> ## Documentation Index
> Fetch the complete documentation index at: https://docs.growthbook.io/llms.txt
> Use this file to discover all available pages before exploring further.
# Stop an experiment
## OpenAPI
````yaml openapi.yaml POST /v1/experiments/{id}/stop
openapi: 3.1.0
info:
version: 5.1.0
title: GrowthBook REST API
description: >
GrowthBook offers a full REST API for interacting with the application.
Request data can use either JSON or Form data encoding (with proper
`Content-Type` headers). All response bodies are JSON-encoded.
The API base URL for GrowthBook Cloud is `https://api.growthbook.io/api`.
For self-hosted deployments, it is the same as your API_HOST environment
variable (defaults to `http://localhost:3100/api`). The rest of these docs
will assume you are using GrowthBook Cloud.
## Versioning
Endpoints are versioned by path prefix:
- `/v1/...` — stable, widely-supported endpoints
- `/v2/...` — updated endpoints with improved shapes (e.g. unified per-rule
environment scope for feature flags)
New integrations should prefer v2 where available. Feature-flag and metric
v1 routes are deprecated: they are tagged **LEGACY** in the sidebar and
collapsed by default. Use the matching v2 group when both exist.
## Authentication
We support both the HTTP Basic and Bearer authentication schemes for
convenience.
You first need to generate a new API Key in GrowthBook. Different keys have
different permissions:
- **Personal Access Tokens**: These are sensitive and provide the same level
of access as the user has to an organization, or less if the token's
permissions are limited. These can be created by going to `Personal Access
Tokens` under your user menu. Members with the API Keys **Manage members'
tokens** permission can disable a member's token under `Settings -> Manage
PATs`, and admins can require new tokens to expire within a maximum
lifetime.
- **Secret Keys**: These are sensitive and provide the level of access of
the role they are given, which can be any built-in or custom role, with
optional per-project role overrides and environment limits. Managing Secret
Keys on behalf of an organization takes the [API Keys
permissions](/account/user-permissions). These can be created by going to
`Settings -> API Keys`, where Admins can also require new Secret Keys to
expire within a maximum lifetime.
If using HTTP Basic auth, pass the Secret Key as the username and leave the
password blank (when using curl, add `:` at the end of the secret to
indicate an empty password)
```bash
curl https://api.growthbook.io/api/v1/features \
-u secret_abc123DEF456:
```
If using Bearer auth, pass the Secret Key as the token:
```bash
curl https://api.growthbook.io/api/v1/features \
-H "Authorization: Bearer secret_abc123DEF456"
```
If a key has an expiration date, every response includes it in the
`X-GrowthBook-Key-Expires-At` header (ISO 8601), so scripts can warn before
the key stops working. Requests made with an expired key fail with a `401`
that says where to create a replacement. See
[Expiration](/integrations/ai-agents/api-keys#expiration) for choosing a
key's expiration, the warnings before it lapses, and the organization
policies.
## Naming who asked
Requests made with a Secret Key may add an `X-GrowthBook-Requested-By`
header naming the organization member who asked for the change, as a user id
or an email address. History, revision logs and webhooks show that member
next to the key.
Each key acts as **System** (the default) or **User**, set under `Settings
-> API Keys`:
- **System**: the header is optional. The key acts as itself, with its own
permissions, and the named member is only recorded in history.
- **User**: a request without the header is rejected with a 400. The request
acts as the named member, with only what both the key and the member can do,
so the header can narrow a key but never widen it. Drafts, owners, `mine`
filters and reviews are the member's. The member's rights over their own
drafts, comments and verdicts apply only where the key's own role allows the
same action.
A System key, or any key that names no one, acts as itself: it can draft and
review with its own permissions, and its verdicts are judged on its role. A
key belongs to no team, so its approval never satisfies a rule that requires
approval from a team. No one can rule on a draft they created, and a key
acting as itself can't rule on a draft it created or edited. An approval a
User key gives counts only where both the key and the member could approve,
and stops counting if the key is disabled, expires or is deleted. Because
the header isn't verified, whoever holds the key can act as any member. A
value that does not match a member of the organization is rejected with a
400. A create request that omits `owner` defaults it to the member a User
key names. Scheduled and auto-publishes run later with the permissions of
the request that armed them, and don't run if that member has left the
organization, the key has expired, or the key no longer acts as User.
Personal Access Tokens and OAuth tokens already act as their user and reject
the header.
```bash
curl https://api.growthbook.io/api/v1/features \
-H "Authorization: Bearer secret_abc123DEF456" \
-H "X-GrowthBook-Requested-By: alice@example.com"
```
## Naming and human review
People review what the API creates in the GrowthBook UI.
- Keep keys and names short (2-5 words). Follow your organization's Feature
Flag key format (`featureRegexValidator` and `featureKeyExample` in `GET
/v1/settings`), or use kebab-case if it has none. Feature Flag keys can't be
changed after creation, and an experiment's tracking key shouldn't change
after launch.
- Put context (why, plan, tickets, links) in `description`, not in names or
keys.
- Group with `tags`, not name prefixes. Projects (`project` or `projects`)
control access and where a resource can be used, so set them deliberately
instead of using them as labels.
- To create a Feature Flag disabled with `POST /v2/features`, list every
environment in `environments` with `enabled: false` (on `POST /v1/features`,
each entry also needs `rules: []`). An environment you leave out uses its
default state, which may be enabled.
- Leave publishing and production changes to a person unless asked.
- If a request is blocked (an approval requirement, a warning, a Custom
Hook, schema validation or a permission error), stop and ask a person. Don't
retry with an override field, a different key or another workaround unless a
person asked you to. If a write answers `202` with a `confirmation`, give a
person its `url` and wait (see [Waiting for
confirmation](#waiting-for-confirmation)).
## Waiting for confirmation
A key can require a person to confirm some writes before they run (see [API
Keys for Agents](/integrations/ai-agents/api-keys#require-confirmation)). A
held write doesn't run. It answers `202 Accepted` with the confirmation, a
`Location` header to poll, and a `Retry-After` header with the seconds to
wait between polls:
```json
{
"confirmation": {
"id": "cnf_abc123",
"status": "pending",
"actions": [
{ "action": "feature.publish", "environments": ["production"] }
],
"summary": "Publish revision 3 of checkout-redesign",
"url": "https://app.growthbook.io/confirmations/cnf_abc123",
"expiresAt": "2026-10-15T18:00:00.000Z",
"response": null,
"note": null
},
"message": "A person must confirm this in GrowthBook. Give them the url, then poll the Location."
