Try to analyze CR by question type. For example, look at a bunch of Assumption questions together. What patterns do you notice in the way that they ask questions, or the kinds of assumptions tested? Here are a few common assumption types, but be on the lookout for others:
- correlation v. causation
- number v. proportion
- substitution of one statistic for another
- no other possibility
- general v. specific data

Then look for patterns in the ways the GMAT constructs wrong answer choices for each question type. Here are a few common wrong answer types for assumptions:
- too extreme
- related to the topic, but not related to the conclusion
- similar to a premise, but not filling a gap

Do this for every question type on CR, and you'll start to notice patterns. This will allow you to say "I know exactly what I'm looking for here."