How do you approach multi-touch attribution, and what are the pros and cons of a first-touch versus a last-touch attribution model?