}
```
Give the person the `url`, then poll [`GET
/v1/confirmations/{id}`](/api/meta/operation/getConfirmation). Pass `wait`
(in seconds, up to 30) to hold the poll open until the status changes:
```bash
curl "https://api.growthbook.io/api/v1/confirmations/cnf_abc123?wait=30" \
-H "Authorization: Bearer secret_abc123DEF456"
```
| `status` |
Meaning
|
| ----------- |
---------------------------------------------------------------------------------------------------------------------
|
| `pending` | Waiting for a person, until `expiresAt` (an hour after the
request) |
| `running` | Confirmed, and running
now
|
| `completed` | It ran. `response` holds the status and body the request
returned, which may be an error |
| `rejected` | A person declined it, and nothing
ran
|
| `expired` | Nobody decided in time, or it acts on something that changed
after it was held (with a `400` `response`). Nothing ran |
`note` carries anything the person left for the agent when they decided.
`actions` lists what the request was held for, each with the environments it
touches.
- Only the key that made the request, or another token of the same person,
can read the confirmation.
- Sending the same request again while it's pending returns the same
confirmation.
- A request that publishes, reverts or archives, or that changes a Feature
Flag or an Experiment's status, schedule, targeting or traffic, isn't run if
what it acts on changed after it was held. Other writes, such as editing
settings or a draft, run as sent, so they can overwrite changes made while
they waited.
- A confirmed request runs once, with the key's permissions (or the member's
role it assumes), not the confirmer's. Send it again to retry, which asks
for a new confirmation.
- Confirming isn't a review approval.
## Errors
The API may return the following error status codes:
- **400** - Bad Request - Often due to a missing required parameter
- **401** - Unauthorized - No valid API key provided
- **402** - Request Failed - The parameters are valid, but the request
failed
- **403** - Forbidden - Provided API key does not have the required access
- **404** - Not Found - Unknown API route or requested resource
- **422** - Unprocessable Entity - The request is valid, but a warning,
validation rule, approval requirement, or another publishing gate blocked
it. Do not assume that `ignoreWarnings` clears every 422 response.
- **429** - Too Many Requests - You exceeded the rate limit of 60 requests
per minute, or the limit your self-hosted server sets. Try again later.
- **5XX** - Server Error - Something went wrong on GrowthBook's end (these
are rare)
The response body will be a JSON object with the following properties:
- **message** - Information about the error
### Publishing gates
Automated callers, including AI agents, should report a blocked response to
a person instead of retrying with an override. Use an override only when a
person has explicitly asked for it.
Publish responses include a `gates` array that explains every blocker:
- `type`, `severity`, and `messages` identify the problem.
- `override` names the request-body field that can bypass it. This is
`ignoreWarnings` for warnings, `skipSchemaValidation` for schema and
invariant failures, or `skipHooks` for Custom Hook rejections. A value of
`null` means there is no request-body override.
- `requiresPermission` identifies any additional permission needed to use
the override.
- `resolution` provides an API action, method, and path when the blocker
must be resolved another way.
For example, an approval gate is cleared by approving the revision or by
using a caller with **Bypass draft approvals** access. A Config lock is
cleared through the unlock route in `resolution`.
When a successful publish bypasses a gate, the response includes
`bypassedGates`. Each entry reports the gate `type` and how it was bypassed
in `via`, which is one of `ignoreWarnings`, `skipSchemaValidation`,
`skipHooks`, `bypassApprovalPermission`, `restApiBypassesReviews`, or
`revertsBypassApproval` (reverts only). This field is omitted when no gates
were bypassed.
servers:
- url: https://api.growthbook.io/api
description: GrowthBook Cloud
- url: https://{domain}/api
description: Self-hosted GrowthBook
variables:
domain:
default: localhost:3100
description: Your self-hosted GrowthBook host (and port)
security:
- bearerAuth: []
- basicAuth: []
tags:
- name: projects
x-displayName: Projects
description: Projects are used to organize your feature flags and experiments
- name: environments
x-displayName: Environments
description: >-
GrowthBook comes with one environment by default (production), but you can
add as many as you need. When used with feature flags, you can
enable/disable feature flags on a per-environment basis.
- name: features-v2
x-displayName: Feature Flags
description: >-
Control your feature flags programmatically.
Rules are returned as a unified top-level array; each rule carries
`allEnvironments` / `environments` scope fields instead of being bucketed
by environment.
- name: feature-revisions-v2
x-displayName: Feature Revisions
description: >-
Draft revisions for feature flags, including rules, scheduling, and
approval workflows.
Revision `rules` is a flat array with per-rule scope fields.
- name: features
x-displayName: Feature Flags (legacy)
description: >-
Control your feature flags programmatically.
**These are v1 endpoints.** New integrations should use the v2 Feature
Flags endpoints, which expose a unified per-rule environment scope instead
of per-environment rule arrays.
- name: feature-revisions
x-displayName: Feature Revisions (legacy)
description: >-
Draft revisions for feature flags, including rules, scheduling, and
approval workflows.
**These are v1 endpoints.** New integrations should use the v2 Feature
Revisions endpoints.
- name: ramp-schedules
x-displayName: Ramp Schedules
description: >-
Multi-step rollout schedules that gradually increase feature rule traffic
over time, with optional real-time monitoring. Each step supports interval
timers, approval gates, and hold conditions. Monitored steps are backed by
a live analysis experiment that can automatically hold, roll back, or
advance the ramp based on guardrail and signal metric health.
- name: data-sources
x-displayName: Data Sources
description: >-
How GrowthBook connects and queries your data, including cached database
schema metadata (information schemas) for tables and columns.
- name: fact-tables
x-displayName: Fact Tables
description: Fact Tables describe the shape of your data warehouse tables
- name: fact-metrics
x-displayName: Fact Metrics
description: Fact Metrics are metrics built on top of Fact Table definitions
- name: metrics
x-displayName: Metrics (legacy)
description: Metrics used as goals and guardrails for experiments
- name: experiments
x-displayName: Experiments
description: Experiments (A/B Tests)
- name: namespaces
x-displayName: Namespaces
description: >-
Namespaces partition your user population into buckets so that experiments
using the same hash attribute do not overlap unintentionally. Each
namespace defines a 0–1 range and individual experiments claim sub-ranges
within it.
