Transition from Google Analytics to Firebase
An article about the experience of switching from Google Analytics (GA) to Firebase in the context of analytics for mobile applications, namely games.
So, consistently on the points:
1. Motive transition
')
Google is actively promoting Firebase as a replacement for GA and in its updates for Android Studio Firebase become an integral part of the environment. If you switch to the latest SDK, then Firebase is there by default (as well as Play Services as well as the latest Admob with the latest native features and ad rewarder), and GA is no longer a separate package.
Thus, to update the old Android project, for example, on the latest advertising SDK (Admob) from Google without GA failure will not work anymore.
The issue of learning Firebase has become an edge.
2. Advantages of Firebase over GA
The main interest that Firebase represents in terms of analytics is the support of automatic export of raw data to BigQuery (shareware!) And the lack of sampling.
What else:
- Audience support, attribution with more advanced mechanisms (+ push and more accurate demographic forecast) for working with a targeted audience - this is in terms of marketing.
- Some ready metrics like ARPU, ARPPU.
- Built-in basic events of the type first_open, app_remove, app_update, which are not always convenient to track manually.
- There are even relatively good funnels.
All the basic functions of Firebase are quite well described in the documentation - it is easy enough to find.
3. And now the disadvantages
In the context of comparison with GA:
- Lack of rialtime.
- Limited segmentation functionality.
- Too exaggerated information on metrics. In GA you can look at any statistics in much more detail.
- No filtering (for example, by IP, to filter the testers grid, city, country).
- There is no API for accessing data to generate super cool and very diverse reports like in GA (unless you immediately go into BigQuery).
Those. if you are interested in reports, you will have to do everything by hand through queries in BiqQuery.
4. Total
Thus, the verdict is as follows: if you start from scratch or you do not have ready GA developments (and even more so the GA working bundle with BigQuery, which costs 100K per year in a Premium Account), then you can just sit down to deploy Firebase (especially if you are using the latest SDK / Play Services from Google).
If there is some groundwork and are not ready to spend a lot of time on reconfiguring reports and rebuilding the analytics system in your project (not just GA-based), then it’s time to wait and wait for Firebase to gain power with Google updates.
Remark
If you are building an analytics system, then ideally it should include all the steps to calculate the cornerstone of gaming analytics: the ratio of user engagement costs (CPI) and revenues from this user (ARPU).
There are a lot of options for calculating these indicators, but the goal is the same: revenues should exceed expenses. The analytical system should be able to receive and process data from marketing accounts to build cost projections, should know who was purchased and should be able to receive income data from these involved users. By calculating CPI, ARPDAU, you can build a forecast on the effectiveness of these purchases and either stop them or continue. And of course to provide any calculations in a digestible form. It's minimum. After all, there are still a lot of people interested in analytics: GD (we need balance checks, new features indicators), testers (statistics of failures, errors), managers (indicators of the product and its updates), marketers (a / b-tests of promotional pictures).
All this is quite simple done using GA + Google Spreadsheets + BigQuery + Procurement System (for example, Adwords or even Facebook Ads) with building beautiful automated reports (Google Ananlytics plugin for Google Spreadsheets).
In the case of Firebase, this will be Google Spreadsheets + BigQuery + Procurement System. In Firebase itself, it’s not even necessary to look, if you make all the metrics you are interested in using BigQuery queries, where all the necessary data on the events will lie. It is also worth noting that the event scheme for BiqQuery is different in GA and Firebase (it’s just not possible to port the database query system).
In itself, the trial in these systems is already a plus in the experience and karma of the future analyst.
Update:
Firebase has turned on real-time (fairly rich): firebase.googleblog.com/2017/03/realtime-analytics-for-everyone.html
So, consistently on the points:
1. Motive transition
')
Google is actively promoting Firebase as a replacement for GA and in its updates for Android Studio Firebase become an integral part of the environment. If you switch to the latest SDK, then Firebase is there by default (as well as Play Services as well as the latest Admob with the latest native features and ad rewarder), and GA is no longer a separate package.
Thus, to update the old Android project, for example, on the latest advertising SDK (Admob) from Google without GA failure will not work anymore.
The issue of learning Firebase has become an edge.
2. Advantages of Firebase over GA
The main interest that Firebase represents in terms of analytics is the support of automatic export of raw data to BigQuery (shareware!) And the lack of sampling.
What else:
- Audience support, attribution with more advanced mechanisms (+ push and more accurate demographic forecast) for working with a targeted audience - this is in terms of marketing.
- Some ready metrics like ARPU, ARPPU.
- Built-in basic events of the type first_open, app_remove, app_update, which are not always convenient to track manually.
- There are even relatively good funnels.
All the basic functions of Firebase are quite well described in the documentation - it is easy enough to find.
3. And now the disadvantages
In the context of comparison with GA:
- Lack of rialtime.
- Limited segmentation functionality.
- Too exaggerated information on metrics. In GA you can look at any statistics in much more detail.
- No filtering (for example, by IP, to filter the testers grid, city, country).
- There is no API for accessing data to generate super cool and very diverse reports like in GA (unless you immediately go into BigQuery).
Those. if you are interested in reports, you will have to do everything by hand through queries in BiqQuery.
4. Total
Thus, the verdict is as follows: if you start from scratch or you do not have ready GA developments (and even more so the GA working bundle with BigQuery, which costs 100K per year in a Premium Account), then you can just sit down to deploy Firebase (especially if you are using the latest SDK / Play Services from Google).
If there is some groundwork and are not ready to spend a lot of time on reconfiguring reports and rebuilding the analytics system in your project (not just GA-based), then it’s time to wait and wait for Firebase to gain power with Google updates.
Remark
If you are building an analytics system, then ideally it should include all the steps to calculate the cornerstone of gaming analytics: the ratio of user engagement costs (CPI) and revenues from this user (ARPU).
There are a lot of options for calculating these indicators, but the goal is the same: revenues should exceed expenses. The analytical system should be able to receive and process data from marketing accounts to build cost projections, should know who was purchased and should be able to receive income data from these involved users. By calculating CPI, ARPDAU, you can build a forecast on the effectiveness of these purchases and either stop them or continue. And of course to provide any calculations in a digestible form. It's minimum. After all, there are still a lot of people interested in analytics: GD (we need balance checks, new features indicators), testers (statistics of failures, errors), managers (indicators of the product and its updates), marketers (a / b-tests of promotional pictures).
All this is quite simple done using GA + Google Spreadsheets + BigQuery + Procurement System (for example, Adwords or even Facebook Ads) with building beautiful automated reports (Google Ananlytics plugin for Google Spreadsheets).
In the case of Firebase, this will be Google Spreadsheets + BigQuery + Procurement System. In Firebase itself, it’s not even necessary to look, if you make all the metrics you are interested in using BigQuery queries, where all the necessary data on the events will lie. It is also worth noting that the event scheme for BiqQuery is different in GA and Firebase (it’s just not possible to port the database query system).
In itself, the trial in these systems is already a plus in the experience and karma of the future analyst.
Update:
Firebase has turned on real-time (fairly rich): firebase.googleblog.com/2017/03/realtime-analytics-for-everyone.html
Source: https://habr.com/ru/post/322082/
All Articles