persistency check

Anonymous , on the way, is I create a month year column and do count distinct month year for rolling three month data. So if If you get 3 it means they got for all three months.

Canceled in next month, check for next month based on cancellation date , You need to  join purchase and cancel both to same date table and the use userelationship and time intelligence 

example

Rolling 3 = CALCULATE(distinctcount(Table[Month Year]),DATESINPERIOD('Date'[Date],max('Date'[Date]),-3,MONTH))

Rolling 3 = CALCULATE(distinctcount(Table[Month Year]),DATESINPERIOD('Date'[Date],eomonth(max('Date'[Date]),0),-3,MONTH))

Purcahse in last three months /Customer id or policy id(if same policy)

Example of this month next month

MTD Sales = CALCULATE(SUM(Sales[Sales Amount]),DATESMTD('Date'[Date]))
next MTD Sales = CALCULATE(SUM(Sales[Sales Amount]),DATESMTD(dateadd('Date'[Date],1,MONTH)))

next month Sales = CALCULATE(SUM(Sales[Sales Amount]),nextmonth('Date'[Date]))

Userelationship : https://www.youtube.com/watch?v=e6Y-l_JtCq4

Power BI — Month on Month with or Without Time Intelligence
https://medium.com/@amitchandak.1978/power-bi-mtd-questions-time-intelligence-3-5-64b0b4a4090e
https://www.youtube.com/watch?v=6LUBbvcxtKA

countx( values(Table[Policy ID]), if( not(isbalnk([Rolling 3])),1,blank()))