- name: snapshots
x-displayName: Experiment Snapshots
description: Experiment Snapshots (the individual updates of an experiment)
- name: dimensions
x-displayName: Dimensions
description: Dimensions used during experiment analysis
- name: segments
x-displayName: Segments
description: Segments used during experiment analysis
- name: reports
x-displayName: Experiment Reports
description: >-
Custom analysis reports built on top of experiment snapshots. Reports let
you re-run analysis with different metrics, date ranges, stats engines,
and other settings without modifying the underlying experiment.
- name: sdk-connections
x-displayName: SDK Connections
description: Client keys and settings for connecting SDKs to a GrowthBook instance
- name: visual-changesets
x-displayName: Visual Changesets
description: Groups of visual changes made by the visual editor to a single page
- name: saved-groups
x-displayName: Saved Groups
description: >-
Defined sets of attribute values which can be used with feature rules for
targeting features at particular users.
- name: saved-group-revisions
x-displayName: Saved Group Revisions
description: >-
Draft revisions for saved groups, including pending changes, approvals,
and lifecycle (publish, discard, revert).
Most callers can interact with these endpoints via shorthand actions
(`/items/add`, `/items/remove`, single-field PUTs) instead of authoring
JSON Patch ops directly. Pass `version: "new"` on edit endpoints to
auto-create a draft.
- name: constants
x-displayName: Constants
description: >-
**Beta** — these endpoints are new and may change in
backwards-incompatible ways.
Reusable named values referenced from feature flag values as `@const:key`
and resolved into the SDK payload at build time. String constants are
interpolated via `{{ @const:key }}`; JSON (object) constants are composed
via an `$extends` array. A constant's own keys **replace** what its
`$extends` bases provide, wholesale — constants are atomic building
blocks. (Config and feature values compose as deep, targeted patches
instead.)
- name: constant-revisions
x-displayName: Constant Revisions
description: >-
**Beta** — these endpoints are new and may change in
backwards-incompatible ways.
Draft revisions for constants, including pending changes, approvals, and
lifecycle (publish, discard, revert). Pass `version: "new"` on edit
endpoints to auto-create a draft.
- name: configs
x-displayName: Configs
description: >-
**Beta** — these endpoints are new and may change in
backwards-incompatible ways.
Reusable, typed, inheritable JSON objects referenced from feature flag
values as `@config:key`. A config carries a field `schema` (with
TypeScript/JSON Schema import-export) and a lineage `parent`. Inheritance
is expressed via `parent`, never an in-value `@config:` entry. Values
layer as a **deep, targeted patch**: a child (or a config-backed feature
value) restates only the leaves it changes and inherits the rest — unlike
a constant's `$extends`, whose own keys replace wholesale. Schema fields
colliding with a published ancestor's key follow 'base wins': identical
re-declarations are stripped with a warning, differing ones are rejected.
- name: config-revisions
x-displayName: Config Revisions
description: >-
**Beta** — these endpoints are new and may change in
backwards-incompatible ways.
Draft revisions for configs, including value and schema edits, schema
import (JSON Schema / TypeScript / inferred), approvals, and lifecycle
(publish, discard, revert). Publishing a schema change cascades the "base
wins" normalization to descendant configs; a publish that removes or
retypes fields descendants still use soft-blocks with a 422 unless the
request body sets `ignoreWarnings: true`. Pass `version: "new"` on edit
endpoints to auto-create a draft.
- name: releases
x-displayName: Releases
description: >-
**Beta** — these endpoints are new and may change in
backwards-incompatible ways.
Coordinated multi-entity publishing: publish a set of revisions across
Feature Flags, Saved Groups, configs, and constants as one all-or-nothing
operation, validated against the combined end-state instead of each
in-between state. Requires the `releases` commercial feature.
- name: custom-hooks
x-displayName: Custom Hooks
description: >-
Sandboxed JavaScript validation hooks that run when features, configs, or
their revisions are saved or published. Throwing an Error blocks the save;
`addWarning(msg)` raises a soft warning. Hooks are scoped by projects, or
pinned to a single feature/config via `entityType`/`entityId`; a
config-scoped hook also runs for every config inheriting from it (its
whole descendant lineage). Scope can be retargeted on update (or cleared
with nulls). Requires an enterprise plan; not available on GrowthBook
Cloud.
- name: organizations
x-displayName: Organizations
description: >-
Organizations are used for multi-org deployments where different teams can
run their own isolated feature flags and experiments. These endpoints are
only via a super-admin's Personal Access Token.
- name: members
x-displayName: Members
description: Members are users who have been invited to an organization.
- name: code-references
x-displayName: Code References
description: >-
Intended for use with our code reference CI utility,
[`gb-find-code-refs`](https://github.com/growthbook/gb-find-code-refs).
- name: archetypes
x-displayName: Archetypes
description: >-
Archetypes allow you to simulate the result of targeting rules on pre-set
user attributes
- name: queries
x-displayName: Queries
description: Retrieve queries used in experiments to calculate results.
- name: settings
x-displayName: Settings
description: Get the organization settings.
- name: attributes
x-displayName: Attributes
description: Used when targeting feature flags and experiments.
- name: usage
x-displayName: Usage
description: Usage information for metrics in experiments.
- name: meta
x-displayName: Meta
description: >-
Server metadata, including the running build's version and commit for
version-skew checks.
- name: ContextualBandits
x-displayName: Contextual Bandits
description: ''
- name: Dashboards
x-displayName: Dashboards
description: ''
- name: ContextualBanditQueries
x-displayName: Contextual Bandit Queries
description: ''
- name: Interleavings
x-displayName: Interleaving Experiments
description: ''
- name: CustomFields
x-displayName: Custom Fields
description: ''
- name: MetricGroups
x-displayName: Metric Groups
description: ''
- name: Teams
x-displayName: Teams
description: ''
- name: ExperimentTemplates
x-displayName: Experiment Templates
description: ''
- name: AnalyticsExplorations
x-displayName: Analytics Explorations
description: ''
- name: RampScheduleTemplates
x-displayName: Ramp Schedule Templates
description: Reusable step configurations for ramp schedules.
