Help please asap - sankey or alternative chart for product change

[Anonymous]

Hi,

My goal is to be able to track customers that changes products historically.

So for a customer number (which can have multiple services under it), there can be x number of Product A's as of March 2021.  For example, if customer has 7 product A's as of March 2021, I want to know if that 7 product A's are still 7 as of April 2021. It is possible that the 7 products A's will be 2 product B's, 3 product C's, and  retain 2 product A's by April 2021.  Or there is a possbility that the 7 Product A's will only be 5 for next month for that customer (signifying a churn). 

I am thinking of representing this using a sankey chart or a tassels or a chord chart. However, I dont think I am getting the right numbers:

Based on the left table, there are two services for that customer but the sankey chart tells me equal proportion for IPMAN, Business IP, and TID. I know this is wrong. I want to track that IPMAN is still IPMAN on the right side of the chart. So, the right table must only total to 2 when I select the IPMAN in the left table. 

[AllisonKennedy]

@Anonymous  Coincidentally I just did a video/blog on Sankey visual, not sure it will entirely help you but we need to know more what your data model/relationships look like to troubleshoot the filering issue you're having. Also, have you put anything in the weight for the Sankey? 

https://excelwithallison.blogspot.com/2021/07/custom-visual-review-sankey.html 

[Anonymous]

Hi @AllisonKennedy , yes I did put weight on it. The problem is it is proportional. What I want is to see where the 7 services under that customer move (if they moved) to the next month.

My data is historical for the last three years. So you will see the behaviour of the customer from Jan 2019 up to now. IN other words, a customer number is not unique for the data. It will appear ~36 times because my data is ~3 years. Now what I did in my sankey chart was to get that customer as of March 2021. Based on the figure, ithas 2 IPMANs. What I want to know is next month, do I still have that 2 IPMANs. Or it became Business IP or Connect IP for example. So my source is product as of March and then the destination is still product but as of April. 

What I did in the data model was to create two tables from the source:

df3 is data as of march

df4 is data as of april

then I connected them by CIDN

I want to know what happened to that CIDN and Product from df3 to df4

I hope this makes sense. 

[AllisonKennedy]

@Anonymous  Do you have a service ID or some other value that can link the change from month to month? Or a change history table? 

With your current data as explained, there's no way to know which products transferred to which, so the best I can get is to connect them to 'null' value. See attached file below signature.