We are revamping our site, and will be live with a new version by next month.

User Level interaction with BigQuery and Firebase

Sometimes, it’s common to be in a place – where you don’t get the best of the analytics tool. However, the conventional ratios just don’t cut in. Ability to understand user-level interactions help you understand the point of friction.

Tools Needed​ – Firebase Blaze Plan.
Prerequisites​: Firebase SDK integrated into the app.

Step 1 –

Go to settings > integrations and click on ​the link (Refer to the screenshot below).

Step 2 –

Please select the attributes that you’d like to pass to BigQuery from firebase. Feel free to add advertising identifiers in the export.

Step 3

Please make sure that the below option in the firebase data flow is turned ON.

Step 4

Once you do that, the data will start flowing into your Big Query table – the table name has a prefix ​_events.


Handling data storage compliance – only applicable if for any reason you’d like to specify the country where the data can be stored.
If you like, you can create a different dataset – in case you plan on storing data into some other location offered by Big Query. (In case your data handling and storage policy requires you to store data in a particular country). Else, by default, US-based data center stores the table.

Optional Step – Visualize the data into visualization software.

Optimizing Costs

  • Use Blaze calculator to estimate the costing.
  • Querying raw data – ​Avoid Select * in the query, use query estimators – by using –dry_run flag.
  • Monitor & Predict​ – I highly recommend you to create another big query dataset with the cloud billing data. The cloud interface is not too user friendly for segmenting the data (in case you are working with multiple apps, datasets – and want to segment the cost by a particular property). Creating a billing export (Label the project Name and dataset name to any name of your choice.)
Related :   Mobile App Target Audience | Target Audience For App 2023

Special Note for apps with over 10M users – In order to control or estimate costs, we run a machine learning model to estimate the pricing. So, you can create a data source using a custom query to train/validate a linear regression model. Although the costs are usually predictable, this would save you from any shocks arising out of exorbitant GCP bills.