- name: Learnings
x-displayName: Learnings
description: >-
Saved learnings captured across experiments, including AI-discovered
patterns.
- name: Holdouts
x-displayName: Holdouts
description: >-
Hold a share of traffic out of all experiments to measure their combined
effect.
- name: AutoRuns
x-displayName: Auto Runs
description: ''
- name: Populations
x-displayName: Populations
description: >-
**Beta** — these endpoints are new and may change in
backwards-incompatible ways.
Populations define a set of units by the steps they must complete, such as
appearing in a fact table. Populations are not yet used in analysis.
- name: AggregatedFactTable_model
x-displayName: Aggregated Fact Table
description:
- name: AnalyticsExploration_model
x-displayName: Analytics Exploration
description:
- name: Archetype_model
x-displayName: Archetype
description:
- name: Attribute_model
x-displayName: Attribute
description:
- name: AutoRun_model
x-displayName: Auto Run
description:
- name: CodeRef_model
x-displayName: Code Ref
description:
- name: Config_model
x-displayName: Config
description:
- name: ConfigKeyUsage_model
x-displayName: Config Key Usage
description:
- name: ConfigLineage_model
x-displayName: Config Lineage
description:
- name: ConfigReferences_model
x-displayName: Config References
description:
- name: ConfigRevision_model
x-displayName: Config Revision
description:
- name: ConfigRevisionActivityLogEntry_model
x-displayName: Config Revision Activity Log Entry
description: >-
- name: ConfigRevisionRef_model
x-displayName: Config Revision Ref
description:
- name: ConfigRevisionReview_model
x-displayName: Config Revision Review
description:
- name: ConfigSchemaExport_model
x-displayName: Config Schema Export
description:
- name: ConfigSchemaSource_model
x-displayName: Config Schema Source
description:
- name: ConfigSchemaVerify_model
x-displayName: Config Schema Verify
description:
- name: ConfigSchemaWarning_model
x-displayName: Config Schema Warning
description:
- name: Confirmation_model
x-displayName: Confirmation
description:
- name: ConfirmationHeld_model
x-displayName: Confirmation Held
description:
- name: Constant_model
x-displayName: Constant
description:
- name: ConstantReferences_model
x-displayName: Constant References
description:
- name: ConstantRevision_model
x-displayName: Constant Revision
description:
- name: ConstantRevisionActivityLogEntry_model
x-displayName: Constant Revision Activity Log Entry
description: >-
- name: ConstantRevisionRef_model
x-displayName: Constant Revision Ref
description:
- name: ConstantRevisionReview_model
x-displayName: Constant Revision Review
description: >-
- name: ContextualBandit_model
x-displayName: Contextual Bandit
description:
- name: ContextualBanditQuery_model
x-displayName: Contextual Bandit Query
description: >-
- name: CustomField_model
x-displayName: Custom Field
description:
- name: CustomHook_model
x-displayName: Custom Hook
description:
- name: Dashboard_model
x-displayName: Dashboard
description:
- name: DataSource_model
x-displayName: Data Source
description:
- name: Dimension_model
x-displayName: Dimension
description:
- name: Environment_model
x-displayName: Environment
description:
- name: EventUser_model
x-displayName: Event User
description:
- name: Experiment_model
x-displayName: Experiment
description:
- name: Experiment Rule_model
x-displayName: Experiment Rule
description:
- name: ExperimentAnalysisSettings_model
x-displayName: Experiment Analysis Settings
description: >-
- name: ExperimentDecisionFrameworkSettings_model
x-displayName: Experiment Decision Framework Settings
description: >-
- name: ExperimentMetric_model
x-displayName: Experiment Metric
description:
- name: ExperimentMetricOverrideEntry_model
x-displayName: Experiment Metric Override Entry
description: >-
- name: ExperimentResults_model
x-displayName: Experiment Results
description:
- name: ExperimentSnapshot_model
x-displayName: Experiment Snapshot
description:
- name: ExperimentTemplate_model
x-displayName: Experiment Template
description:
- name: ExperimentWithEnhancedStatus_model
x-displayName: Experiment With Enhanced Status
description: >-
- name: FactMetric_model
x-displayName: Fact Metric
description:
- name: FactTable_model
x-displayName: Fact Table
description:
- name: FactTableColumn_model
x-displayName: Fact Table Column
description:
- name: FactTableFilter_model
x-displayName: Fact Table Filter
description:
- name: FeatureBaseRule_model
x-displayName: Feature Base Rule
description:
- name: FeatureContextualBanditRefRule_model
x-displayName: Feature Contextual Bandit Ref Rule
description: >-
- name: FeatureDefinition_model
x-displayName: Feature Definition
description:
- name: FeatureEnvironmentV1_model
x-displayName: Feature Environment V1
description:
- name: FeatureEnvironmentV2_model
x-displayName: Feature Environment V2
description:
- name: FeatureExperimentRefRule_model
x-displayName: Feature Experiment Ref Rule
description: >-
- name: FeatureExperimentRule_model
x-displayName: Feature Experiment Rule
description: >-
- name: FeatureForceRule_model
x-displayName: Feature Force Rule
description:
- name: FeatureRevisionRef_model
x-displayName: Feature Revision Ref
description:
- name: FeatureRevisionSummary_model
x-displayName: Feature Revision Summary
description: >-
- name: FeatureRevisionV1_model
x-displayName: Feature Revision V1
description:
- name: FeatureRevisionV2_model
x-displayName: Feature Revision V2
description:
- name: FeatureRolloutRule_model
x-displayName: Feature Rollout Rule
description:
- name: FeatureRuleV1_model
x-displayName: Feature Rule V1
description:
- name: FeatureRuleV2_model
x-displayName: Feature Rule V2
description:
- name: FeatureSafeRolloutRule_model
x-displayName: Feature Safe Rollout Rule
description: >-
- name: FeatureV1_model
x-displayName: Feature V1
description:
- name: FeatureV2_model
x-displayName: Feature V2
description:
- name: FeatureWithRevisionsV1_model
x-displayName: Feature With Revisions V1
description: >-
- name: FeatureWithRevisionsV2_model
x-displayName: Feature With Revisions V2
description: >-
- name: Holdout_model
x-displayName: Holdout
description:
- name: InformationSchema_model
x-displayName: Information Schema
description:
- name: InformationSchemaTable_model
x-displayName: Information Schema Table
description: >-
- name: Interleaving_model
x-displayName: Interleaving
description:
- name: Learning_model
x-displayName: Learning
description:
- name: LookbackOverride_model
x-displayName: Lookback Override
description:
- name: Member_model
x-displayName: Member
description:
- name: Metric_model
x-displayName: Metric
description:
