Tools › Finance and Business › Marketing › Customer Metrics (NPS) › Example

Net Promoter Score of an illustrative 400-response file: Rocks' Adjusted Wald interval, segment and period tests, key drivers

Tool	Customer Metrics (NPS)
Source	Rocks, B. (2016). "Interval Estimation for the 'Net Promoter Score'." The American Statistician, Vol. 70, No. 4, pp. 365–372. Sections 1.4 and 2.1.3; Reichheld, F. F. (2003). "The One Number You Need to Grow." Harvard Business Review, Vol. 81, No. 12, pp. 46–54; Sauro, J., and Lewis, J. R. (2016). Quantifying the User Experience, 2nd ed. Morgan Kaufmann, Cambridge, MA (the adjusted-Wald NPS interval).
Level	Professional

Problem

A relationship survey of 400 customers in three segments (Consumer, Enterprise, SMB) over four quarters records the 0–10 likelihood to recommend and three driver ratings (support, price, quality, 1–10). Find the NPS with its standard error and 95 % interval, test it against an industry benchmark of +30 and a prior score of +20 from 350 responses, compare the segments and the quarters, find the sample size for a ± 5-point interval and rank the drivers.

Inputs to Enter

Metric NPS, layout respondent rows; the default 400-line file (illustrative, seeded; References/Customer-Metrics/default_data_illustrative.csv)
Confidence 95 %; benchmark 30; prior score 20 with n 350; target half-width 5; segments Enterprise against Consumer; periods first against last

Open the Customer Metrics (NPS) tool (loaded with this example)

Answers

191 promoters, 124 passives, 85 detractors: NPS = 100 (0.4775 − 0.2125) = +26.5; SE = 100 √((0.4775 + 0.2125 − 0.265²)/400) = 3.936 (Rocks eq. 1.4); Wald interval +18.8 to +34.2
Adjusted Wald AW(3, T): n 403, adjusted shares 0.4758 / 0.2128, centre +26.30, SE 3.920, interval +18.6 to +34.0 (Rocks' recommended closed form; Sauro and Lewis's seventeen printed widths reproduced in the test)
Benchmark +30: z = −0.89, p = 0.374 (not distinguishable); prior +20 of 350: SE bound 5.24, z = 0.99, p = 0.321
Enterprise +66.9 against Consumer −8.2: difference 75.2, SE 9.08, z = 8.27; X² = 71.05 on 4 df (scipy), Cramér's V 0.298; change point after Q2 (+5.9 to +52.0, z = 6.19); WLS slope +15.0 points per quarter (SE 3.26)
Sample size for ± 5 points: (100 × 1.96)² × 0.6198/25 = 952.3 → 953; key drivers quality r 0.653 and support r 0.620 (Pratt shares 54.8 % and 43.7 % of R² = 0.529), price r 0.260 low priority

Notes

The tool's default inputs are this file; press Submit. The data are illustrative (not survey data); every number is checked in test_customer_metrics.php (282 checks) against numpy / scipy (References/Customer-Metrics/cm_check.py) and the published values of Rocks, Sauro and Lewis, Newcombe, Agresti and NIST StRD.