Net Promoter Score NPS
A loyalty score built from one question -- how likely are you to recommend us -- on a 0-to-10 scale.
The fork, why two teams get different numbers
The STUDY TYPE forks first: RELATIONAL NPS surveys the whole customer base
periodically about the overall relationship, while TRANSACTIONAL NPS fires
after a specific interaction (a purchase, a support ticket) and measures that
touchpoint -- the two produce different scores for the same company and answer
different questions. The CALCULATION is standardized (%Promoters[9-10] minus
%Detractors[0-6], with Passives[7-8] ignored, range -100 to +100) but the
SCALE gets abused: teams run a 1-5 or 1-10 survey and map buckets their own
way, or report the raw 0-10 MEAN and call it "NPS." And the SAMPLING frame --
who was surveyed, response rate, in-product vs email -- swings the result more
than any real loyalty change.
The trap
NPS is NOT a percentage and not on a 0-100 scale -- it runs -100 to +100, so "our NPS is 45" cannot be read as "45% happy," yet it constantly is.
- The full trap, worked on a real export
- Every formula variant, spelled out
- The reconciliation anchor, what to tie it to and when to refuse
Reference: Net Promoter Score survey framework (Reichheld / Bain origin) cited as reference · not reproduced; survey-methodology conventions on sampling and response bias