- name: MetricAnalysis_model
x-displayName: Metric Analysis
description:
- name: MetricGroup_model
x-displayName: Metric Group
description:
- name: MetricUsage_model
x-displayName: Metric Usage
description:
- name: Namespace_model
x-displayName: Namespace
description:
- name: NamespaceExperimentMember_model
x-displayName: Namespace Experiment Member
description: >-
- name: Organization_model
x-displayName: Organization
description:
- name: PaginationFields_model
x-displayName: Pagination Fields
description:
- name: Population_model
x-displayName: Population
description:
- name: Project_model
x-displayName: Project
description:
- name: Query_model
x-displayName: Query
description:
- name: RampSchedule_model
x-displayName: Ramp Schedule
description:
- name: RampScheduleTemplate_model
x-displayName: Ramp Schedule Template
description:
- name: Report_model
x-displayName: Report
description:
- name: RequireReviewRule_model
x-displayName: Require Review Rule
description:
- name: RequireReviewRuleInput_model
x-displayName: Require Review Rule Input
description: >-
- name: RevisionIdRef_model
x-displayName: Revision Id Ref
description:
- name: Safe Rollout Rule_model
x-displayName: Safe Rollout Rule
description:
- name: SavedGroup_model
x-displayName: Saved Group
description:
- name: SavedGroupApprovalRule_model
x-displayName: Saved Group Approval Rule
description: >-
- name: SavedGroupReferenceResource_model
x-displayName: Saved Group Reference Resource
description: >-
- name: SavedGroupReferences_model
x-displayName: Saved Group References
description:
- name: SavedGroupRevision_model
x-displayName: Saved Group Revision
description:
- name: SavedGroupRevisionActivityLogEntry_model
x-displayName: Saved Group Revision Activity Log Entry
description: >-
- name: SavedGroupRevisionRef_model
x-displayName: Saved Group Revision Ref
description: >-
- name: SavedGroupRevisionReview_model
x-displayName: Saved Group Revision Review
description: >-
- name: ScheduleRule_model
x-displayName: Schedule Rule
description:
- name: ScheduledStopPlan_model
x-displayName: Scheduled Stop Plan
description:
- name: SdkConnection_model
x-displayName: Sdk Connection
description:
- name: Segment_model
x-displayName: Segment
description:
- name: Settings_model
x-displayName: Settings
description:
- name: Targeting Rule_model
x-displayName: Targeting Rule
description:
- name: TargetingReviewRule_model
x-displayName: Targeting Review Rule
description:
- name: Team_model
x-displayName: Team
description:
- name: VisualChange_model
x-displayName: Visual Change
description:
- name: VisualChangeset_model
x-displayName: Visual Changeset
description:
- name: VisualEditorCbExperimentStub_model
x-displayName: Visual Editor Cb Experiment Stub
description: >-
paths:
/v1/experiments/{id}/stop:
post:
tags:
- experiments
summary: Stop an experiment
operationId: postExperimentStop
parameters:
- $ref: '#/components/parameters/id'
requestBody:
required: true
content:
application/json:
schema:
type: object
properties:
results:
description: The experiment conclusion status.
type: string
enum:
- dnf
- won
- lost
- inconclusive
enableTemporaryRollout:
description: >-
If true, include this stopped experiment in SDK payload and
force the release variation (`releasedVariationId`) to all
traffic.
type: boolean
releasedVariationId:
description: >-
Required if enableTemporaryRollout is true. Variation ID
(e.g. var_abc123) to release to 100% of traffic eligible for
this experiment.
type: string
winnerVariationId:
description: >-
Variation ID (e.g. var_abc123) of the winning variation.
Used only as metadata. Required if results is 'won' and
there are multiple test variations. Otherwise, defaults to
the test variation when results is 'won' and to the baseline
variation for other results.
type: string
analysis:
description: >-
Optional markdown summary displayed on the experiment
results page.
type: string
reason:
description: >-
Optional reason for ending the phase stored on the latest
phase metadata.
type: string
dateEnded:
description: >-
Optional ISO datetime for ending the latest phase. Defaults
to the current date and time.
type: string
ignoreWarnings:
description: >-
Set to true to acknowledge the warnings listed in a blocked
response and continue. This covers experiment guards, locked
dependents, and references affected by an archive. When the
organization treats schema failures as warnings, it also
covers schema and invariant warnings. It never bypasses a
rejected Custom Hook. On revision publish endpoints, it can
also force-publish an out-of-date draft when the caller has
Bypass draft approvals access. Only set this when a person
has explicitly asked you to. If a check blocks you, stop and
ask a person instead of working around it.
type: boolean
required:
- results
additionalProperties: false
responses:
'200':
description: Resource created
content:
application/json:
schema:
type: object
properties:
experiment:
$ref: '#/components/schemas/ExperimentWithEnhancedStatus'
required:
- experiment
additionalProperties: false
'202':
$ref: '#/components/responses/ConfirmationHeld'
'409':
description: Conflict
content:
application/json:
schema:
type: object
description: Resource is not in a valid status for this operation
properties:
message:
type: string
code:
type: string
enum:
- invalid_status
details:
$ref: '#/components/schemas/ApiErrorInvalidStatusDetails'
required:
- message
- code
- details
x-codeSamples:
- lang: cURL
source: >-
curl -X POST
'https://api.growthbook.io/api/v1/experiments/exp_abc123/stop' \
-H 'Authorization: Bearer YOUR_API_KEY' \
-H 'Content-Type: application/json' \
-d '{"results":"won","releasedVariationId":"var_treatment","winnerVariationId":"var_treatment","enableTemporaryRollout":true,"analysis":"Reached desired sample size with statistically significant positive lift; shipping treatment"}'
components:
parameters:
id:
name: id
in: path
required: true
description: The id of the requested resource
schema:
description: The id of the requested resource
type: string
schemas:
ExperimentWithEnhancedStatus:
allOf:
- type: object
properties:
id:
type: string
trackingKey:
type: string
dateCreated:
format: date-time
type: string
dateUpdated:
format: date-time
type: string
name:
type: string
type:
type: string
enum:
- standard
- multi-armed-bandit
- holdout
project:
type: string
hypothesis:
type: string
description:
type: string
maxLength: 10000
tags:
type: array
items:
type: string
owner:
description: >-
The userId of the owner (or raw owner name/email for legacy
records)
type: string
ownerEmail:
description: >-
The email address of the owner, when the owner can be resolved
to a known user.
