Skip to main content

In our service, non-logged-in users can perform certain actions, and for the purpose of tracking metrics such as conversion rates, we need to uniquely identify these non-logged-in users.

Previously, we identified these non-logged-in users as individual users through Device IDs. However, the problem is that these Device IDs frequently change due to reasons such as app updates, app/cache deletion, etc. This has resulted in instances where a single actual user is assigned multiple Device IDs, causing them to be recognized as multiple users.

Fortunately, in Amplitude, even non-logged-in users with multiple Device IDs (but not assigned a User ID) can be uniquely identified as individual users through a single Amplitude ID.

The issue arises when analyzing log data via queries in Google BigQuery, as Amplitude IDs are not separately stored. Therefore, users must be identified based on Device ID or User ID, and since non-logged-in users can only be identified by Device ID, the above-mentioned issue persists.

My thought is that if we could export the Device IDs and Amplitude IDs from Amplitude for each user's event logs, and use these Device IDs as keys to join with the log data stored in BigQuery (where Device IDs are recorded in the log tables), we could uniquely identify non-logged-in users(with multiple Device IDs) using a single merged Amplitude ID.

 

This was a lengthy background explanation, but with this context in mind, I am inquiring about whether it is possible to extract the multiple Device IDs assigned to each user and their merged Amplitude IDs from Amplitude to an external source (e.g., Google BigQuery).

  1. From some individual research, I have confirmed that there is a method to export data to Google BigQuery as a destination. I am curious if there are any other methods available besides this method.

  2. Additionally, when exporting data using the above method, I would like to know if it is possible to selectively extract only the Device ID and Amplitude ID per unique user, rather than the entire event data column.

Hi Root,

Thank you for contacting Amplitude Support. Happy to help!

1. When you export data to BigQurey, you can choose to export all merged Amplitude IDs. This will include a table of merged user IDs, which tracks all the Amplitude IDs that have been merged for each user. This table can be used to accurately count unique users, as it prevents a single user from being counted multiple times just because they logged events on multiple devices or as an anonymous user.

For more details, please check out our documentation.
https://amplitude.com/docs/data/destination-catalog/google-bigquery#merged-user-table-schema

2. Alternatively, there is an option to export device IDs linked to an Amplitude ID in Cohort Export CSV. This is behind a feature flag. If you would like to explore this option, please let me know and I can turn on the feature for your org.

 



I hope this helps. Please don't hesitate to let me know if you have any questions.

Best regards,
Thao


P.S. Checkout upcoming events and user meetups on our events page.

Reply


Terms & ConditionsCookie settings