with the minimumVisitLength setting.

For example, if my page ping settings are snowplow_name_here('enableActivityTracking', 30, 20);, this line becomes:

20 * COUNT(DISTINCT(FLOOR(EXTRACT(EPOCH FROM ev.derived_tstamp)/20))) - 30 AS time_engaged_in_s

In the sql-runner variant of the model, this can be achieved by setting the playbook variables:

first_ping should reflect the minimumVisitLength argument (30 in our example above)

heartbeat should reflect the heartBeat argument (20 in our example above).

Find more information on how to set up Snowplow tracking via the JavaScript tracker.

Find more information on the web page context.

Find more information on activity tracking/page pings

2.1 Recommended requirements

To be able to run the model 'as is' and take full advantage of all options, you will need to be using version 2.6.0+ of the Snowplow JavaScript tracker and Snowplow 71+. The web page context must be enabled. You should also enable the performance timing context, and the ua parser, campaign attribution, IP lookups and referer parser enrichments.

2.2 Minimum requirements

You must be using version 2.5.0+ of the JavaScript tracker for the web page context to work. This context is required and you will not be able to use the model without it.

The performance timing context, and the referer parser, ua parser and campaign attribution enrichments are optional. You will still be able to use the model without them but you will have to comment out the relevant parts from the SQL code.

The user agent utils enrichment has been depracated but if you have enabled it manually, you can include an additional filter (currently commented out) on the br_family field which is populated by this enrichment.

By default, the model uses the more accurate derived_tstamp which makes allowance for inaccurate device clocks. This timestamp is only available in version 2.6.0+ of the JavaScript tracker and Snowplow 71+. If you are using older versions, you will need to change the relevant timestamps in the SQL files.

3. Understanding the individual fields

In order to analyse Snowplow data, it is important to understand how it is structured.

3.1 Page Views table

3.1.1 User fields

Field Type Description Example
user_custom_id text Unique ID set by business, user_id atomic field 'jon.doe@email.com'
user_snowplow_domain_id text User ID set by Snowplow using 1st party cookie 'bc2e92ec6c204a14'
user_snowplow_crossdomain_id text User ID set by Snowplow using 3rd party cookie 'ecdff4d0-9175-40ac-a8bb-325c49733607'

3.1.2 Session fields

Field Type Description Example
session_id text A visit / session identifier 'c6ef3124-b53a-4b13-a233-0088f79dcbcb'
session_index int A visit / session index 3

session_index is the number of the current user session. For example, an event occurring during a user's first session would have session_index set to 1.

3.1.3 Page view fields

Field Type Description Example
page_view_id text UUID 'e7c41b46-f449-49a0-a8ef-9a4037944004'
page_view_index int A page view index 3
page_view_in_session_index int A page view index within a single session 1

page_view_index is the number of the current page view. For example, if this is the second page viewed by the user, page_view_index will be set to 2.

page_view_in_session_index is the number of the current page view within the current session. Consider the example of a user who views two pages on their first visit and three more on their second visit. The second page view of the second visit will be their fourth overall. Hence page_view_index will be set to 4 but page_view_in_session_index will be set to 2.

3.1.4 Time fields

3.1.4.1 Core time fields
Field Type Description Example
page_view_start timestamp Timestamp for the start of the page view on your preferred clock '2015-11-17 04:13:53.000000'
page_view_end timestamp Timestamp for the end of the page view on your preferred clock '2015-11-17 04:16:03.997000'
page_view_start_local timestamp Timestamp for the start of the page view on the user's clock '2015-11-17 15:13:53.000000'
page_view_end_local timestamp Timestamp for the end of the page view on the user's clock '2015-11-17 15:16:03.000000'

The timestamps are calculated by default from the derived_tstamp. Change this in the SQL code if you want to use a different timestamp.

The time zone for page_view_start and page_view_end is set in the SQL code. The time zone for page_view_start_local and page_view_end_local is derived from the time zone of the OS.

