"pointing to a higher net promoter score"
I think I've found the problem.
Net Promoter Score is a proprietary technique that can work well, but is unfortunately easily manipulated, simply by choosing who is asked to rate the service, and at what point they are asked. Anybody familiar with a helpdesk run under SLA will notice that requests for feedback seem to crop up for all the easy, first time fixes, but never when you're querying an unresolved, complex ticket that's months overdue.
NPS can work very well, but it is a bit like immunisations - you need to have a very high coverage for it to be effective at a population level. In practice, companies latch onto to NPS, pay the licence fee, and then incentivise managers to improve the figure. The NPS score improves, and the companies kid themselves that they are improving, even though nobody outside the company believes this. I speak from experience, working for one of a number of energy suppliers who use NPS, and our scores have (apparently) improved considerably and consistently over recent years. Until you look at impartial and external data, like customer complaints, time to complaint resolution, ombsudman referrals, and customer losses, all of which show no improvement at all. Looks the same over at Toadafone.