Build an Event Property that uses the label method and breaks event down and aggregates them into Free (direct, organic_search, other) and Paid referrals (advertisement, paid_search, social). Then use that Event Property in an Explore query to return a pie view that shows the count of users who generate those event types over all time.
Build an Event Property that uses the label method. Value 1 can be named Free Referrals and it should be filtered to events with Referrer that matches direct, organic_search, and other_web. Value 2 can be named Paid Referrals and it should be filtered to events with Referrer that matches Advertisement, paid_search, and social then save it. In the Explore view, build a query that uses Pie view that returns a count of unique user actors filtered to all user actors that is split by the Event Property you created and saved in the first step ran from the beginning of time to now.
Event Properties are critical in organizing and getting specific with your data set. You can add new roll-up columns, create static filter sets, rename columns or values and have them as persistent saved objects, and more. They’re similar to Actor Properties, however, Event Properties are solely concerned with events and event meta-data, whereas Actor Properties can be thought of as Segments of Users who do similar behaviors or are in similar demographics.
Note: The numbers and result does not need to be an exact match as the data changes over time. The query construction is what is most important. See images below.
*Click to enlarge*