type: string
archived:
type: boolean
status:
type: string
autoRefresh:
type: boolean
hashAttribute:
type: string
fallbackAttribute:
type: string
hashVersion:
anyOf:
- type: number
const: 1
- type: number
const: 2
disableStickyBucketing:
type: boolean
bucketVersion:
type: number
minBucketVersion:
type: number
variations:
type: array
items:
type: object
properties:
variationId:
type: string
key:
type: string
name:
type: string
description:
type: string
maxLength: 10000
screenshots:
type: array
items:
type: string
required:
- variationId
- key
- name
- description
- screenshots
additionalProperties: false
phases:
type: array
items:
type: object
properties:
name:
type: string
dateStarted:
type: string
dateEnded:
type: string
reasonForStopping:
type: string
seed:
type: string
coverage:
type: number
trafficSplit:
type: array
items:
type: object
properties:
variationId:
type: string
weight:
type: number
required:
- variationId
- weight
additionalProperties: false
namespace:
type: object
properties:
namespaceId:
type: string
enabled:
type: boolean
range:
minItems: 2
maxItems: 2
type: array
items:
type: number
ranges:
type: array
items:
type: array
prefixItems:
- type: number
- type: number
required:
- namespaceId
additionalProperties: false
targetingCondition:
type: string
prerequisites:
type: array
items:
type: object
properties:
id:
type: string
condition:
type: string
required:
- id
- condition
additionalProperties: false
savedGroupTargeting:
type: array
items:
type: object
properties:
matchType:
type: string
enum:
- all
- any
- none
savedGroups:
type: array
items:
type: string
required:
- matchType
- savedGroups
additionalProperties: false
required:
- name
- dateStarted
- dateEnded
- reasonForStopping
- seed
- coverage
- trafficSplit
- targetingCondition
additionalProperties: false
settings:
$ref: '#/components/schemas/ExperimentAnalysisSettings'
resultSummary:
type: object
properties:
status:
type: string
winner:
type: string
conclusions:
type: string
releasedVariationId:
type: string
excludeFromPayload:
type: boolean
required:
- status
- winner
- conclusions
- releasedVariationId
- excludeFromPayload
additionalProperties: false
shareLevel:
type: string
enum:
- public
- organization
publicUrl:
type: string
banditScheduleValue:
type: number
banditScheduleUnit:
type: string
enum:
- days
- hours
banditBurnInValue:
type: number
banditBurnInUnit:
type: string
enum:
- days
- hours
banditConversionWindowValue:
type: number
banditConversionWindowUnit:
type: string
enum:
- days
- hours
linkedFeatures:
type: array
items:
type: string
attributeScopeAllProjects:
type: boolean
hasVisualChangesets:
type: boolean
hasURLRedirects:
type: boolean
customFields:
type: object
propertyNames:
type: string
additionalProperties: {}
customMetricSlices:
description: >-
Custom slices that apply to ALL applicable metrics in the
experiment
type: array
items:
type: object
properties:
slices:
type: array
items:
type: object
properties:
column:
type: string
levels:
type: array
items:
type: string
required:
- column
- levels
additionalProperties: false
required:
- slices
additionalProperties: false
precomputedUnitDimensionIds:
maxItems: 3
type: array
items:
type: string
defaultDashboardId:
description: ID of the default dashboard for this experiment.
type: string
templateId:
type: string
statusUpdateSchedule:
anyOf:
- description: >-
Scheduled start/end for an experiment. All fields optional;
the end may be an absolute `stopAt` or a deferred relative
`stopAfter`, but not both.
type: object
properties:
startAt:
description: >-
ISO datetime when the experiment should start. Must be
in the future. Setting or clearing this field
invalidates any existing staged start
(`nextScheduledStatusUpdate`); call POST
/experiments/{id}/start to stage the new schedule.
format: date-time
type: string
stopAt:
description: >-
ISO datetime when the experiment should stop. Resolved
from `stopAfter` at start when a relative end was set.
format: date-time
type: string
stopAfter:
description: >-
Relative end offset. Deferred: resolved to a concrete
`stopAt` at the experiment's actual start (or off
`dateStarted` when already running).
type: object
properties:
value:
type: integer
exclusiveMinimum: 0
unit:
type: string
enum:
- hours
- days
required:
- value
- unit
additionalProperties: false
scheduledStopPlan:
$ref: '#/components/schemas/ScheduledStopPlan'
description: >-
End-of-experiment automation applied at the scheduled
end. Required whenever a `stopAt` or `stopAfter` is set
(any mode, including `notify`).
additionalProperties: false
- type: 'null'
nextScheduledStatusUpdate:
anyOf:
- type: object
properties:
type:
type: string
enum:
- start
- stop
date:
format: date-time
type: string
required:
- type
- date
additionalProperties: false
- type: 'null'
required:
- id
- trackingKey
- dateCreated
- dateUpdated
- name
- type
- project
- hypothesis
- description
- tags
- owner
- archived
- status
- autoRefresh
- hashAttribute
- hashVersion
- variations
- phases
- settings
additionalProperties: false
- type: object
properties:
enhancedStatus:
type: object
properties:
status:
type: string
enum:
- Running
- Stopped
- Draft
- Scheduled
- Archived
detailedStatus:
type: string
required:
- status
additionalProperties: false
additionalProperties: false
ApiErrorInvalidStatusDetails:
type: object
properties:
currentStatus:
type: string
expectedStatuses:
type: array
items:
type: string
required:
- currentStatus
- expectedStatuses
additionalProperties: false
ExperimentAnalysisSettings:
type: object
properties:
datasourceId:
type: string
assignmentQuery:
description: >-
The assignment query, grouping its ID with the identifier type
analyzed on.
