The score is the least useful output
Most small businesses that run a customer survey end up with a number, a mild sense of how it compares with last quarter, and very little else. The number moves slowly, is heavily influenced by who happened to respond, and in a small base can swing on a handful of answers. Treating it as the output is what makes NPS programmes feel like an administrative ritual.
The output that matters is the set of conversations the survey creates. Those conversations produce specific, actionable information, and they also produce the recovery of individual relationships. If you have to choose between running the survey well and following up well, follow up well on a smaller sample.