A Sift Score is a number between 0 and 100 that indicates the riskiness of an action (what Sift refers to as an "event") taken by your users. The lower the score, the less likely the event is high risk.
To find your ideal block/review/accept score thresholds, look for a score where you're seeing the majority of the entities above it are fraud. Since each business has different considerations for the right block percentage, false positives to allow (good entities blocked), and fraud to catch based on automation, you'll need to do some evaluation - we're happy to help so feel free to reach out.
A high score in one abuse type does not tell you that a user will go on to commit fraud in another. So, a user could have a low score for Content Integrity and a high score for Payment Protect, suggesting you're at risk of payment fraud but not content fraud. However, you'll often only see one score at a time as you would check for Content Integrity score content is posted and the Payment Protect score at the point of transaction. As such, your score thresholds will likely vary between the different fraud and abuse types you're using Sift to prevent.