The results of an attribution model calculation are stored in a Google BigQuery dataset:

Attribution_tables_in_BigQuery.png

By default, the dataset contains three tables with intermediate attribution results (actions_probabilities_, and purchases_) and a partitioned table values_ with the final calculation results data. Up until August 1, 2018, the values' data is being duplicated to non-partitioned values_ tables as well.

The name of each table ends with a prefix denoting the end date of the transaction window selected during the attribution model calculation. For example, “actions_20160601”.

The values table contains the information about sessions that were attributed and the value assigned to them and the information about the attributed values that were created as the result of regular model calculations grouped by months.

The actions_ table contains information about all sessions and user interactions with your site. The probabilities table is based on the actions_ table and contains data about the probability of users getting throughout each step of the conversion funnel. The purchases_ table contains data on conversions. The data in these three tables is intermediate and is used only to deliver you the attribution calculation results stored in the values table.

The values table schema

An example of a values table in Google BigQuery

Field name Data type Description
session_id STRING Session ID
value_from_sid STRING  
source STRING Source
medium STRING Medium
campaign STRING Campaign
date STRING The latest date within the calculation period in the “YYYYMMDD” format. Note: this is not the session start time. For the session start time, see the field time.
keyword STRING Ad campaign keyword
adContent STRING Ad campaign contents
user_id STRING User ID
active_step INTEGER The order number of the deepest step in the funnel the user visited during the session
product_id STRING Product ID (if a model attributed the value of a product, not of a transaction)
transaction_id STRING Transaction ID
time INTEGER Session start time in the unix time format with microseconds
value FLOAT Total value of a session
revenue FLOAT Revenue from a product
data_source STRING Source of information about session: OWOX BI Pipeline, Google Analytics export to BigQuery or custom transactions data source
user_type  STRING Type of a customer: new or returned.
device  STRING Device category: desktop only, tablet only, mobile only or cross-device.
region  STRING Region
lag_active_step INTEGER The order number of next to last funnel step. If a user made no actions in the funnel, the field will get a "-1" value.

The probabilities table schema

Field name Data type Description
device STRING User device category
region STRING Region
user_type STRING User type
lag_active_step INTEGER The order number of next to last funnel step. If a user made no actions in the funnel, the field will get a "-1" value.
probability FLOAT Step probability that was used in the model calculation
reg_usr_dev_conf_interval FLOAT Confidence interval in the segment grouped by all the dimensions: Device Category, User Type, and Region
reg_usr_dev_probability FLOAT Probability in the segment grouped by all the dimensions
reg_usr_dev_lag_actions INTEGER Number of actions on the previous step in the segment grouped by all the dimensions
reg_usr_dev_actions INTEGER Number of actions on the step in the segment grouped by all the dimensions
user_dev_conf_interval FLOAT Confidence interval in the segment grouped by the dimensions: Device Category and User Type
user_dev_lag_actions INTEGER Number of actions on the previous step in the segment grouped by the dimensions: Device Category and User Type
user_dev_actions INTEGER Number of actions on the step in the segment grouped by the dimensions: Device Category and User Type
user_dev_probability FLOAT Probability in the segment grouped by the dimensions: Device Category and User Type
dev_conf_interval FLOAT Confidence interval in the segment grouped by Device Category
dev_lag_actions INTEGER Number of actions on the previous step in the segment grouped by Device Category
dev_actions INTEGER Number of actions on the step in the segment grouped by Device Category
dev_probability FLOAT Probability in the segment grouped by Device Category
global_actions INTEGER Total number of actions on the step
global_lag_actions INTEGER Total number of actions on the previous step
global_probability FLOAT Average probability without segmentation

The actions table schema

Field name Data type Description
session_id STRING Session ID
active_step INTEGER The order number of the deepest step in the funnel the user visited during the session
user_id STRING User ID
transaction_id STRING Transaction ID
time INTEGER Session start time in unix time format with microseconds
date STRING Session date in format “YYYY-MM-DD”
is_attributed BOOLEAN Shows if the session didn't get value because of the model settings.
False when the value was assigned to another source.
device STRING Device category
user_type STRING User type
region STRING Region
data_source STRING Source of the session data: OWOX BI Pipeline, BigQuery Export for Google Analytics 360 or a custom data source.
lag_active_step INTEGER The order number of next to last funnel step. If a user made no actions in the funnel, the field will get a "-1" value.
Was this article helpful?
0 out of 0 found this helpful
Have more questions? Submit a request

0 Comments

Please sign in to leave a comment.