Context Kind

What it means

A context kind describes the type of subject a flag evaluation is about. The classic subject is a user, but many rollouts need to target an organization, an account, a device, a request, or even an environment. Each of these is a distinct context kind, carrying its own attributes that targeting rules can match against, such as a plan tier on an organization or an OS version on a device.

Why it matters

Real products rarely roll out purely by individual user. You often want a feature on for an entire company at once, or gated by device capability, or scoped to a region. Modeling those as separate context kinds lets targeting rules reason about the right entity instead of forcing everything through a single user object. It also keeps billing and analytics honest, since the tool can count and segment by the kind that actually matters to your business.

In practice

Imagine shipping a beta only to organizations on an enterprise plan. With an organization context kind you write one rule that matches the org attribute, and every user inside those orgs sees the feature automatically. LaunchDarkly popularized multi-context evaluation, letting a single flag consider a user and an organization together. Our feature flag management guide and how to implement feature flags show how to choose the right context before you write targeting logic.