A toggle is a switch that has two outcomes: it’s either on or off. It’s used in almost every aspect of computing whenever there’s an options or preferences list—think of the Caps Lock and Num Lock keys, or the controls for painting features like brush size, paint drop-off, and opacity. Generally, toggles are set to be on and left in that state, but they can be turned off by clicking or tapping them.

The term was first recorded in the 18th century as a noun meaning “a pin passed transversely through an eye or loop in a rope or chain, to hold it fast.” Nowadays, it refers to a kind of toggle button that lets people choose between opposing values that affect the state of content or a view. It’s usually configured with a distinct appearance to indicate its current state, and platforms may provide guidance for implementing toggle styles.

People perceive a toggle to be either on or off, so it’s important that its design makes clear which of these states it currently represents. This is especially true for visual cues like icons, which should communicate the toggle’s current state to users. It’s also vital that the toggle’s surrounding context provides sufficient information to help people manage its state.

Toggle configuration should ideally be baked into the release and not change over time. That’s why savvy teams often add a task to their backlog when they introduce a new toggle, and then use the team’s process of deploying changes to roll out a new release with that toggle either on or off as necessary. This can have the effect of reducing the need to test new releases and helps keep their total number low.

However, there are some cases when a more dynamic approach to toggle configuration is needed. One common way to achieve this is by modifying code with comments. The disadvantage of this approach is that it requires a restart or re-deployment of the artifact to toggle it, and so can have a negative impact on your validation cycle times and overall testing speed.

Another method is to use an in-memory toggle manager, which allows you to flip a toggle on-demand and without restarting the artifact. This can be a good solution when you’re creating experiments, for example, and don’t need to deploy a new version to validate the experiment. However, it’s also not a good option for a production system, as it can disrupt the flow of the CI/CD pipeline. If you do need to run a new release with the toggle either on or off, be sure to include a test that checks its current state before continuing on with the rest of your tests. This will ensure that your QA team isn’t accidentally overrunning your toggle-flipping tests with unneeded changes. For this reason, some teams even add expiration dates to their toggles in order to ensure that they’re removed as soon as they’re no longer needed.