Uploading ML user scoring into Google Analytics
This document describes how you can upload SegmentStream ML-based predictions about the probability to convert from Google BigQuery into Google Analytics.
Uploaded predictions can be used to build retargeting audiences, for example:
- Users that have >10% likelihood to convert in the next X days;
- Users that have <1% likelihood to convert within the next X days;
These retargeting audiences can be then transferred to Google Ads to improve bidding strategy or building wide look-a-like audiences for better targeting.
Before you begin
- Make sure that BigQuery is connected to SegmentStream.
- Make sure that table with user probabilities is created in Google BigQuery (usually is done by SegmentStream onboarding team).
Creating custom dimension for Google clientId
- In the Admin section, select the account and property you want to import data to and select Custom Definitions > Custom Dimensions:
- Create custom dimension on user level. You can name it anyway you want.
- Use Custom Dimension index while setting up Data Import
- Click Save
Creating custom metric for conversion probability
- In the Admin section, select the account and property you want to import data to and select Custom Definitions > Custom Metrics:
- Create custom metric with Integer “Formatting Type”. You can name it anyway you want.
- Use Custom Metric index while setting up Data Import
- Click Save
Setting up Google clintId collection into Custom Dimension
Creating data set for user data
In the Admin section, select the account and property you want to import data to and select Data Import:
Click Create to create a new Data Set:
Select Data Set type as User Data and click Continue: Image here
Enter the Data Set Name and choose the Enabled Views. Cost data will be available only for the selected views: Image here
(draft) Choose all the dimensions and metrics as shown in the screenshot, select Import Behavior as Overwrite and click Save: Image here
Enabling Google Analytics data destination
- Go to SegmentStream admin panel ▸.
- Open Data Destination ▸ Add Data Destination.
- Select Google Analytics.
- Click Authenticate with Google and go through the authentication flow.
- Select your Google Analytics Account and Property using dropdowns.
- Choose Data Set Type as User Data
- Select Data Set which you would like to use for the user data import.
- Define BigQuery source table name where you user scroing data is stored.
- Map you BigQuery columns with Google clientId and scoring value to proper Google Analytics custom dimensions and metrics. (Image here)
- Click Save.
This document was last updated on August 4, 2020. If you notice any gaps, outdated information or simply want to leave some feedback to help us improve our documentation, please let us know!