Ruby

The posthog-ruby library provides tracking functionality on the server-side for applications built in Ruby.

It uses an internal queue to make calls fast and non-blocking. It also batches requests and flushes asynchronously, making it perfect to use in any part of your web app or other server-side application that needs performance.

Use a single client instance (singleton) — Create the PostHog client once and reuse it throughout your application. Multiple client instances with the same API key can cause dropped events and inconsistent behavior. The SDK logs a warning if it detects multiple instances.

Installation

Add this to your Gemfile:

Terminal

In your app, set your API key before making any calls. If setting a custom host, make sure to include the protocol (e.g. https://).

Ruby

You can find your project token and instance address in the project settings (opens in a new tab) page in PostHog.

Configuration

Initialize the client with your project token before making any calls:

Ruby

Available client options:

Filtering or modifying events before sending

Use before_send to add, modify, or drop events immediately before the SDK queues or sends them:

Ruby

Flushing and shutting down

For short-lived scripts, call flush before the process exits. Call shutdown when your application is stopping to flush pending events and stop background resources.

Ruby

Identifying users

Identifying users is required. Backend events need a distinct_id that matches the ID your frontend uses when calling posthog.identify(). Without this, backend events are orphaned — they can’t be linked to frontend event captures, session replays (opens in a new tab), LLM traces (opens in a new tab), or error tracking (opens in a new tab).

See our guide on identifying users (opens in a new tab) for how to set this up.

Identify a user and set their person properties with identify:

Ruby

Capturing events

You can send custom events using capture:

Ruby

Tip: We recommend using a [object] [verb] format for your event names, where [object] is the entity that the behavior relates to, and [verb] is the behavior itself. For example, project created, user signed up, or invite sent.

Setting event properties

Optionally, you can include additional information with the event by including a properties (opens in a new tab) object:

Ruby

Sending pageviews

If you’re aiming for a backend-only implementation of PostHog and won’t be capturing events from your frontend, you can send pageviews from your backend like so:

Ruby

capture accepts these fields:

Person profiles and properties

The Ruby SDK captures identified events by default. These create person profiles (opens in a new tab). To set person properties (opens in a new tab) in these profiles, include them when capturing an event:

Ruby

For more details on the difference between $set and $set_once, see our person properties docs (opens in a new tab).

To capture anonymous events (opens in a new tab) without person profiles, set the event’s $process_person_profile property to false:

Ruby

Alias

Sometimes, you want to assign multiple distinct IDs to a single user. This is helpful when your primary distinct ID is inaccessible. For example, if a distinct ID used on the frontend is not available in your backend.

In this case, you can use alias to assign another distinct ID to the same user.

Ruby

We strongly recommend reading our docs on alias (opens in a new tab) to best understand how to correctly use this method.

Feature flags

PostHog’s feature flags (opens in a new tab) enable you to safely deploy and roll back new features as well as target specific users and groups with them.

There are two steps to implement feature flags in Ruby:

Step 1: Evaluate flags once

Call posthog.evaluate_flags() once for the user, then read values from the returned snapshot.

Boolean feature flags

Ruby

Multivariate feature flags

Ruby

flags.get_flag() returns the variant string for multivariate flags, true for enabled boolean flags, false for disabled flags, and nil when the flag wasn’t returned by the evaluation.

Note: posthog.is_feature_enabled(), posthog.get_feature_flag(), posthog.get_feature_flag_result(), posthog.get_feature_flag_payload(), and capture({ ..., send_feature_flags: true }) still work during the migration period, but they’re deprecated. Prefer evaluate_flags() for new code.

Step 2: Include feature flag information when capturing events

If you want use your feature flag to breakdown or filter events in your insights (opens in a new tab), you’ll need to include feature flag information in those events. This ensures that the feature flag value is attributed correctly to the event.

Note: This step is only required for events captured using our server-side SDKs or API (opens in a new tab).

There are two methods you can use to include feature flag information in your events:

Method 1: Pass the evaluated flags snapshot to capture()

Pass the same flags object that you used for branching. This attaches the exact flag values from that evaluation and doesn’t make another /flags request.

Ruby

By default, this attaches every flag in the snapshot using $feature/<flag-key> properties and $active_feature_flags.

To reduce event property bloat, pass a filtered snapshot:

Ruby

only_accessed is order-dependent. If you call it before accessing any flags with enabled?() or get_flag(), no feature flag properties are attached.

Method 2: Include the $feature/feature_flag_name property manually

In the event properties, include $feature/feature_flag_name: variant_key:

Ruby

Evaluating only specific flags

By default, evaluate_flags() evaluates every flag for the user. If you only need a few flags, pass flag_keys to request only those flags:

Ruby

Evaluating locally only

If you want to skip the remote /flags request and only use locally cached definitions, pass only_evaluate_locally: true:

Ruby

Disabling GeoIP for flag evaluation

Pass disable_geoip: true to disable GeoIP lookup for remote flag evaluation:

Ruby

Sending $feature_flag_called events

Capturing $feature_flag_called events enables PostHog to know when a flag was accessed by a user and provide analytics and insights (opens in a new tab) on the flag. With evaluate_flags(), the SDK sends this event when you call flags.enabled?() or flags.get_flag() for a flag.

The SDK deduplicates these events per (distinct_id, flag, value) in a local cache. If you reinitialize the PostHog client, the cache resets and $feature_flag_called events may be sent again. PostHog handles duplicates, so duplicate $feature_flag_called events don’t affect your analytics.

flags.get_flag_payload() doesn’t send $feature_flag_called events and doesn’t count as an access for only_accessed.

