BigQuery destination for batch exports

Batch exports can be used to export data to a BigQuery table.

Setting up BigQuery access

To set up the right permissions for a batch export targeting BigQuery, you need:

1. A Service Account.
2. A dataset which has permissions allowing the service account to access it.

Here's how to set these up so that the destination has access only to the dataset it needs:

1. Create a Service Account.

2. Create a key for the Service Account you created in the previous step.
3. Save the key file as JSON to upload it when configuring a batch export.

4. Create a role which has only the specific permissions the batch export requires:
   • bigquery.datasets.get
   • bigquery.jobs.create
   • bigquery.tables.create
   • bigquery.tables.get
   • bigquery.tables.list
   • bigquery.tables.updateData
   • (Optional, for mutable models) bigquery.tables.delete

This step can be skipped if using the built-in roles BigQuery Data Editor and BigQuery Job User in the steps that follow.

5. Grant the Service Account access to run jobs in your Google Cloud project. This can be done by granting the BigQuery Jobs User role or the role we created in the previous step on your project.

Navigate to IAM and click on Grant Access to arrive at this screen:

In the screenshot above, we have used a custom role named Testing PostHog BatchExports with the permissions listed in the previous step.

6. Create a dataset within a BigQuery project (ours is called BatchExports, but any name will do).
7. Use the Sharing and Add Principal buttons to grant access to your dataset with your Service Account created in step 1. Next, assign either the BigQuery Data Editor role or your custom role created in step 4 to provide permissions for the dataset access. Read the full instructions on granting access to the dataset in BigQuery if unclear.

In the screenshot below, we grant our Service Account access to the BatchExports data set and assign the Testing PostHog BatchExports role permissions for it.

8. All done! After completing these steps you can create a BigQuery batch export in PostHog and your data will start flowing from PostHog to BigQuery.

Models

This section describes the models that can be exported to BigQuery.

Events model

This is the default model for BigQuery batch exports. The schema of the model as created in BigQuery is:

Some fields can be either STRING or JSON type depending on whether the corresponding checkbox is marked or not when creating the batch export.

Persons model

The schema of the model as created in BigQuery is:

The BigQuery table will contain one row per (team_id, distinct_id) pair, and each pair is mapped to their corresponding person_id and latest properties. The properties field can be either STRING or JSON, depending on whether the corresponding checkbox is marked or not when creating the batch export.

How is the persons model kept up to date?

Exporting mutable data (like the persons model) requires executing a merge operation to apply new updates to existing rows. Executing a merge in BigQuery involves the following steps:

1. Creating a stage table.
2. Inserting new data into stage table.
3. Execute a merge operation between existing table and stage table. a. Any rows that match in the final table and for which the stage table's version is higher are updated. b. Any new rows not found in the final table are inserted.
4. Drop the stage table.

Besides the permissions required for exporting the events model, and since we need to clean-up a stage table, exporting the persons model requires also bigquery.tables.delete permissions.

Creating the batch export

1. Subscribe to data pipelines add-on in your billing settings if you haven't already.
2. Click Data pipelines in the navigation and go to the exports tab in your PostHog instance.
3. Click "Create export workflow".
4. Select BigQuery as the batch export destination.
5. Fill in the necessary configuration details.
6. Finalize the creation by clicking on "Create".
7. Done! The batch export will schedule its first run on the start of the next period.

BigQuery configuration

Configuring a batch export targeting BigQuery requires the following BigQuery-specific configuration values:

• Table ID: The ID of the destination BigQuery table. This is not the fully-qualified name of a table, so omit the dataset and project IDs. For example for the fully-qualified table name project-123:dataset:MyExportTable, use only MyExportTable as the table ID.
• Dataset ID: The ID of the BigQuery dataset which contains the destination table. Only the dataset ID is required, so omit the project ID if present. For example for the fully-qualified dataset project-123:my-dataset, use only my-dataset as the dataset ID.
• Google Cloud JSON key file: The JSON key file for your BigQuery Service Account to access your instance. Generated on Service Account creation. See here for more information.

Examples

These examples illustrate how to use the data from batch exports in BigQuery.

Requirements

Two batch exports need to be created:

• An events model batch export.
• A persons model batch export.

For the purposes of these examples, assume that these two batch exports have already been created and have exported some data to BigQuery in tables example.events and example.persons.

Example: Count unique persons that have triggered an event

The following query can be used to count the number of unique persons that have triggered events:

SELECT
  event,
  COUNT(persons.person_id) AS unique_persons_count
FROM
  example.events AS events
LEFT JOIN
  example.persons AS persons ON events.distinct_id = persons.distinct_id AND events.team_id = persons.team_id
WHERE
  persons.person_id IS NOT NULL
GROUP BY
  event
ORDER BY
  unique_persons_count DESC