Quickstart

Access

Decode GA4 is distributed via BigQuery Analytics Hub, available in all regions and multi-regions. Currently access is only on a private-offer basis, with billing via your existing linked billing account on a flat rate per TiB processed basis. Contact ga4@decodedata.io to request private offer terms.

Pre-requisites

In order to deploy Decode GA4, you first need to create a Google Cloud Storage (GCS) bucket. This will be the storage location for compressed, transformed events data. Creating a new bucket can be achieved through the Google Cloud user interface, command line or client libraries, and the executing user or service account must have storage.objects.create permissions on the bucket.

Note that GCS bucket names need to be globally unique, so it makes sense to use a common, company-specific prefix (e.g. company-name-project-name) and then e.g. decode-ga4 as a suffix.

Installation

Setting the configuration options and executing the script will install decode-ga4 onto your GA4 property.

Configuration

The following options are required:

Deployment

Decode GA4 is installed by copying the following script to your BigQuery console, updating the configuration options and running the updated script.

Note that all configuration options are set using the options JSON variable.

DECLARE options JSON;

SET options = JSON """
    {
        "ga4_dataset_id": "project_id.ga4_dataset_id",
        "gcs_bucket_name": "bucketname"
    }
    """;

EXECUTE IMMEDIATE (
    SELECT `decode-ga4.eu.deploy_installer`(options)
    );

Executing this script will create the BigQuery dataset decode_analytics_XXXXXXX and then profile your historic data in GA4.

It will then build and deploy a set of custom functions - derived from the historic profile - into the dataset, including the stored procedure RUN(). This is the only function you should ever need to call, as it will both execute the initial data export and subsequent incremental runs.

It can also to be used to trigger a full data refresh or specific date range data refresh. Run the function using the following syntax:

CALL `project_id.decode_analytics_XXXXXXX.RUN()`

This can be invoked manually, on a schedule or on an event-driven basis. Note that if this is invoked and there is no new data detected, only negligible metadata query costs are incurred.