If you “know” that one is going to have higher impact than the other you don’t need this tool. If you’re not full sure about impact (can you ever be?) then the ICE score gives you a hint — knowing what you know now which is the better bet. Having said that you don’t have to move forward strictly by ICE score order. ICE can penalize big ideas that have low confidence, so it’s ok to “boost” specific ideas and slot them for further testing. As long as you don’t commit to building the full thing, just to validate it with the next minimal experiment, you should be good.