Lost customers

[Khalidc]

Hi 

I try to calculate the lost customers over time in a specific week. I feel like I am almost there but cannot get it to work. I am for sure overseeing something in the measure, hopefully you can help.

The assumption used it that all customers who have a last sale older than 12 weeks should be considered lost. The measure should be able to present the lost customer in a stack column chart but also show which customers are lost in a matrix table when selecting the period in the chart.

# lost customers week = 
var SelectedWeek = SELECTEDVALUE('FS_Vertaling weken naar perioden'[Index - week]) 

var allcustomers = values(FS_Omzet_uren_all[Groepnaam]) --take all customers
var customerinactiveperiod = CALCULATETABLE( values(FS_Omzet_uren_all[Groepnaam]), --select all customers who match below filter
filter( all( 'FS_Vertaling weken naar perioden'), --remove all date filters
'FS_Vertaling weken naar perioden'[Index - week] > SelectedWeek-12 &&
'FS_Vertaling weken naar perioden'[Index - week] < SelectedWeek)) --take all customers with sales in the last 12 weeks

return

COUNTROWS(
EXCEPT(allcustomers,customerinactiveperiod)) --take all customers and remove all active customers (customers in active period) 

This is what I get:

This is what I would expect:

[amitchandak]

@Khalidc , refer my apporch here

Customer Retention Part 1:
https://community.powerbi.com/t5/Community-Blog/Customer-Retention-Part-1-Month-on-Month-Retention/b...
Customer Retention Part 2: Period over Period Retention :https://community.powerbi.com/t5/Community-Blog/Customer-Retention-Part-2-Period-over-Period-Retenti...

dealing with week refer

Power BI — Week on Week and WTD
https://medium.com/@amitchandak.1978/power-bi-wtd-questions-time-intelligence-4-5-98c30fab69d3
https://community.powerbi.com/t5/Community-Blog/Week-Is-Not-So-Weak-WTD-Last-WTD-and-This-Week-vs-La...
https://www.youtube.com/watch?v=pnAesWxYgJ8

[Khalidc]

Unfortunately Amit, I tried it but unfortunately doesn't seem te work.

I am working with non-date periods. I have indexed the weeks, which start at week 1 in 2019 (index 1) and counting until today.

The rolling 3 and rolling 9 before 3 seem to give the correct result. However the Lost customer in period remain blank everywhere.

I copied the measures used below. Groepnaam is the customer name which is in the sales transaction table (hence customer/groepnaam is not unique in the table).

Hopefully you see what is missing.

#Rolling 3 = 

var MaxIndex = SELECTEDVALUE('FS_Vertaling weken naar perioden'[Index - week])
var MinIndex = MaxIndex - 12

return

calculate('03 FS measures'[Omzet totaal],
Filter (all('FS_Vertaling weken naar perioden'),
'FS_Vertaling weken naar perioden'[Index - week] <= MaxIndex &&
'FS_Vertaling weken naar perioden'[Index - week] > MinIndex))

#Rolling 9 before three = 
var MaxIndex = SELECTEDVALUE('FS_Vertaling weken naar perioden'[Index - week])-12
var MinIndex = MaxIndex - 52

return

calculate('03 FS measures'[Omzet totaal],
Filter (all('FS_Vertaling weken naar perioden'),
'FS_Vertaling weken naar perioden'[Index - week] <= MaxIndex &&
'FS_Vertaling weken naar perioden'[Index - week] > MinIndex))

#Lost Customer This Period = 
Sumx(VALUES(FS_Omzet_uren_all[Groepnaam]),  
if(ISBLANK([#Rolling 3]) && not(ISBLANK([#Rolling 9 before three])) , 1,BLANK()))