3.1.4.2 Optional time fields

As well as the core timestamp fields, you can also have a variety of derived time fields. Some examples are already included in the code for this model. They are:

Field Type Description Example
page_view_time text Time derived from page_view_start '2015-11-17 04:13:53'
page_view_minute text Time derived from page_view_start, truncated to the minute '2015-11-17 04:13'
page_view_hour text Time derived from page_view_start, truncated to the hour '2015-11-17 04'
page_view_date text Date derived from page_view_start '2015-11-17'
page_view_week text Year, month and week number derived from page_view_start '2015-11-16'
page_view_month text Year and month derived from page_view_start '2015-11'
page_view_quarter text Quarter derived from page_view_start '2015-10'
page_view_year int Year derived from page_view_start 2015
page_view_local_time text Time derived from page_view_start_local '2015-11-17 15:13:53'
page_view_local_time_of_day text Hour and minute derived from page_view_start_local '15:13'
page_view_local_hour_of_day int Hour derived from page_view_start_local 15
page_view_local_day_of_week text Day of week derived from page_view_start_local, on a scale of 1 to 7 '4'
page_view_local_day_of_week_index int Day of week derived from page_view_start_local, on a scale of 0 to 6 0

page_view_quarter indicates the quarter by the starting month, so '2015-10' is the 'October quarter', or 'Q4'.

3.1.5 Engagement fields

Field Type Description Example
time_engaged_in_s int Time spent by the user on the page 70
time_engaged_in_s_tier text Range of time spent by the user on the page '60s or more'
horizontal_pixels_scrolled int Distance the user scrolled horizontally in pixels 0
vertical_pixels_scrolled int Distance the user scrolled vertically in pixels 1948
horizontal_percentage_scrolled int Percentage of page scrolled horizontally 100
vertical_percentage_scrolled int Percentage of page scrolled vertically 52
vertical_percentage_scrolled_tier text Range of percentage of page scrolled vertically '50% to 74%'
user_bounced boolean Did the user bounce? FALSE
user_engaged boolean Did the user engage with the page? TRUE

time_engaged_in_s assumes page_pings have been set to fire every 10 seconds. Change the setting in the SQL code to match your own settings.

user_bounced and user_engaged are derived from time_engaged_in_s and vertical_percentage_scrolled. Users who engage with the page for 0 seconds are counted as 'bounced'. Users who engage with the page for 30 seconds or more and scroll to see at least 25% of the content on the page are counted as 'engaged'.

3.1.6 Page fields

Field Type Description Example
page_url text The page URL 'http://www.example.com'
page_url_scheme text Scheme aka protocol 'https'
page_url_host text Host aka domain 'snowplowanalytics.com'
page_url_port int Port if specified, 80 if not 80
page_url_path text Path to page '/product/index.html'
page_url_query text Querystring 'id=GTM-DLRG'
page_url_fragment text Fragment aka anchor '4-conclusion'
page_title text Web page title 'Using ChartIO to visualize and interrogate Snowplow data - Snowplow Analytics'
page_width int The page's width in pixels 1024
page_height int The page's height in pixels 3000

3.1.7 Referer fields

Field Type Description Example
referer_url text The referer URL 'www.google.de/'
referer_url_scheme text Referer scheme 'http'
referer_url_host text Referer host 'www.bing.com'
referer_url_port int Referer port 80
referer_url_path text Referer page path '/images/search'
referer_url_query text Referer URL querystring 'q=psychic+oracle+cards'
referer_url_fragment text Referer URL fragment 'process-that-data'
referer_medium text Type of referer 'search', 'internal'
referer_source text Name of referer if recognised 'Bing images'
referer_term text Keywords if source is a search engine 'psychic oracle cards'

These fields require the the referer parser enrichment. Find more information about it here

3.1.8 Marketing fields

Field Type Description Example
marketing_medium text Type of traffic source 'cpc', 'affiliate', 'organic', 'social'
marketing_source text The company / website where the traffic came from 'Google', 'Facebook'
marketing_term text Any keywords associated with the referer 'new age tarot decks'
marketing_content text The content of the ad. (Or an ID so that it can be looked up.) '13894723'
marketing_campaign text The campaign ID 'diageo-123'
marketing_click_id text The click ID 'ac3d8e459'
marketing_network text The ad network to which the click ID belongs 'DoubleClick'

These fields require the campaign attribution enrichment. Find more information about it here.

