My recommendation: don’t over think it. For impact and confidence it’s better to have a short discussion and choose the number by group decision (you’ll update it later when you have more data). For confidence there should be less guessing — it should come from available evidence. ICE is all about making better and better estimations as we’re subjecting the idea to more rigorous testing. See example in this article: https://hackernoon.com/finding-winning-ideas-using-the-confidence-tool-d8f2d8cc2c15.