Glossary

A/A Experiment: A/A testing is a form of A/B testing where two identical versions of a product or design are tested against each other to check the consistency of the testing platform and methodology.

A/B Experiment: A/B testing is a method of comparing two versions of a webpage or other product to determine which one performs better by splitting traffic between the two versions and measuring their performance using statistical analysis.

Analysis attribute: Analysis attribute in GrowthBook is a user-specific attribute (like user id, device id, etc.) used for segmenting and analyzing experiment results based on different user characteristics.

Approval Flow: Approval flow is a feature in GrowthBook that requires changes to features to be reviewed and approved before they are published, reducing errors and ensuring changes have been viewed by someone else in the organization.

Assignment: Assignment in GrowthBook refers to the process of allocating users to different variations of an experiment, typically done using a consistent hashing algorithm to ensure the same user always gets the same variation given the same experiment settings.

Assignment attribute: Assignment attribute in GrowthBook is a user-specific attribute (like user id, device id, etc.) used to consistently assign the same user to the same variation across multiple sessions or devices.

Audit Logs: Audit logs in GrowthBook are records of all actions taken on the platform, useful for auditing user activities and troubleshooting issues. These logs can be accessed from the Settings → Log in the left navigation.

Binomial Metric: A binomial metric in GrowthBook is a type of legacy metric that represents a binary outcome (yes/no or success/failure), such as whether a user converted or not, and is used to compute the proportion of users in an experiment variation who converted. The new name for this metric type is a Proportion Metric.

Carryover Bias: Carryover bias in GrowthBook refers to the contamination of experiment results when users are not re-randomized between different phases of an experiment, causing the effects of the first phase to influence the results of the subsequent phase.

Cloud-Hosted: Cloud-hosted in GrowthBook refers to the GrowthBook platform hosted on the cloud, providing a managed service that takes care of infrastructure, maintenance, and updates, allowing users to focus on running and analyzing experiments.

Code Reference: Code Reference in GrowthBook allows teams to quickly see instances of feature flags being leveraged in their codebase, helping surface valuable information such as flagging stale feature flags more accurately, and directing devs to the exact lines of code that need to be cleaned up.

Config.yml File: The config.yml file in GrowthBook is used to define data sources, metrics, and dimensions, making it easier to version control the configuration and share between environments when self-hosting the platform.

Control: The control in GrowthBook refers to the baseline or original version in an A/B test against which other variations are compared to determine their effectiveness.

Count Metric: A count metric in GrowthBook is a type of legacy metric that represents the number or magnitude of conversions (like pages per visit or points earned), where the value of each conversion event is summed together at the user level before taking an average per variation. The new name for this metric type is a Mean Metric.

Data Source: A Data Source in GrowthBook is how the platform connects to your data warehouse to pull aggregated statistics in order to compute metrics and experiment results, supporting various SQL versions and event tracker software.

Data Warehouse: A Data Warehouse in GrowthBook is your secure storage where your data resides, and only aggregated statistics are transmitted to GrowthBook servers or your self-hosted environment for analysis.

Dimension: A Dimension in GrowthBook is a way to slice and dice your data, allowing you to drill down into your results by different user attributes or experiment characteristics for deeper analysis.

Duration Metric: A duration metric in GrowthBook is a type of metric that measures the time it takes to do something (like page load time or time on site), where the value of each event is summed together at the user level before taking an average per variation.

Edge SDK: The Edge SDK in GrowthBook is a set of libraries designed to run on edge workers, enabling server-side or hybrid experimentation with features like visual and URL redirect experiments without flicker or delay.

Environment: An Environment in GrowthBook is a specific context (like production, staging, or development) where you can enable or disable feature flags, allowing you to control and segment feature delivery across different stages of your application lifecycle.

Event Webhooks: Event Webhooks in GrowthBook are outbound notifications that trigger when the state of GrowthBook changes, such as when a feature is modified or a new experiment is created, enabling real-time communication from GrowthBook to your applications.

Experiment Assignment Query: An Experiment Assignment Query in GrowthBook is a SQL query that returns which users were part of which experiment, what variation they saw, and when they saw it, helping to generate experiment results.

Experimentation: Experimentation in GrowthBook refers to the process of testing different variations of a feature or product to determine which one performs better, using statistical analysis to make data-driven decisions.