3.1.9 Geo fields

Field Type Description Example
geo_country text ISO 3166-1 code for the country the visitor is located in 'GB', 'US'
geo_region text ISO-3166-2 code for country region the visitor is in 'I9', 'TX'
geo_region_name text Visitor region name 'Florida'
geo_city text City the visitor is in 'New York', 'London'
geo_zipcode text Postcode the visitor is in '94109'
geo_latitude text Visitor location latitude '37.443604'
geo_longitude text Visitor location longitude '-122.4124'
geo_timezone text Visitor timezone name 'Europe/London'

These fields require the IP lookups enrichment. Find more information about it here.

3.1.10 IP address fields

Field Type Description Example
ip_address text The IP address of the visitor '125.22.43.11'
ip_isp text Visitor's ISP 'FDN Communications'
ip_organization text Organization associated with the visitor's IP address - defaults to ISP name if none is found 'Bouygues Telecom'
ip_domain text Second level domain name associated with the visitor's IP address 'nuvox.net'
ip_net_speed text Visitor's connection type 'Cable/DSL'

These fields require the IP lookups enrichment. Find more information about it here.

3.1.11 Browser fields

Field Type Description Example
browser text The name and version of the visitor's browser 'Chrome 49.0.2623'
browser_name text The name, or family, of the visitor's browser 'Chrome'
browser_major_version int Browser major version 49
browser_minor_version int Browser minor version 0
browser_build_version int Browser build version 2623
browser_engine text Browser rendering engine 'WEBKIT'
browser_window_width int Viewport width 1433
browser_window_height int Viewport height 567
browser_language text Language the browser is set to 'en-GB'

browser, browser_name, browser_major_version, browser_minor_version and browser_build_version require the ua parser enrichment. Find more information about it here.

3.1.12 OS fields

Field Type Description Example
os text The name and version of the visitor's OS 'Mac OS X 10.10.1'
os_name text The name of the visitor's OS 'Mac OS X'
os_major_version int OS major version 10
os_minor_version int OS minor version 10
os_build_version int OS build version 1
os_manufacturer text OS manufacturer 'Apple Inc.'
os_timezone text Time zone the OS is set to 'America/Chicago'

All OS fields except os_timezone and os_manufacturer require the ua parser enrichment. Find more information about it here.

3.1.13 Device fields

Field Type Description Example
device text Device model 'XT1022'
device_type text Type of device 'Computer'
device_is_mobile boolean Is the device mobile? TRUE

device requires the ua parser enrichment. Find more information about it here.

3.1.14 Performance timing fields

Field Type Description Example
redirect_time_in_ms int Time to redirect the visitor 6864
unload_time_in_ms int Time to complete the unload event 1
app_cache_time_in_ms int Time to fetch resource from relevant application cache 3
dns_time_in_ms int Time to complete domain lookup 1
tcp_time_in_ms int Time to establish connection 380
request_time_in_ms int Time between the user agent sending a request and receiving the first byte of the response 747
response_time_in_ms int Time to complete the response 633
processing_time_in_ms int Processing time 10849
dom_loading_to_interactive_time_in_ms int Time for the current document readiness to change from 'loading' to 'interactive' 3996
dom_interactive_to_complete_time_in_ms int Time for the current document readiness to change from 'interactive' to 'complete' 6853
onload_time_in_ms int Time to complete the load event 8
total_time_in_ms int Total time from navigation start to load event completion 18470

These fields require the performanceTiming context. Find more information about it here.

All times are measured in milliseconds.

3.1.15 Other fields

Field Type Description Example
app_id text Application ID 'snowplowweb'

The application ID is used to distinguish different applications that are being tracked by the same Snowplow stack, eg 'production' versus 'dev'. This is hardcooded in the tracker's appId argument.

3.2 Sessions table

3.2.1 User fields

Field Type Description Example
user_custom_id text Unique ID set by business, user_id 'jon.doe@email.com'
user_snowplow_domain_id text User ID set by Snowplow using 1st party cookie 'bc2e92ec6c204a14'
user_snowplow_crossdomain_id text User ID set by Snowplow using 3rd party cookie 'ecdff4d0-9175-40ac-a8bb-325c49733607'

3.2.2 Session fields

Field Type Description Example
session_id text A visit / session identifier 'c6ef3124-b53a-4b13-a233-0088f79dcbcb'
session_index int A visit / session index 3

session_index is the number of the current user session. For example, an event occurring during a user's first session would have session_index set to 1.

