*I know I can then make further branching logic to and create additional embedded data fields (ie if they are tagged with fair + poor = poor, or good + poor = poor, or good + fair = fair). I am trying to avoid making dozens of branches and multiple embedded data point. 

If it should be based off their least satisfied response then simply reverse the branch logic you have in the post question. That means it should first check for sales health good then fair and lastly poor.

@Shashi that didn’t work for me - I reversed the order as seen here (was sure that I hit “apply” prior to creating my test distribution)
 

I then took the survey, making sure that I hit all 3 of the sentiments as again I only want the “sales health” embedded data to = the lowest score that was selected, in this case is should have been been as I selected both dissatisfied and very dissatisfied as shown here 

But the when I go into the response their embedded data is “good.” I am missing something here...

@Shashi that didn’t work for me - I reversed the order as seen here (was sure that I hit “apply” prior to creating my test distribution)
 

 then took the survey, making sure that I hit all 3 of the sentiments as again I only want the “sales health” embedded data to = the lowest score that was selected, in this case is should have been been as I selected both dissatisfied and very dissatisfied as shown here 

But the when I go into the response their embedded data is “good.” I am missing something here...

If you took the survey via distribution, just wanted to check if you published the survey?

Hi @kgillis,

I recommend creating additional branching logic and generating extra embedded data fields. For instance, if responses are tagged with fair + poor, categorize them as poor. Similarly, if responses are tagged with good + poor or good + fair, categorize them accordingly. While this may result in numerous branches and multiple embedded data points, it is the current suggested approach to address this issue.

Thanks @Sachin Nandikol that was my original thought, I was hoping there was a “sleek” way to condense it so if someone else was to come in and look it through they would be able to quickly make logical sense of it. I think this is what I’m going to end up doing since I can only get 2 out of the 3 to work correct - super strange, I’ve even deleted and rebuilt the one that isn’t placing the embedded data correctly.