My organization does! Here's some things we have implemented I'd recommend you keep in mind:

1. Ensure questions are phrased the exact same way to ensure you're measuring the same thing. So if the relationship survey asks for liklihood to recommend the company/product.... the post-interaction survey should say the same thing, not "how likely are you to recommend this call center/team." An NPS for the org and an NPS for a service may be two different ratings in the mind of a customer.


2. Bear in mind you may want to weight responses. A straight aggregation, where every survey and every customer response counts for equal weight might skew towards the survey where a lot of responses are coming through (e.g. if more people take the relationship survey than the interaction survey, the results will skew on the relationship side). My org has multiple relationship survyes, and multiple interaction surveys. Within those groups- the weight is by n-count. But to achieve an org-wide-single-NPS aggregate, we weight relationship surveys 50% and interaction 50%.