3.2.3 Time fields

3.2.3.1 Core time fields
Field Type Description Example
session_start timestamp Timestamp for the start of the session on your preferred clock '2015-07-20 10:35:22.441000'
session_end timestamp Timestamp for the end of the session on your preferred clock '2015-07-20 12:38:17.238000'
session_start_local timestamp Timestamp for the start of the session on the user's clock '2015-07-20 15:05:22.441000'
session_end_local timestamp Timestamp for the end of the session on the user's clock '2015-07-20 17:08:17.238000'

session_start corresponds to the first value of page_view_start within that session; and session_end corresponds to the last page_view_end within that session.

The timestamps are calculated by default from the derived_tstamp. Change this in the SQL code if you want to use a different timestamp.

The time zone for session_start and session_end is set in the SQL code when defining the page_view_start and page_view_end fields. The time zone for session_start_local and session_end_local is derived from the time zone of the OS (by way of page_view_start_local and page_view_end_local).

3.2.3.2 Optional time fields

As well as the core timestamp fields, you can also have a variety of derived time fields. Some examples are already included in the code for this model. They are:

Field Type Description Example
session_time text Time derived from session_start '2015-09-03 22:22:02'
session_minute text Time derived from session_start, truncated to the minute '2015-09-03 22:22'
session_hour text Time derived from session_start, truncated to the hour '2015-09-03 22'
session_date text Date derived from session_start '2015-09-03'
session_week text Year, month and week number derived from session_start '2015-08-31'
session_month text Year and month derived from session_start '2015-09'
session_quarter text Quarter derived from session_start '2015-07'
session_year int Year derived from session_start 2015
session_local_time text Time derived from session_start_local '2015-11-17 15:13:53'
session_local_time_of_day text Hour and minute derived from session_start_local '15:13'
session_local_hour_of_day int Hour derived from session_start_local 15
session_local_day_of_week text Day of week derived from session_start_local, on a scale of 1 to 7 '4'
session_local_day_of_week_index int Day of week derived from session_start_local, on a scale of 0 to 6 0

session_quarter indicates the quarter by the starting month, so '2015-07' is the 'July quarter', or 'Q3'.

3.2.4 Engagement fields

Field Type Description Example
page_views int Number of page views within this session 8
bounced_page_views int Number of page views within this session where the visitor bounced 0
engaged_page_views int Number of page views within this session where the visitor engaged 1
time_engaged_in_s int Time spent by the user on this visit 320
time_engaged_in_s_tier text Range of time spent by the user on this visit '240s or more'
user_bounced boolean Did the user bounce on their first page view in this session? FALSE
user_engaged boolean Did the user engage with the website in this session? TRUE

time_engaged_in_s assumes page_pings have been set to fire every 10 seconds. Change the setting in the SQL code to match your own settings.

user_bounced and user_engaged are derived from page_views and time_engaged_in_s. Users who only viewed 1 page and bounced on that page view, are counted as 'bounced'. Users who viewed more than 2 pages and spent at least 60 seconds in the same session are counted as 'engaged'.

3.2.5 First page fields

Field Type Description Example
first_page_url text The page URL 'http://www.example.com'
first_page_url_scheme text Scheme aka protocol 'https'
first_page_url_host text Host aka domain 'snowplowanalytics.com'
first_page_url_port int Port if specified, 80 if not 80
first_page_url_path text Path to page '/product/index.html'
first_page_url_query text Querystring 'id=GTM-DLRG'
first_page_url_fragment text Fragment aka anchor '4-conclusion'
first_page_title text Web page title 'Using ChartIO to visualize and interrogate Snowplow data - Snowplow Analytics'

These fields are derived from the page fields of the Page Views table to represent the first page view within the session.

3.2.6 Referer fields

Field Type Description Example
referer_url text The referer URL 'www.google.de/'
referer_url_scheme text Referer scheme 'http'
referer_url_host text Referer host 'www.bing.com'
referer_url_port int Referer port 80
referer_url_path text Referer page path '/images/search'
referer_url_query text Referer URL querystring 'q=psychic+oracle+cards'
referer_url_fragment text Referer URL fragment 'process-that-data'
referer_medium text Type of referer 'search', 'internal'
referer_source text Name of referer if recognised 'Bing images'
referer_term text Keywords if source is a search engine 'psychic oracle cards'

These fields carry information about the first page view within the session.

