Reports glossary
Last updated
Last updated
A user in Tabnine private installation can be in one of these states:
Invited User (Status=Invited): Has a pending email invitation to join Tabnine but has not yet registered. This user is authorized to use Tabnine but cannot connect and use Tabnine until registered.
Registered User: This user has signed up to Tabnine after an invitation (by email or link), by self-registration, or through SSO. This user is authorized to connect and use Tabnine. Registered users can be in one of the following statuses based on how advanced they are in onboarding to Tannine.
Status=Registered: This is a registered user who has not yet connected to Tabnine in the IDE.
Status=Connected: This is a registered user who has connected to Tabnine in the IDE but has not actually used Tabnine yet.
Status=Used: This is a registered user who has connected to the Tabnine in the IDE and has completed at least one chat interaction or code completion.
Deactivated User (Status=Deactivated): This user had been a registered user at some point, but has been deactivated by an admin since then. This user is no longer an authorized user and cannot use Tabnine.
Notes:
Admin can deactivate Registered users (changing their status to Deactivated) or activate Deactivated users (changing their status to back to Registered, Connected, or Used).
The transition between the Registered, Connected, and Used statuses is one-directional, meaning that once a user has connected Tabnine in the IDE, they cannot be back in "Registered" status. Once a user has used Tabnine, they cannot be back in "Connected" status.
An active user refers to a user who has completed at least one chat interaction or code completion in a specific time frame. For example, a user can be active in March but not in April. Another example is that a user can be an active user in the last 30 days, but not in the last 7 days.
The code completion automation factor is a metric representing which percentage of the code was generated using Tabnine code completions and is calculated in the following manner:
For example, if the user typed 20 characters and Tabnine generated 5 characters:
AF = 5 / [25 =(20+5)] => 20% automated
where the sum of typed characters is proxied by the number of requests sent from the binary.
The automation factor can be calculated for the entire organization, for a specific team, or for individuals, and in different time frames.
However, the automation factor calculation relies on minimal code completion activity over a period of time, and in some cases, it is not available.
Chat interaction: Refers to a user action that triggers Tabnine chat, either through a free text prompt or using a built-in command (i.e., "The user asks a question").
Useful chat interaction: Refers to any engagement with the chat answer that suggests that the user found the chat answer useful (i.e., "The user found the chat answer to be useful").
Note: The useful chat interaction data has been collected since version 5.6.0
Copy code: The user copied code from the answer by clicking copy/insert code or by free copying parts of the code in the chat answer.
Follow-up click: The user clicked a related question offered by the chat, which further suggests that the chat helped the user discover more relevant information or context on the problem they're trying to solve.
Navigate to a reference: The user navigated to a code reference included in the chat answer, which helped the user navigate and find the reference more easily.
Review: The chat helps the user review and compare a specific suggestion to their existing code.
Thumbs up: Explicit expression of content from the user by clicking thumbs up for a specific answer.
The Chat Consumption Rate is a metric that measures the usefulness of Tabnine chat answers. It is calculated as the ratio between the total number of times a user copied code from a chat answer and the total number of questions the user asked:
Since a user can copy code more than once from the same chat answer (for example, when the answer contains multiple code blocks), the Chat Consumption Rate can exceed 100%.
This rate can be calculated at various levels: for the entire organization, specific teams, or individual users, and across different time frames.
Note: Chat Consumption Rate data has been collected since version 5.6.0.
The total number of lines of code accepted from code completions or useful chat interactions (specifically of type "copy code").
Note: Lines of code data has been collected since version 5.7.0.