Fact Table: A Fact Table in GrowthBook is a feature that allows you to write SQL once for an event and from that, you can create multiple related metrics, simplifying the process of defining metrics and enabling advanced performance and cost optimizations.

Feature: Features in GrowthBook are configurable elements of your product that can be controlled via the GrowthBook UI, allowing you to change your application's behavior, run experiments, and gradually roll out new functionality to your users.

Feature Flag: A Feature Flag in GrowthBook is a powerful developer tool that allows you to control how and when new functionality is released to your users, supporting different types (Boolean, Number, String, JSON) and environments (dev, production).

Feature Key: A Feature Key in GrowthBook is a unique identifier that you reference in your application to control a specific feature flag, and it cannot be changed later once it's created.

Forced Value: A Forced Value in GrowthBook is a type of feature rule that assigns a specific value to a feature for a subset of users based on targeting attributes, useful for testing or targeting specific user groups.

Global SDK Webhooks: Global SDK Webhooks in GrowthBook are notifications that fire for all SDK Connection changes, not limited to a single SDK Connection, and are particularly useful for larger organizations with multi-org installations of GrowthBook.

GrowthBook Chrome Extension: The GrowthBook Chrome Extension integrates with the JavaScript SDKs, allowing you to QA and debug feature flags and experiments from GrowthBook's JavaScript SDKs directly in your browser.

GrowthBook Proxy: The GrowthBook Proxy is a server that sits between your application and GrowthBook, providing speed, scalability, security, and real-time feature rollouts by caching feature lookups and streaming updates in real-time.

Hashing Algorithm: The Hashing Algorithm in GrowthBook is a deterministic method used to consistently assign users to experiment variations, ensuring the same user always gets the same variation given the same experiment settings.

Holdout Experiments: Holdout Experiments in GrowthBook are a method of measuring the long-term impact of one or more features by keeping a set of users (the holdout group) from seeing new features and comparing their behavior to another set of users who do see the new features.

Mean Metric: A Mean Metric in GrowthBook is a type of metric that measures an average value among all experiment users. For example, the average revenue per user.

Metric: A Metric in GrowthBook is a key performance indicator that your experiments are trying to improve, defined via SQL or a simple query builder, and can be of various types like proportion, mean, quantile, or ratio.

Metric Query: A Metric Query in GrowthBook is a SQL or simple query that defines how to fetch data for a specific metric from your data source, returning one row per "conversion event" and supporting various types of identifiers.

Multiple Exposures: Multiple Exposures in GrowthBook refers to a situation where a substantial number of users in an experiment have been exposed to multiple variations, indicating a potential issue with the experiment assignment or tracking implementation.

Namespace: A Namespace in GrowthBook is a feature that allows you to make multiple experiments mutually exclusive, ensuring that users are only part of one experiment within the same namespace at a time, useful for avoiding conflicts between experiments.

Peeking Problem: The Peeking Problem in GrowthBook refers to the issue of experimenters making decisions about the results of an experiment based on early data, which can lead to inflated false positive rates, and can be mitigated using techniques like Sequential Testing.

Percentage Rollout: A Percentage Rollout in GrowthBook is a rule that allows you to gradually release a feature to a random sample of your users, useful for ensuring a new feature doesn't break your app or site and for controlling the exposure of your users to the new feature.

Proportion Metric: A Proportion Metric in GrowthBook is a type of metric that measures the percent of experiment users who match a specific criteria (like the percent of users who purchased something).

Prerequisite Feature: A Prerequisite Feature in GrowthBook is a feature that controls the state of other features, rules, and experiments based on its own state, allowing for grouping of related features, creating a hierarchy of features, and enabling features based on user bucketing.

Project: A Project in GrowthBook is a way to organizationally separate features, metrics, experiments, and data sources by team or product feature within an organization, allowing for focused views and management of specific sections of GrowthBook.

Quantile Metric: A Quantile Metric in GrowthBook is a type of metric that measures a specific quantile (Median, P90, etc.) of a value. For example, P99 Latency or Median Revenue per User.

Ratio Metric: A Ratio Metric in GrowthBook is a type of metric that measures the ratio between two values. For example, Average Order Value (sum of revenue divided by number of orders).

Revenue Metric: A Revenue Metric in GrowthBook is a type of legacy metric that measures the amount of revenue earned (like revenue per user or average order value), where the value of each event is summed together at the user level before taking an average per variation. The new name for this is a Mean Metric.

