@dxconnamnguyen I’ve seen it done both ways; and not seen documentation on the Qualtrics end in regards to dashboards.  I know many who prefer to consider the last measure as the primary one - however in my experience and through BAIN methodology who created NPS the main measure or overall score would be the combined average across all touchpoints, not the final one. This is the method I follow myself

You might like to refer to documentation by BAIN on this,  perhaps check out figure 2 on page 5 of the following.  BAIN_BRIEF_Mastering_the_episodes_that_count_with_customers.pdf

The other argument for not making it the last touchpoint; is that is skewed more positively typically as it only includes people who successfully made it through the full journey.  

I definitely agree with @ScottG, the last NPS question that is based on a touchpoint won’t be capturing the overall as it is a question pointed to the specific touchpoint.

The Overall could either be created as a separate question and if the survey is already live, the weighted average of all NPS scores might be a better measure of the overall NPS.

I would aggregate tNPS and rNPS separately, since one measure is representing a respondent’s experience with a rband or service as a whole and the other is based on a specific touchpoint/interaction. So you could show an overall tNPS metric and then an overall rNPS metric separately.

Hi guys @ScottG @Rishabh_Singh05 @ash123 
Thank you for your responses!
I’ve been asking info recently with a Qualtrics XM Scientist and this is his response

“1/ On a customer journey, can the transactional NPSs of all the touchpoints be summed up to the Relational NPS? (As far as I know, it is not.)
I could see why this question would come up. People generally think that the sum of the experiences "experienced" across transactions will sum up to a relational measure. If the company is very transactional focused then the interactions on those transactions will be seen as the single most important aspect of a customer's experience. 

I would agree with you that this is not a replacement for a relational measure. A customer's relationship with an organization is made up of many more aspects than just the experience of the transaction. Some of these could include:

• Brand reputation
• Product quality
• Customer communications (outside of transactions)
• Personalization
• Transparency
• Community approach
• Approach to Equality, Diversity and Inclusion.
• Environmental stance

2/ Can we use a ratio/weight for transactional NPS of more or less important touchpoints to have a more accurate relational NPS result? Is this acceptable?

I generally advise against blending metrics but if all of the measurements are NPS then I don't see any issue with the weighting of certain touchpoints against others. This is a form of prioritization and this is a good practice. Organizations should not try and fix all of their issues and there will be some that are more critical to the experience than others. What this does do is complicate communication of the NPS results but I think conversations around which touchpoints are a priority for the customer and for the organization is ultimately a good thing. “

So rNPS can be shown as a big overall number and tNPS should be placed on journey map to identify gap. I think this is good enough to convince my client on what to do.
Many thanks,

@dxconnamnguyen Glad you have resolved your question on this.  I would agree with XM scientist that I would not combine an rNPS and tNPS measure - as there is typically different objectives with each of these NPS measurements.

From my example however what I was thinking of was more an episode NPS measurement; where we do aggregate the score from multiple sub-episodes.  I.e. in banking we would have a customer episode of ‘Buying a home’.  For this we measure the episode at multiple touch points; booking an appointment, making an application, approval of loan and finalisation of loan.  We would have an episode NPS measure for each moment of the customer journey; but our overall NPS for the episode will be an average of each of these sub-episode moments.  Ourselves, we would not weight each of these episodes differently, unless there was a substantial difference in volume of responses at each sub-episode and we then may choose to weight based on response volumes to even these out.

Just thought I’d give extra context to my earlier comments and how we do it.  Appreciate you sharing the response you got from XM scientist as well.