3.2.7 Marketing fields

Field Type Description Example
marketing_medium text Type of traffic source 'cpc', 'affiliate', 'organic', 'social'
marketing_source text The company / website where the traffic came from 'Google', 'Facebook'
marketing_term text Any keywords associated with the referer 'new age tarot decks'
marketing_content text The content of the ad. (Or an ID so that it can be looked up.) '13894723'
marketing_campaign text The campaign ID 'diageo-123'
marketing_click_id text The click ID 'ac3d8e459'
marketing_network text The ad network to which the click ID belongs 'DoubleClick'

These fields carry information about the first page view within the session.

These fields require the campaign attribution enrichment. Find more information about it here.

3.2.8 Geo fields

Field Type Description Example
geo_country text ISO 3166-1 code for the country the visitor is located in 'GB', 'US'
geo_region text ISO-3166-2 code for country region the visitor is in 'I9', 'TX'
geo_region_name text Visitor region name 'Florida'
geo_city text City the visitor is in 'New York', 'London'
geo_zipcode text Postcode the visitor is in '94109'
geo_latitude text Visitor location latitude '37.443604'
geo_longitude text Visitor location longitude '-122.4124'
geo_timezone text Visitor timezone name 'Europe/London'

These fields require the IP lookups enrichment. Find more information about it here.

3.2.9 IP address fields

Field Type Description Example
ip_address text The IP address of the visitor '125.22.43.11'
ip_isp text Visitor's ISP 'FDN Communications'
ip_organization text Organization associated with the visitor's IP address - defaults to ISP name if none is found 'Bouygues Telecom'
ip_domain text Second level domain name associated with the visitor's IP address 'nuvox.net'
ip_net_speed text Visitor's connection type 'Cable/DSL'

3.2.10 Browser fields

Field Type Description Example
browser text The name and version of the visitor's browser 'Chrome 49.0.2623'
browser_name text The name, or family, of the visitor's browser 'Chrome'
browser_major_version int Browser major version 49
browser_minor_version int Browser minor version 0
browser_build_version int Browser build version 2623
browser_engine text Browser rendering engine 'WEBKIT'
browser_language text Language the browser is set to 'en-GB'

browser, browser_name, browser_major_version, browser_minor_version and browser_build_version require the ua parser enrichment. Find more information about it here.

3.2.11 OS fields

Field Type Description Example
os text The name and version of the visitor's OS 'Mac OS X 10.10.1'
os_name text The name of the visitor's OS 'Mac OS X'
os_major_version int OS major version 10
os_minor_version int OS minor version 10
os_build_version int OS build version 1
os_manufacturer text OS manufacturer 'Apple Inc.'
os_timezone text Time zone the OS is set to 'America/Chicago'

All OS fields except os_timezone and os_manufacturer require the ua parser enrichment. Find more information about it here.

3.2.12 Device fields

Field Type Description Example
device text Device model 'XT1022'
device_type text Type of device 'Computer'
device_is_mobile boolean Is the device mobile? TRUE

device requires the ua parser enrichment. Find more information about it here.

3.2.13 Other fields

Field Type Description Example
app_id text Application ID 'snowplowweb'

The application ID is used to distinguish different applications that are being tracked by the same Snowplow stack, eg 'production' versus 'dev'. This is hardcooded in the tracker's appId argument.

3.3 Users table

3.3.1 User fields

Field Type Description Example
user_custom_id text Unique ID set by business 'jon.doe@email.com'
user_snowplow_domain_id text User ID set by Snowplow using 1st party cookie 'bc2e92ec6c204a14'
user_snowplow_crossdomain_id text User ID set by Snowplow using 3rd party cookie 'ecdff4d0-9175-40ac-a8bb-325c49733607'

3.3.2 Time fields

3.3.2.1 Core time fields
Field Type Description Example
first_session_start timestamp Timestamp for the start of the user's first session on your preferred clock '2015-07-20 10:35:22.441000'
first_session_start_local timestamp Timestamp for the start of the user's first session on the user's clock '2015-07-20 15:05:22.441000'
last_session_end timestamp Timestamp for the end of the user's last session on your preferred clock '2015-07-20 12:38:17.238000'

first_session_start corresponds to the first value of session_start for this user from the Sessions table; and last_session_end corresponds to the last session_end for this user from the Sessions table.

The timestamps are calculated by default from the derived_tstamp. Change this in the SQL code if you want to use a different timestamp.