type: object
properties:
id:
description: The ID of one of the data source's assignment queries.
type: string
identifierType:
description: >-
The identifier type analyzed on. Null when none can be resolved:
no assignment query is selected, or a record saved before
identifier types were stored points at a query that no longer
exists.
anyOf:
- type: string
- type: 'null'
required:
- id
- identifierType
additionalProperties: false
assignmentQueryId:
description: 'Deprecated: use `assignmentQuery`.'
deprecated: true
type: string
experimentId:
type: string
segmentId:
type: string
queryFilter:
type: string
inProgressConversions:
type: string
enum:
- include
- exclude
attributionModel:
description: >-
Setting attribution model to `"experimentDuration"` is the same as
selecting "Ignore Conversion Windows" for the Conversion Window
Override. Setting it to `"lookbackOverride"` requires a
`lookbackOverride` object to be provided.
type: string
enum:
- firstExposure
- experimentDuration
- lookbackOverride
lookbackOverride:
$ref: '#/components/schemas/LookbackOverride'
statsEngine:
type: string
enum:
- bayesian
- frequentist
regressionAdjustmentEnabled:
type: boolean
sequentialTestingEnabled:
type: boolean
sequentialTestingTuningParameter:
type: number
postStratificationEnabled:
description: When null, the organization default is used.
anyOf:
- description: When null, the organization default is used.
type: boolean
- description: When null, the organization default is used.
type: 'null'
decisionFrameworkSettings:
$ref: '#/components/schemas/ExperimentDecisionFrameworkSettings'
description: >-
Controls the decision framework and metric overrides for the
experiment. Replaces the entire stored object on update (does not
patch individual fields).
metricOverrides:
description: >-
Per-metric analysis overrides; also reflected in
goals/secondaryMetrics/guardrails overrides when applicable. On
create/update, this replaces the entire stored array (it does not
patch individual entries).
type: array
items:
$ref: '#/components/schemas/ExperimentMetricOverrideEntry'
goals:
type: array
items:
$ref: '#/components/schemas/ExperimentMetric'
secondaryMetrics:
type: array
items:
$ref: '#/components/schemas/ExperimentMetric'
guardrails:
type: array
items:
$ref: '#/components/schemas/ExperimentMetric'
activationMetric:
$ref: '#/components/schemas/ExperimentMetric'
required:
- datasourceId
- assignmentQuery
- assignmentQueryId
- experimentId
- segmentId
- queryFilter
- inProgressConversions
- attributionModel
- statsEngine
- goals
- secondaryMetrics
- guardrails
additionalProperties: false
ScheduledStopPlan:
description: >-
What happens at the scheduled end date. `notify` keeps the experiment
running and just notifies (soft). `auto-ship` (requires the Decision
Framework) ships the winning variation and stops; multi-winner ties
break on `tiebreakerMetricId` (higher lift); with no clear winner,
`fallback` either keeps running (`notify`) or ships
`fallbackVariationId`. `force-ship` stops and rolls out
`fallbackVariationId`. `stop` is a hard deadline that stops with no
rollout. For `force-ship` and `stop`, the Decision Framework verdict
(won/lost/inconclusive) is recorded as metadata when available.
type: object
properties:
mode:
type: string
enum:
- notify
- auto-ship
- force-ship
- stop
tiebreakerMetricId:
type: string
fallback:
description: >-
What to do at the scheduled end when there is no clear winner.
Required when `mode` is `"auto-ship"`; ignored for other modes.
type: string
enum:
- notify
- force-ship
fallbackVariationId:
type: string
required:
- mode
additionalProperties: false
ConfirmationHeld:
type: object
properties:
confirmation:
$ref: '#/components/schemas/Confirmation'
message:
type: string
required:
- confirmation
- message
additionalProperties: false
$defs:
Confirmation:
id: Confirmation
type: object
properties:
id:
type: string
status:
type: string
enum:
- pending
- running
- completed
- rejected
- expired
actions:
type: array
items:
type: object
properties:
action:
type: string
enum:
- feature.create
- feature.draft
- feature.review
- feature.publish
- feature.archive
- feature.delete
- feature.other
- experiment.create
- experiment.draft
- experiment.start
- experiment.stop
- experiment.targeting
- experiment.traffic
- experiment.phases
- experiment.visualEditor
- experiment.other
- savedGroup.create
- savedGroup.draft
- savedGroup.review
- savedGroup.publish
- savedGroup.archive
- savedGroup.delete
- savedGroup.other
- constant.create
- constant.draft
- constant.review
- constant.publish
- constant.archive
- constant.delete
- constant.other
- config.create
- config.draft
- config.review
- config.publish
- config.archive
- config.delete
- config.other
- rampSchedule.create
- rampSchedule.start
- rampSchedule.pause
- rampSchedule.progression
- rampSchedule.approve
- rampSchedule.edit
- rampSchedule.delete
- rampSchedule.other
- settings.approvals
- settings.other
- access.members
- access.teams
- customHook.create
- customHook.edit
- customHook.delete
- environment.create
- environment.edit
- environment.delete
- project.create
- project.edit
- project.delete
- attribute.create
- attribute.edit
- attribute.delete
- customField.create
- customField.edit
- customField.delete
- sdkConnection.create
- sdkConnection.edit
- sdkConnection.delete
- metric.create
- metric.edit
- metric.delete
- factTable.create
- factTable.edit
- factTable.delete
- segment.create
- segment.edit
- segment.delete
- dimension.create
- dimension.edit
- dimension.delete
- override.ignoreWarnings
- override.skipSchemaValidation
- override.skipHooks
- override.bypassApproval
- override.restBypassesReviews
- override.draftLimit
- override.skipChecklist
- override.bypassDuplicateKeyCheck
- override.unscopedRules
- other.write
environments:
type: array
items:
type: string
required:
- action
- environments
additionalProperties: false
summary:
type: string
url:
type: string
expiresAt:
type: string
response:
anyOf:
- type: object
properties:
status:
type: number
body: {}
required:
- status
- body
additionalProperties: false
- type: 'null'
note:
anyOf:
- type: string
- type: 'null'
required:
- id
- status
- actions
- summary
- url
- expiresAt
- response
- note
additionalProperties: false
LookbackOverride:
description: >-
Controls the lookback override for the experiment. For type "window",
value must be a non-negative number and valueUnit is required.
type: object
properties:
type:
type: string
enum:
- date
- window
value:
description: >-
For "window" type - non-negative numeric value (e.g. 7 for 7 days).
