Net Promoter Score

Definition

Across a Bain study of more than four thousand customers in six industries, one question predicted repeat purchase and referral behavior better than any other candidate on a twenty-question loyalty survey: "How likely is it that you would recommend Company X to a friend or colleague?" Scored zero to ten, customers cluster into three behaviorally distinct groups — promoters (9–10), passives (7–8), and detractors (0–6) — and the Net Promoter Score is simply the percentage of promoters minus the percentage of detractors.

In the Book

Chapter 2 tells the story through Intuit cofounder Scott Cook, who could sense customer sentiment personally in Intuit's early days but lost that visibility as the company scaled — while "vague statistics of satisfaction" gave managers nothing reliable to act on, so they defaulted to the numbers they could measure: cost per call, revenue, staffing levels. Bain's research process ran fourteen case studies matching customers' actual survey answers against their later purchase and referral behavior; the recommend question won because it demands both a rational judgment (does this offer real value?) and an emotional one (does the company treat me, and would it treat my friend, well?) — "engaging the customer's head" and "engaging the heart" at once. The book explicitly rejects retention rate (measures the bucket emptying, not filling) and satisfaction surveys (GM won J.D. Power awards in 2005 while its market share and bond ratings collapsed) as reliable growth predictors.

Why It Matters

It converts an amorphous concept — "loyalty," "satisfaction," "engagement" — into a single number simple enough for frontline employees to act on daily and rigorous enough for a CFO to invest against. The three-way segmentation matters as much as the score itself: "increase promoters, reduce detractors" is a directive an employee can act on this week, in a way "raise the satisfaction index by one standard deviation" never was. The underlying move — finding the one behavioral question that predicts an outcome better than a battery of proxy questions — is portable to any domain drowning in surveys that generate reports instead of action.