The time zone for first_session_start and last_session_end is set in the SQL code when defining the page_view_start and page_view_end fields. The time zone for first_session_start_local is derived from the time zone of the OS (by way of page_view_start_local).

3.3.2.2 Optional time fields

As well as the core timestamp fields, you can also have a variety of derived time fields. Some examples are already included in the code for this model. They are:

Field Type Description Example
first_session_time text Time derived from first_session_start '2015-09-03 22:22:02'
first_session_minute text Time derived from first_session_start, truncated to the minute '2015-09-03 22:22'
first_session_hour text Time derived from first_session_start, truncated to the hour '2015-09-03 22'
first_session_date text Date derived from first_session_start '2015-09-03'
first_session_week text Year, month and week number derived from first_session_start '2015-08-31'
first_session_month text Year and month derived from first_session_start '2015-09'
first_session_quarter text Quarter derived from first_session_start '2015-07'
first_session_year int Year derived from first_session_start 2015
first_session_local_time text Time derived from first_session_start_local '2015-11-17 15:13:53'
first_session_local_time_of_day text Hour and minute derived from first_session_start_local '15:13'
first_session_local_hour_of_day int Hour derived from first_session_start_local 15
first_session_local_day_of_week text Day of week derived from first_session_start_local, on a scale of 1 to 7 '4'
first_session_local_day_of_week_index int Day of week derived from first_session_start_local, on a scale of 0 to 6 0

first_session_quarter indicates the quarter by the starting month, so '2015-07' is the 'July quarter', or 'Q3'.

3.3.3 Engagement fields

Field Type Description Example
page_views int Number of page views by this user 8
sessions int Number of sessions / visits by this user 2
time_engaged_in_s int Time spent by the user across all visits 320

time_engaged_in_s assumes page_pings have been set to fire every 10 seconds. Change the setting in the SQL code to match your own settings.

3.3.4 First page fields

Field Type Description Example
first_page_url text The page URL 'http://www.example.com'
first_page_url_scheme text Scheme aka protocol 'https'
first_page_url_host text Host aka domain 'snowplowanalytics.com'
first_page_url_port int Port if specified, 80 if not 80
first_page_url_path text Path to page '/product/index.html'
first_page_url_query text Querystring 'id=GTM-DLRG'
first_page_url_fragment text Fragment aka anchor '4-conclusion'
first_page_title text Web page title 'Using ChartIO to visualize and interrogate Snowplow data - Snowplow Analytics'

These fields are derived from the page fields of the Page Views table to represent the first page view for this user.

3.3.5 Referer fields

Field Type Description Example
referer_url text The referer URL 'www.google.de/'
referer_url_scheme text Referer scheme 'http'
referer_url_host text Referer host 'www.bing.com'
referer_url_port int Referer port 80
referer_url_path text Referer page path '/images/search'
referer_url_query text Referer URL querystring 'q=psychic+oracle+cards'
referer_url_fragment text Referer URL fragment 'process-that-data'
referer_medium text Type of referer 'search', 'internal'
referer_source text Name of referer if recognised 'Bing images'
referer_term text Keywords if source is a search engine 'psychic oracle cards'

These fields carry information about the first page view by this user.

3.3.6 Marketing fields

Field Type Description Example
marketing_medium text Type of traffic source 'cpc', 'affiliate', 'organic', 'social'
marketing_source text The company / website where the traffic came from 'Google', 'Facebook'
marketing_term text Any keywords associated with the referer 'new age tarot decks'
marketing_content text The content of the ad. (Or an ID so that it can be looked up.) '13894723'
marketing_campaign text The campaign ID 'diageo-123'
marketing_click_id text The click ID 'ac3d8e459'
marketing_network text The ad network to which the click ID belongs 'DoubleClick'

These fields carry information about the first page view by this user.

These fields require the campaign attribution enrichment. Find more information about it here.

3.3.7 Other fields

Field Type Description Example
app_id text Application ID 'snowplowweb'

The application ID is used to distinguish different applications that are being tracked by the same Snowplow stack, eg 'production' versus 'dev'. This is hardcooded in the tracker's appId argument.

About

SQL data model for working with Snowplow web data. Supports Redshift and Looker. Snowflake and BigQuery coming soon

Topics

Resources

Stars

Watchers

Forks

Packages

No packages published