Metrics are arbitrary datasets that can be displayed alongside any component, or on the homepage of your status page. Currently, all metrics are tied to specific components. There are various constructs you should be aware of when working with Hund metrics.
Metric providers (also called metric sources) provide various metrics to a component via the component's watchdog. A component may have several metric providers, some of which may only report certain metrics (e.g. the Updown metric provider only provides "apdex", and the Pingdom/Uptime Robot providers both provide only "res"). However, there exist general metric providers, such as the webhook metric provider, which can provide any arbitrary metrics you wish.
Metric definitions define the basic properties of a metric, such as the title, axes types, interpolation mode, aggregation function, etc. Every metric definition is uniquely characterized by its "slug" (e.g. "res", "apdex", "percent_uptime", etc.), which is used by the rest of the application to reference that metric.
In the case of general metric providers, such as webhooks, metric definitions can be created and subsequently instantiated on-the-fly by selecting the Define new metric... option of the Definition Slug field when adding a metric to a provider. Naturally, these definitions can be instantiated on any other general metric providers on any other components.
Metric instances are instantiations of particular metric definitions. They are added to metric providers to declare what is provided, as well as any extensions to the base definition. Instances inherit all options from their definition but can override whatever options must differ from the definition (e.g. overriding the title of a metric to be specific to the component).