Export Crashlytics Data to BigQuery

You can export your Crashlytics data into BigQuery for further analysis. BigQuery allows you to analyze the data using BigQuery SQL, export it to another cloud provider, and use it for visualization and custom dashboards with Google Data Studio.

Enabling BigQuery export

1. Go to the Integrations page in the Firebase console.
2. In the BigQuery card, click Link.
3. Follow the on-screen instructions to enable BigQuery.

When you link your project to BiqQuery:

• Firebase exports a copy of your existing data to BigQuery.
• Firebase sets up daily syncs of your data from your Firebase project to BigQuery.
• By default, all apps in your project are linked to BigQuery and any apps that you later add to the project are automatically linked to BigQuery. You can manage which apps send data.

To deactivate BigQuery export, unlink your project in the Firebase console.

What data is exported to BigQuery?

Firebase Crashlytics data is exported into a BigQuery dataset named firebase_crashlytics. By default, individual tables will be created inside the Crashlytics data set for each app in your project. Firebase names the tables based on the app's bundle identifier, with periods converted to underscores, and a platform name appended to the end. For example, data for an app with the ID com.google.test would be in a table named com_google_test_ANDROID.

Each row in a table represents an event that occurred in the app, including both fatal and non-fatal errors.

What can you do with the exported data?

BigQuery exports contain raw crash data including device type, operating system, exceptions (Android apps) or errors (iOS apps), and Crashlytics logs, as well as other data.

Working with Firebase Crashlytics data in BigQuery

The following examples demonstrate queries you can run on your Crashlytics data. These queries generate reports that aren't available in the Crashlytics dashboard.

Examples of Crashlytics queries

The following examples demonstrate how to generate reports that aggregate crash event data into more easily-understood summaries.

Example 1: Crashes by day

After working to fix as many bugs as possible, a lead developer thinks her team is finally ready to launch their new photo-sharing app. Before they do, they want to check the number of crashes per day for the past month, to be sure their bug-bash made the app more stable over time:

SELECT
  COUNT(DISTINCT event_id) AS number_of_crashes,
  FORMAT_TIMESTAMP("%F", event_timestamp) AS date_of_crashes
FROM
 `projectId.firebase_crashlytics.package_name_ANDROID`
GROUP BY
  date_of_crashes
ORDER BY
  date_of_crashes DESC
LIMIT 30;

Example 2: Find most pervasive crashes

To properly prioritize production plans, a project manager ponders how to point out the top 10 most pervasive crashes in their product. They produce a query that provides the pertinent points of data:

SELECT
  DISTINCT issue_id,
  COUNT(DISTINCT event_id) AS number_of_crashes,
  COUNT(DISTINCT installation_uuid) AS number_of_impacted_user,
  blame_frame.file,
  blame_frame.line
FROM
  `projectId.firebase_crashlytics.package_name_ANDROID`
WHERE
  event_timestamp >= TIMESTAMP_SUB(CURRENT_TIMESTAMP(),INTERVAL 168 HOUR)
  AND event_timestamp < CURRENT_TIMESTAMP()
GROUP BY
  issue_id,
  blame_frame.file,
  blame_frame.line
ORDER BY
  number_of_crashes DESC
LIMIT 10;

Example 3: Top 10 crashing devices

Fall is new phone season! A developer knows that also means it's new device-specific issues season. To get ahead of the looming compatibility concerns, they put together a query that identifies the 10 devices that experienced the most crashes in the past week:

SELECT
  device.model,
COUNT(DISTINCT event_id) AS number_of_crashes
FROM
  `projectId.firebase_crashlytics.package_name_ANDROID`
WHERE
  event_timestamp >= TIMESTAMP_SUB(CURRENT_TIMESTAMP(), INTERVAL 168 HOUR)
  AND event_timestamp < CURRENT_TIMESTAMP()
GROUP BY
  device.model
ORDER BY
  number_of_crashes DESC
LIMIT 10;

Example 4: Filter by custom key

A game developer wants to know which level of their game experiences the most crashes. To help them track that stat, they set a custom Crashlytics key current_level, and update it every time the user reaches a new level.

