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

BigQuery raw data can be quite useful for analysing your user’s data and mine some important patterns. Some of the most interesting use cases are:

Exploring Mobile Analytics Use Cases with BigQuery

1. Identify if the same user has downloaded your app multiple times, and on multiple devices.
2. Identify ad fraud, by capturing some specific identifiers.
3. Identify if the user has posted review on your app. (Android only, you need Google login or capture e-mail with some identifiers).
4. Connect user with the FCM token to build your own customer engagement tool.
5. Stay a step ahead of your crash numbers on Google Play Console. We’ve tried reconciling numbers, and they match.
6. Send purchase identifier and then use Google Play Billing Api to get subscription status – very handy – if you don’t have your own CRM, and recommended – even if you do.

Point 3, 4, 6 – requires passing some identifiers, but other than that – all the data is available in BQ.

What makes BQ preferable compared to other tools?

Lots of useful data, and some very special events – like uninstall event. Firebase transfer gets you exact timestamp of uninstall – something that you don’t get on other tools. Sure, you can write a cloud function and send that event to your preferred tool – but this requires engineering efforts. You get this piece of data, ready to use, in BQ.

While other analytics tools are brilliant, BQ gives you access to raw database, and you can create pipelines with it.

Related :   Simplified Data Storage with Apache Parquet Compression

Powerful Workflows With BQ – Use Cases (Intraday table – Blaze Plan is Required)

Since you can write raw SQL queries, you can do anything with it. Below are some use cases that can probably be done just by BQ:

  1. Stop your ads when your crash rate goes beyond 2%. There are Airflow connectors for both Google Ads & BigQuery.
  2. Stop your ads when the user repeat ratio goes beyond x%.
  3. Send a notification when % of app_remove events increase.
  4. Send a notification to slack whenever there’s unusually high number of downloads from a certain device.
  5. A crisper analysis of your uninstalls. Yes, there’s a cost to sending notifications – you can potentially lose a user. How do you do such analysis without getting access to raw data?

It takes a few minutes to enable transfer of Firebase data to BigQuery.

To use BQ, you must consider hiring an analyst – who is good with marketing. They can help you make sense of the nebula of data. Or, pick up SQL.

How NextGrowth Labs Works With BQ Data

While you can do all this using BQ, we find regular querying cost prohibitive. At NextGrowth Labs, we use our own BQ connector to transfer data to our DBs. Doing ETL of BQ data to our infra is a tough problem, so we maintain and regularly update how we normalize the data. We often fork the code – depending upon category of the app (E commerce vs others). We tried writing a universal normaliser, but it didn’t work – and resulted in huge number of columns. If you are interested in working with us on this problem, do reach out to us. We believe that it can greatly benefit indie developers, who could spin up a storage server and do all their aggregation on open source dash-boarding tools.

Related :   Google Play Data Studio Connector by NextGrowth Labs.