Launch an automated BigQuery machine learning model in minutes.
Fill in the fields below with details about your Google Cloud and Google Analytics accounts. Upon completing all fields, click download to generate a customized JSON file. Upload the file to CRMint to automate your pipelines.
CRMint Setup
CRMint is a Google open source dataflow platform that orchestrates the
pipeline generated by Instant BQML/Vertex.
It has simple and intuitive web UI that gives you full control and transparency into the underlying data processing jobs.

It has simple and intuitive web UI that gives you full control and transparency into the underlying data processing jobs.
keyboard_arrow_down
CRMint can be deployed via App Engine or Cloud Run.
We suggest using App Engine since deploying with Cloud Run requires a Cloud Organization (through Workspace or Cloud Identity).
If you're unsure about this or haven't done it yet, it's best to stick with CRMint on App Engine.
We suggest using App Engine since deploying with Cloud Run requires a Cloud Organization (through Workspace or Cloud Identity).
If you're unsure about this or haven't done it yet, it's best to stick with CRMint on App Engine.
Did you run the command below in the Google Cloud Shell terminal, yet?
To activate the Cloud Shell, visit your Google Cloud Console and look for the "Activate Cloud Shell" button in the top right corner. It looks like a small terminal icon.

keyboard_arrow_down
This is a boosted tree regression model.
Trained on last 12 months of data updated weekly.
Scores updated daily for visitors.
This model contains the following features:
The model output is a score between 0 and 1000 (or a given multiplier).
Higher scores denote higher likelihood.
Trained on last 12 months of data updated weekly.
Scores updated daily for visitors.
This model contains the following features:
- # Different Days Visited
- Average Session Depth
- Bounce Rate
- Browser
- Day of week
- Distinct Regions
- Medium
- Mobile
- Pageviews
- Region
- Total Sessions
The model output is a score between 0 and 1000 (or a given multiplier).
Higher scores denote higher likelihood.






keyboard_arrow_down

keyboard_arrow_down

keyboard_arrow_down


Your Google Analytics Dataset ought to be a Custom Dataset, shared with your BigQuery Enabled View at least using Query Time Import where the Key is the GA Client ID or User ID custom dimension & the Imported Data is the custom dimension placeholder for the propensity score.
keyboard_arrow_down
Google Analytics audiences can be published to Google Ads and DV360.
Cloud Storage Bucket Acknowledgement
Predictions from the pipeline are output to Cloud Storage prior to Google Analytics Data Import. You must create the Cloud Storage bucket first in order to upload files to it during the pipeline.

Did you create a bucket in Cloud Storage in the Google Cloud Platform Project, my-sample-project-191923, named crmint-bqml-202201219, yet?
Permissions Acknowledgement
Did you add editor permissions for the Service Account, my-sample-project-191923@appspot.gserviceaccount.com, to the Google Analytics property under Property or Account Access Management, yet?
Google Cloud's service account must be added with editor permissions to the Google Analytics property. This can be done at the Account or Property level.
