Advanced Email Subscriber Churn Rate - Taking into Account Inactivity and Cohorts

Hi.  While I've mastered most of the basics in DAX and M Query for churn, there's a scenario that I find quite interesting that I wanted to see if people had feedback for.

Consider a churn analysis for a typical large (30m monthly unique visitor) publisher's email subscriber list - snapshot below:

Here's what these columns represent and why they should be all that's needed to determine churn:

1. Profile ID represents a unique subscriber (in this case it's just a hashed email address).  For DAX churn calcs, we'll basically be counting these under various contexts.

2. Profile Created Date in this instance determines cohort since this is the date someone joins.   What I'm going to suggest below is a churn metric that uses a median of the churn rate across certain cohorts.

3. Engagement status.   The most basic level of churn happens when someone goes (i) "hardbounce" or (ii) "optout".  I call this "Hard" Churn.   The Latest Interaction date captures the date of this event.

4.  (and here's the hardest part) Latest Interaction.   Most subscribers in these email lists never hardbounce or optout and yet need to be counted as what I call "Soft" Churn because they disengage to the point where don't visit and need to be counted as churn.    

Accordingly, does anyone have any suggestions on how I'd write a DAX query to get the following: 

A.  Median 28 Day "Hard" Churn for the Prior 12 Months:  taking into memory for each of the last 12 fully completed monthly cohorts (ie since it's now Jan 2019, we would look at the cohorts of users who signed up during 2018-01, 2018-02 .... 2018-12), calculating the percent of Profile ID's that "hard churn" -- defined as "hardbounce" and "optout" -- in each cohort, and from that creating a measure of the median of  those 12 churn percentages.

(note I've done this -- harder part follows below)

B.  Setting up the 28 day period as a variable or parameter so that we can look at churn over any number of days while automatically counting only the last 12 periods that are fully completed given that number of days.  In other words, let's say we want to change 28 Day Hard Churn to 90 Day Hard Churn, the 12 cohorts analyzed need to be 90 days earlier (2017-09, 2017-10 .... 2018-09) because a cohort will only be measurable for churn purposes if its last day is older than the # of days history being looked at.  

(and here's the hardest part)

C. Adding "Soft" Churn" to #1 and #2, above.     "Soft" Churn is someone who should be counted as having churned because they havent visited the site in over X days (let's use 120 days for now - although ideally this is Parameter).  This is the part that has made my brain hurt the most because the DAX needs (I think) to draw a variable based the difference between a fixed date that the numbers were drawn (ie 1/20/2019) and Latest Interaction to see if the Profile ID should be deemed churned due to inactivity.  I have done this manually through tremendous amounts of calculated columns.  Do any DAX geniuses know how to do this with measures from just the 4 primary columns, above?

Note: since anyone interested in this is probably interested in what we see as churn, where's what the numbers look like for us (based on my more manual and more M-Query calcs):   see chart below: we typically see hard churn ~16% in the first month of any cohort, then an additional 6% in month 2, then about another 20% over the rest of the year for ~40% in the first year if we count 6 month inactives as having churned.   6 month inactives come back only about 5% of the time.

\

If I were super-skilled DAX, ideally I'd be able to show a stacked area chart for the median of any given set of cohorts where the Y axis is % churn, the X axis is time, and the stacked areas are "hard churn" and "soft churn" (judged by X day inactivity)