Skip to content

Latest commit

 

History

History
169 lines (119 loc) · 6.3 KB

ANALYTICS.md

File metadata and controls

169 lines (119 loc) · 6.3 KB
Raw

Analytics Best Practices

Analytics are a crucial component of tracking and understanding user behavior, system performance, and other key metrics. To ensure that our analytics are effective and actionable, we follow a set of best practices. These practices help us maintain consistency, accuracy, and clarity in our analytics data.

When adding or updating a new analytic event utilize the following spreadsheet to ensure that the event is consistent with the rest of the events in the system: Analytics Event Spreadsheet

We utilize Honeycomb for web to track analytic events. Using span attributes in OpenTelemetry, we can provide additional context to our analytics data. Below are some best practices for defining span attributes in OpenTelemetry.

OpenTelemetry Span Attribute Best Practices

1. Use Namespaces for Context

When defining span attributes, it's essential to provide context by using namespaces. This helps to clarify the origin and purpose of the attribute, making it easier to interpret in analytics. Read more about Namespaces here

Bad Examples:

  • status
  • identifier
  • function.name
  • function.status

Good Examples:

  • task.status
  • version.status
  • project.identifier
  • section.function.name
  • section.function.status

By including a namespace (e.g., task, version, project, section), the attribute is contextualized, making it clear where it fits within the larger system. It also allows for more straightforward data filtering and analysis. e.g. task.status=success or project.identifier=evergreen.

2. Avoid Excessive Dot Nesting

While namespaces are useful, over-nesting with too many dots can lead to overly complex attribute names. This not only makes them harder to read but can also complicate data queries and analysis.

Bad Examples:

  • host.is.volume.migration
  • host.is.unexpirable

Good Examples:

  • host.is_volume_migration
  • host.is_unexpirable

In these examples, using underscores instead of excessive dot nesting keeps the attribute names concise and easier to manage.

3. Snake case

When defining multi word attributes, use snake_case to separate words. This ensures that the attribute is readable and easy to understand.

4. Strive for Balance

When creating span attributes, aim to balance the need for context with simplicity. Use namespaces thoughtfully to provide clarity without overcomplicating the attribute structure. This approach will help maintain both readability and utility in our analytics. Read more about OpenTelemetry attribute naming here.

Querying and Analyzing Data

To begin querying and analyzing data in Honeycomb, follow these steps:

  1. Select the Correct Dataset and Environment: Ensure you have selected the appropriate dataset and environment before proceeding.

  2. Querying Analytics Events:

    • All analytics events are stored in the library.name = analytics namespace.
    • To query analytics events, use the WHERE clause to filter by specific attributes. Including library.name = analytics ensures that only analytics events are queried.
    • Each event has a name attribute, which can be used to filter for a specific event.
    • The production environment is further divided into production and beta environment events. Use the environment attribute to filter for events in a specific environment.

  3. Grouping Data:

    • Use the GROUP BY clause to organize data by a specific attribute, providing a clearer overview of the data.

Example Query: Grouping by name Attribute

Example Query

This query groups data by the name attribute, showing how often each event has been triggered. To find a specific name attribute or to find which ones exist reference the Analytics Event Spreadsheet.

Grouping Data by Custom Attributes

You can also group data by other custom event specific attributes, such as status or log.type, to see the data organized by custom attributes in the analytics event.

Example Query: Grouping by log.viewer

Example Query

This query groups Clicked log link events by the log.viewer attribute, showing how often a specific log viewer was chosen when viewing a task log.

Grouping Data by analytics.identifier

Analytics events in our apps are split by the specific page or component where the event is triggered, using the analytics.identifier attribute. You can use this attribute to group data by the page or component where the event occurred.

For example, if you want to see how many times a user clicked a button on a specific page, group the data by analytics.identifier to get detailed insights.

Example Query: Grouping by analytics.identifier

Example Query

This query searches for the Click patch link event, which exists on multiple pages. It then groups the data by analytics.identifier to show how often each event was triggered on specific pages.

Default Attributes

All analytics events have several default attributes that can be used for querying. Here are some common default attributes:

  • analytics.identifier
  • name
  • browser.name
  • url.path
  • user.id
  • environment

Additional Resources