Delve works best when the players immerse themselves in a sense of place, effectively seeing through their characters' eyes. When faced with fictional situations, you have no stats or metagame knowledge to impact your decisions. It's just, "I'm here in Lendrhald, here's what I see, now what do I want to do?"
For an individual player, creating immersion is largely a matter of priority, asking the GM environmental questions until a clear enough image comes together to inform your choices. For a group of players, another task becomes necessary: clear communication.
Starting and Stopping – "Dive In", "Up For Air"
When checking to see if everyone's ready to immerse, I ask the group, "Ready to Dive In?" When checking to see if everyone's ready to get their heads out of the game world and into the real world, I ask, "Up For Air?" If you already have phrases you prefer, use those. It doesn't matter what your system looks like, only that you have one.
It is important that everyone knows whether a player is speaking in-character or out-of-character at all times. This becomes an issue in a few situations:
1. While everyone in the group is trying to immerse, there are many ways to signal "this comment is out of character", such as physical gestures (e.g. hand on head) or bookending with certain phrases (e.g. "up for air" . . . "back in-game").
2. If you want to speak in-character before the group has communicated that you're immersing, DON'T. A confused state of fuzzy fiction results. Instead, ask the group, "Ready to Dive In?" first, let the group frame the scene, and then deliver your line.