CrashlyticsKit setIntValue:3 forKey:@"current_level";

Crashlytics.sharedInstance().setIntValue(3, forKey: "current_level");

Crashlytics.setInt("current_level", 3);

With that key in their BigQuery export, they then write a query to report the distribution of current_level values associated with each crash event:

SELECT
COUNT(DISTINCT event_id) AS num_of_crashes,
  value
FROM
  `projectId.firebase_crashlytics.package_name_ANDROID`
UNNEST(custom_keys)
WHERE
  key = "current_level"
GROUP BY
  key,
  value
ORDER BY
  num_of_crashes DESC

Example 5: User ID extraction

A developer has an app in early access. Most of their users love it, but three have experienced an unusual number of crashes. To get to the bottom of the problem, they write a query that pulls all the crash events for those users, using their user IDs:

SELECT *
FROM
  `projectId.firebase_crashlytics.package_name_ANDROID`
WHERE
  user.id IN ("userid1", "userid2", "userid3")
ORDER BY
  user.id
 

Example 6: Find all users facing a particular crash issue

A developer has released a critical bug to a group of beta testers. The team was able to use the query from Example 2 above to identify the specific crash issue ID. Now they would like to run a query to extract the list of app users who were impacted by this crash:

SELECT user.id as user_id
FROM
  `projectId.firebase_crashlytics.package_name_ANDROID`
WHERE
  issue_id = "YOUR_ISSUE_ID"
  AND application.display_version = ""
  AND user.id != ""
ORDER BY
  user.id;

Example 7: Number of users impacted by a crash issue, broken down by country

Now the team has detected a critical bug during the rollout of a new release. They were able to use the query from Example 2 above to identify the specific crash issue ID. The team would now like to see if this crash has spread to users in different countries around the world.

To write this query, the team will need to:

1. Enable BigQuery exports for Google Analytics for Firebase. See Export project data to BigQuery.

2. Update their app to pass a user ID into both the Google Analytics for Firebase SDK and the Crashlytics SDK.

   Objective-C

   CrashlyticsKit setUserIdentifier:@"123456789";
   FIRAnalytics setUserID:@"123456789";
   

   Swift

   Crashlytics.sharedInstance().setUserIdentifier("123456789");
   Analytics.setUserID("123456789");
   

   Java

   Crashlytics.setUserIdentifier("123456789");
   mFirebaseAnalytics.setUserId("123456789");
   

3. Write a query that uses the user ID field to join events in the Google Analytics BigQuery data set with crashes in the Crashlytics BigQuery data set:

SELECT DISTINCT c.issue_id, a.geo.country, COUNT(DISTINCT c.user.id) as num_users_impacted
FROM `projectId.firebase_crashlytics.package_name_ANDROID` c
INNER JOIN  `projectId.analytics_YOUR_TABLE.events_*` a on c.user.id = a.user_id 
WHERE
  c.issue_id = "YOUR_ISSUE_ID"
  AND a._TABLE_SUFFIX BETWEEN '20190101'
  AND '20200101'
GROUP BY
  c.issue_id,
  a.geo.country,
  c.user.id

Understanding the Firebase Crashlytics schema in BigQuery

When you link Crashlytics with BigQuery, Firebase exports recent fatal and non-fatal crash events, including events from up to two days before the link.

From that point until you disable the link, Firebase exports Crashlytics events on a daily basis. It can take a few minutes for the data to be available in BigQuery after each export.

Datasets

Firebase Crashlytics creates a new dataset in BigQuery for Crashlytics data. The dataset covers your entire project, even if it has multiple apps.

Tables

Firebase Crashlytics creates a table in the dataset for each app in your project, unless you've opted out of exporting data for that app. Firebase names the tables based on the app's bundle identifier, with periods converted to underscores, and a platform name appended to the end. For example, data for an app with the ID com.google.test would be in a table named com_google_test_ANDROID.

Tables contain a standard set of Crashlytics data in addition to any custom Crashlytics keys defined by the developers.

Rows

Each row in a table represents an error the app encountered.

Columns

The columns in a table are identical for fatal and non-fatal errors. The columns within the export are listed below.

Without stack traces

Columns present in rows that represent events without stack traces.

Visualizing exported Crashlytics data with Data Studio

Google Data Studio turns your Crashlytics datasets in BigQuery into reports that are easy to read, easy to share, and fully customizable.

To learn more about using Data Studio, try the Data Studio quickstart guide, Welcome to Data Studio.

Using a Crashlytics report template

Data Studio has a sample report for Crashlytics that includes a comprehensive set of dimensions and metrics from the exported Crashlytics BigQuery schema. You can use the sample as a template to quickly create new reports and visualizations based on your own app's raw crash data:

1. Open the Crashlytics Data Studio Dashboard template.
2. Click Use Template in the upper-right corner.
3. In the New Data Source dropdown, select Create New Data Source.
4. Click Select on the BigQuery card.
5. Select a table containing exported Crashlytics data by choosing My Projects > [your-project-name] > firebase_crashlytics > [your-table-name].
6. Under Configuration, set Crashlytics Template level to Default.
7. Click Connect to create the new data source.
8. Click Add to Report to return to the Crashlytics template.
9. Finally, click Create Report to create your copy of the Crashlytics Data Studio Dashboard template.