For "date" type a date string.
anyOf:
- description: >-
For "window" type - non-negative numeric value (e.g. 7 for 7
days). For "date" type a date string.
type: number
- format: date-time
description: >-
For "window" type - non-negative numeric value (e.g. 7 for 7
days). For "date" type a date string.
type: string
valueUnit:
description: Used when type is "window". Defaults to "days".
type: string
enum:
- minutes
- hours
- days
- weeks
required:
- type
- value
additionalProperties: false
ExperimentDecisionFrameworkSettings:
description: >-
Controls the decision framework and metric overrides for the experiment.
Replaces the entire stored object on update (does not patch individual
fields).
type: object
properties:
decisionCriteriaId:
type: string
decisionFrameworkMetricOverrides:
type: array
items:
type: object
properties:
id:
description: ID of the metric to override settings for.
type: string
targetMDE:
description: >-
The target relative MDE to use for the metric, expressed as
proportions (e.g. use 0.1 for 10%). Must be greater than 0.
type: number
exclusiveMinimum: 0
required:
- id
additionalProperties: false
additionalProperties: false
ExperimentMetricOverrideEntry:
description: >-
Per-metric analysis overrides stored on the experiment (matches internal
metricOverrides).
type: object
properties:
id:
description: ID of the metric to override settings for.
type: string
windowType:
type: string
enum:
- conversion
- lookback
- ''
windowHours:
type: number
delayHours:
type: number
properPriorOverride:
description: >-
Must be true for the override to take effect. If true, the other
proper prior settings in this object will be used if present.
type: boolean
properPriorEnabled:
type: boolean
properPriorMean:
type: number
properPriorStdDev:
type: number
regressionAdjustmentOverride:
description: >-
Must be true for the override to take effect. If true, the other
regression adjustment settings in this object will be used if
present.
type: boolean
regressionAdjustmentEnabled:
type: boolean
regressionAdjustmentDays:
type: number
required:
- id
additionalProperties: false
ExperimentMetric:
type: object
properties:
metricId:
type: string
overrides:
type: object
properties:
delayHours:
type: number
windowHours:
type: number
window:
type: string
enum:
- conversion
- lookback
- ''
winRiskThreshold:
deprecated: true
type: number
loseRiskThreshold:
deprecated: true
type: number
properPriorOverride:
type: boolean
properPriorEnabled:
type: boolean
properPriorMean:
type: number
properPriorStdDev:
type: number
regressionAdjustmentOverride:
type: boolean
regressionAdjustmentEnabled:
type: boolean
regressionAdjustmentDays:
type: number
additionalProperties: false
required:
- metricId
- overrides
additionalProperties: false
Confirmation:
type: object
properties:
id:
type: string
status:
type: string
enum:
- pending
- running
- completed
- rejected
- expired
actions:
type: array
items:
type: object
properties:
action:
type: string
enum:
- feature.create
- feature.draft
- feature.review
- feature.publish
- feature.archive
- feature.delete
- feature.other
- experiment.create
- experiment.draft
- experiment.start
- experiment.stop
- experiment.targeting
- experiment.traffic
- experiment.phases
- experiment.visualEditor
- experiment.other
- savedGroup.create
- savedGroup.draft
- savedGroup.review
- savedGroup.publish
- savedGroup.archive
- savedGroup.delete
- savedGroup.other
- constant.create
- constant.draft
- constant.review
- constant.publish
- constant.archive
- constant.delete
- constant.other
- config.create
- config.draft
- config.review
- config.publish
- config.archive
- config.delete
- config.other
- rampSchedule.create
- rampSchedule.start
- rampSchedule.pause
- rampSchedule.progression
- rampSchedule.approve
- rampSchedule.edit
- rampSchedule.delete
- rampSchedule.other
- settings.approvals
- settings.other
- access.members
- access.teams
- customHook.create
- customHook.edit
- customHook.delete
- environment.create
- environment.edit
- environment.delete
- project.create
- project.edit
- project.delete
- attribute.create
- attribute.edit
- attribute.delete
- customField.create
- customField.edit
- customField.delete
- sdkConnection.create
- sdkConnection.edit
- sdkConnection.delete
- metric.create
- metric.edit
- metric.delete
- factTable.create
- factTable.edit
- factTable.delete
- segment.create
- segment.edit
- segment.delete
- dimension.create
- dimension.edit
- dimension.delete
- override.ignoreWarnings
- override.skipSchemaValidation
- override.skipHooks
- override.bypassApproval
- override.restBypassesReviews
- override.draftLimit
- override.skipChecklist
- override.bypassDuplicateKeyCheck
- override.unscopedRules
- other.write
environments:
type: array
items:
type: string
required:
- action
- environments
additionalProperties: false
summary:
type: string
url:
type: string
expiresAt:
type: string
response:
anyOf:
- type: object
properties:
status:
type: number
body: {}
required:
- status
- body
additionalProperties: false
- type: 'null'
note:
anyOf:
- type: string
- type: 'null'
required:
- id
- status
- actions
- summary
- url
- expiresAt
- response
- note
additionalProperties: false
responses:
ConfirmationHeld:
description: >-
Held until a person confirms it in GrowthBook. Poll the Location until
the confirmation completes.
headers:
Location:
description: The confirmation to poll
schema:
type: string
Retry-After:
description: Seconds to wait before polling
schema:
type: integer
content:
application/json:
schema:
$ref: '#/components/schemas/ConfirmationHeld'
securitySchemes:
bearerAuth:
type: http
scheme: bearer
description: >
If using Bearer auth, pass the Secret Key as the token:
```bash
curl https://api.growthbook.io/api/v1/features -H "Authorization:
Bearer secret_abc123DEF456"
```
basicAuth:
type: http
scheme: basic
description: >
If using HTTP Basic auth, pass the Secret Key as the username and leave
the password blank:
```bash
curl https://api.growthbook.io/api/v1/features -u secret_abc123DEF456:
# The ":" at the end stops curl from asking for a password
```
````