Throughout your experimentation, you will probably publish, start, and pause your experiments many times. This article explains the differences among these three actions and how they affect your experiment's status.
Publish saves and “stages” your experiment’s changes. If you have not started the experiment or the experiment is paused, your changes will not be visible to visitors when you publish. A common use case for publishing changes to experiments that are not running is using Optimizely's Experiment Scheduler to start your experiments.
If you publish changes to an experiment that is running (in other words, an experiment you have started but have not paused), visitors will see your changes immediately. We do not recommend publishing changes to a running experiment.
Start makes your experiment live to all visitors in your targeted audience and changes the experiment status to "Running."
Pause stops your experiment so new visitors will not see it and changes the experiment status to "Paused." No new visitors will see a paused experiment until you restart it.