1. Make sure you have alignment with your sales organization as to what, and how, you score behavior. How much value do you put on a specific form fill, a whitepaper download, an event registration? Knowing these types of values can help you get a better idea of where to create your threshold. You may also wish to have separate scoring models for leads, and for users who already have a rep assigned.
ITEM OF NOTE: You should also consider degradation scoring - negative scores for things like extended periods of no activity, unsubscribes from email, etc.
2. For us, we use score to show how many leads reached the score threshold in a given period of time. This helps us get a bit of a look at our lead's journey looks from entry to qualification. Example: Was created x time frame, score changed to over 100 in the last week.
3. Score your web visits off of what you can, and also do your scores for forms, emails, events/webinars, etc. You're not going to be able to do a full historic scoring of everyone, but you are going to be able to get a good look at those who have been active most recently, and have a good starting ground for all of your lead scoring going forward.
So you don't really need the behavior score values for the analysis, if you know which activities are being scored you can just look directly at those (i.e. campaign members and their statuses).
Other great comments here too!