Rule: A Feature Rule in GrowthBook is a tool to change the default value of a feature for specific users or groups, supporting different types like Forced Value, Percentage Rollout, and A/B Experiment, and providing fine-grained control over feature assignment.

Sample Ratio Mismatch (SRM): A Sample Ratio Mismatch (SRM) in GrowthBook is a health check that warns you when the actual traffic split between experiment variations significantly deviates from the expected split, indicating potential issues with the experiment configuration or implementation.

Saved Groups: Saved Groups in GrowthBook are reusable lists of users, based on specific attributes, that can be targeted across multiple features or experiments, supporting two types: ID Lists and Condition Groups, and allowing for efficient and consistent user targeting.

SCIM: SCIM (System for Cross-domain Identity Management) in GrowthBook is a standard for managing users and groups across multiple applications, enabling automation of user provisioning and deprovisioning through your identity provider, currently supporting Okta as the identity provider.

SDK: The SDK (Software Development Kit) in GrowthBook is a set of libraries for different programming languages that allows you to integrate GrowthBook into your application, enabling feature flagging, running experiments, and sending data back to GrowthBook for analysis.

SDK Connection: An SDK Connection in GrowthBook is a unique link between your application and GrowthBook, generating a clientKey for read-only access to feature flags in a specific environment, and used by the SDKs to fetch feature flag states and rules from the GrowthBook API.

SDK Connection Endpoint: The SDK Connection Endpoint in GrowthBook is a readonly API endpoint that provides access to a subset of your feature flag data, just enough for the GrowthBook SDKs to assign values to users, and does not require authentication to view.

SDK Webhooks: SDK Webhooks in GrowthBook are notifications that trigger a script on your server whenever something changes within GrowthBook that will affect that SDK, allowing for real-time updates and synchronization of feature flag states.

Segment: A Segment in GrowthBook is a subset of users that match a particular attribute, allowing you to filter experiment results to only show users that match the segment, and can be created with SQL from the "Data and Metrics → Segments" page.

Self-Hosted: Self-Hosted in GrowthBook refers to the option of hosting the GrowthBook platform on your own infrastructure, giving you full control over your data and allowing you to customize the platform to suit your specific needs.

SSO: Single Sign-On (SSO) in GrowthBook is an enterprise feature that allows users to authenticate with a single set of credentials across multiple applications, available on GrowthBook Cloud or Self-hosted via OpenID Connect, and configurable with various identity providers.

Stale Feature Flag: A Stale Feature Flag in GrowthBook is a feature flag that is no longer actively used or relevant, and GrowthBook's Stale Feature Flag Detection feature helps identify and manage these flags, ensuring your feature flag ecosystem remains clean and efficient.

Sticky Bucketing: Sticky Bucketing in GrowthBook is a feature that ensures users continue to see the same variation when changes are made to a running experiment, allowing for consistent user experiences and accurate experiment results.

Targeting Condition: A Targeting Condition in GrowthBook is a rule that determines which users are included in an experiment or receive a specific feature value, based on user attributes, and can be defined using a simple UI or advanced JSON syntax inspired by MongoDB query syntax.

TrackingCallback(): The trackingCallback() in GrowthBook is a function that is called every time an experiment is run, allowing you to record the assigned value in your event tracking or analytics system, and is used across various SDKs for tracking experiment results.

Traffic Split: A Traffic Split in GrowthBook is a configuration that determines how users are divided among different variations in an experiment, allowing you to control both the percentage of users included in the experiment and the distribution of users among the variations.

URL Redirect: A URL Redirect in GrowthBook is a type of experiment that redirects users to different URLs based on their assigned variation, ideal for testing big changes or complete page redesigns, and requires integration with GrowthBook's SDKs.

URL Split Test: A URL Split Test in GrowthBook is a method of A/B testing where traffic is split between two different URLs to compare their performance, typically implemented using feature flags and custom JavaScript to redirect users based on their assigned variation.

Variation: A Variation in GrowthBook is a version of a feature or product that is being tested in an experiment, with users being assigned to different variations to compare their performance and impact on key metrics.

Visual Editor: The Visual Editor in GrowthBook is a tool that allows you to design A/B tests directly in your browser without writing a single line of code, enabling you to run experiments on your site and analyze results in real-time.