Advanced: Overriding server properties

Sometimes, you may want to evaluate feature flags using person properties (opens in a new tab), groups (opens in a new tab), or group properties that haven’t been ingested yet, or were set incorrectly earlier.

You can provide properties to evaluate the flag with by using the person properties, groups, and group properties arguments. PostHog will then use these values to evaluate the flag, instead of any properties currently stored on your PostHog server.

For example:

Ruby

Overriding GeoIP properties

By default, a user’s GeoIP properties are set using the IP address they use to capture events on the frontend. You may want to override the these properties when evaluating feature flags. A common reason to do this is when you’re not using PostHog on your frontend, so the user has no GeoIP properties.

You can override GeoIP properties by including them in the person_properties parameter when evaluating feature flags. This is useful when you’re evaluating flags on your backend and want to use the client’s location instead of your server’s location.

The following GeoIP properties can be overridden:

• $geoip_country_code
• $geoip_country_name
• $geoip_city_name
• $geoip_city_confidence
• $geoip_continent_code
• $geoip_continent_name
• $geoip_latitude
• $geoip_longitude
• $geoip_postal_code
• $geoip_subdivision_1_code
• $geoip_subdivision_1_name
• $geoip_subdivision_2_code
• $geoip_subdivision_2_name
• $geoip_subdivision_3_code
• $geoip_subdivision_3_name
• $geoip_time_zone

Simply include any of these properties in the person_properties parameter alongside your other person properties when calling feature flags.

Request timeout

You can configure the feature_flag_request_timeout_seconds parameter when initializing your PostHog client to set a flag request timeout. This helps prevent your code from being blocked if PostHog’s servers are too slow to respond. By default, this is set to 3 seconds.

Ruby

Legacy single-flag methods

The following methods are still available during the migration period, but are deprecated. Prefer evaluate_flags for new code.

Local Evaluation

Evaluating feature flags requires making a request to PostHog for each flag. However, you can improve performance by evaluating flags locally. Instead of making a request for each flag, PostHog will periodically request and store feature flag definitions locally, enabling you to evaluate flags without making additional requests.

It is best practice to use local evaluation flags when possible, since this enables you to resolve flags faster and with fewer API calls.

For details on how to implement local evaluation, see our local evaluation guide (opens in a new tab).

Evaluating feature flags locally in unicorn server

If you have preload_app true in your unicorn config, you can use the after_fork (opens in a new tab) hook (which is part of the unicorn’s configuration) to enable the feature flag cache to receive the updates from PostHog.

Ruby

Evaluating feature flags locally in a Puma server

If you use Puma with multiple workers, you can use the on_worker_boot hook (which is part of Puma’s configuration) to enable the feature flag cache to receive updates from PostHog.

Ruby

Distributed flag definition caching

flag_definition_cache_provider shares locally evaluated feature flag definitions across multiple workers or processes. The provider object must implement:

• flag_definitions – returns cached definitions as a hash with :flags, :group_type_mapping, and :cohorts, or nil if empty.
• should_fetch_flag_definitions? – returns true if this process should fetch fresh definitions from PostHog.
• on_flag_definitions_received(data) – stores freshly fetched definitions.
• shutdown – releases locks or other resources.

Ruby

Remote config payloads

Use get_remote_config_payload to fetch the decrypted remote config payload for a flag. This requires personal_api_key.

Ruby

Experiments (A/B tests)

Since experiments (opens in a new tab) use feature flags, the code for running an experiment is very similar to the feature flags code:

Ruby

It’s also possible to run experiments without using feature flags (opens in a new tab).

Group analytics

Group analytics allows you to associate an event with a group (e.g. teams, organizations, etc.). Read the Group Analytics (opens in a new tab) guide for more information.

Note: This is a paid feature and is not available on the open-source or free cloud plan. Learn more on the pricing page (opens in a new tab).

Capture an event and associate it with a group:

Ruby

Update properties on a group:

Ruby

The name is a special property which is used in the PostHog UI for the name of the group. If you don’t specify a name property, the group ID will be used instead.

If the optional distinct_id is not provided in the group identify call, it defaults to $#{group_type}_#{group_key} (e.g., $company_company_id_in_your_db in the example above). This default behavior will result in each group appearing as a separate person in PostHog. To avoid this, it’s often more practical to use a consistent distinct_id, such as group_identifier.

Exception capture

You can capture exceptions using the posthog-ruby library. This enables you to see stack traces and debug errors in your application. Learn more in our error tracking docs (opens in a new tab).

Using Rails?

The posthog-rails (opens in a new tab) gem provides automatic exception capture, ActiveJob instrumentation, and user context out of the box. See our Rails error tracking guide (opens in a new tab) for details.

For non-Rails Ruby applications, you can manually capture exceptions with capture_exception:

Ruby

The capture_exception method accepts the following parameters:

You can also override the fingerprint (opens in a new tab) to customize how exceptions are grouped into issues:

Ruby

Debug mode

The Ruby SDK logs warnings by default. You can change the log level to DEBUG to debug the client:

Ruby

You can also replace the SDK logger globally:

Ruby

Test helpers

When test_mode: true, events remain queued. You can inspect and clear the queue in tests:

Ruby

Thank you

This library is largely based on the analytics-ruby package.

Still have questions?

Ask PostHog AI

Was this page useful?

HelpfulCould be better