Net Promoter Score: calculate, weight, track

In DataLion the NPS is a transparent codebook formula — exact, weightable and reproducible. Segment it and test for significance, smooth the trend, and use driver analysis to see what moves it. ISO 27001-certified hosting, GDPR-compliant, made in Munich.

Try DataLion free
Interactive DataLion dashboard with an NPS score, promoter/detractor split and filters

DataLion computes the Net Promoter Score directly in the codebook as 100 × (share of promoters 9–10 − share of detractors 0–6) over valid answers — optionally weighted. You break the score down by segment, test wave and group differences at confidence levels from 80% to 99%, smooth the trend with SMA/EMA, and use relative-importance analysis on the R engine to see what drives it.

  • 🇩🇪 Made in Munich
  • GDPR-compliant
  • DPA included
  • Hosted in Germany

Trusted by research, insights & media teams

  • GfK
  • L’Oréal
  • Deutsche Telekom
  • eBay
  • Hubert Burda Media
  • SevenOne Media
  • mobile.de
  • Psyma
  • 0–10 recommendation scale
  • −100…+100 NPS range
  • 80–99% significance levels
  • SMA / EMA tracking smoothing

Why the NPS is often worth less than it could be

  • The score is computed by hand in Excel — sometimes weighted, sometimes not, with no documented formula.
  • Segment differences get interpreted without testing whether they are statistically significant.
  • The why-answers get collected but are never distilled into drivers.

The NPS — computed exactly, in the codebook

DataLion does not treat the NPS as a black box but as a transparent codebook formula: 100 × (share of promoters [9–10] − share of detractors [0–6]) ÷ valid answers. Anyone answering 7 or 8 counts as a passive.

The formula excludes missing values, so non-responders do not distort the score. If a weight variable is present, the weighted NPS is computed (each case × its weight) — one or several weights are supported.

  • Promoters (9–10), passives (7–8), detractors (0–6) as top-box/net rows
  • A reproducible codebook formula, not an opaque metric
  • Missing values cleanly excluded
  • Weighted NPS over one or several weight variables
DataLion table with NPS net rows and percentages

Break it down by segment — with significance

Break the NPS down with live filters by region, touchpoint, product or customer group. Differences between segments are significance-tested right in the table — at four confidence levels (80%, 90%, 95%, 99%), as stars or letters.

Behind the scenes run the z-test, chi² and t-test; pairwise and complement comparisons are built in, with an optional Yates correction for chi². So you know whether a higher segment NPS is real or noise.

  • Live filters by region, touchpoint, product or customer group
  • Significance at 80/90/95/99% — as stars (*/**/***) or letters
  • z-test, chi² and t-test; pairwise and complement comparisons
  • Optional Yates correction for chi²
Interactive DataLion dashboard comparing NPS across segments

NPS tracking with a smoothed trend

In tracking, DataLion shows the NPS as a timeline and, on request, smooths it with a moving average (SMA) or an exponentially weighted average (EMA) — with a freely chosen window, to separate sampling noise from the real trend.

New waves are imported automatically, the dashboard refreshes itself, and wave-over-wave change can again be significance-tested.

  • NPS as a timeline, wave after wave
  • SMA/EMA smoothing with a freely chosen window
  • Wave-over-wave change significance-tested
  • Automatic wave import — more on tracking studies
Timeline of an NPS across several waves in DataLion

Driver analysis, not gut feeling

The score does not tell you why it moves. With relative-importance analysis — one of the predefined models on the R engine — you find which satisfaction or experience drivers explain the NPS most, complemented by regressions.

The open follow-up on the why is analyzed in a structured way — as net codes or categorized mentions — instead of just collected.

  • Relative importance: which drivers explain the NPS
  • Linear, ordinal & other regressions on R
  • Analyze and categorize open-ended responses in a structured way
  • Compare drivers by segment

The NPS & eNPS question — with no recodes

Collect the NPS with the NPS question type (0–10) and an open follow-up. On publishing, DataLion automatically builds the project, dataset and codebook with value labels — answers are analyzable immediately, with no recoding.

The eNPS for employees uses the same workflow. Invite by anonymous link, QR code or one-time token.

  • NPS question type (0–10) plus an open follow-up
  • Auto-codebook with value labels — chartable at once, no recodes
  • eNPS for employees in the same workflow
  • Invite by anonymous link, QR code or one-time token
DataLion survey editor with the NPS question type

What you can build with DataLion

  • NPS tracking dashboard

    Score, net rows and a smoothed trend — wave after wave, significance-checked.

    See how →

  • Driver analysis

    Relative importance on R — which factors explain the NPS.

    See how →

  • Weighted NPS

    Weight to your target frame — from the dataset, a table or computed.

    See how →

See DataLion with your own data

Start a free trial or book a personal demo — from raw data to a finished dashboard.

Try it free Book a demo
We now work much more efficiently, giving us more time to take care of the derivations and insights from the data for the customers.
Jens Falkenau, Vice President of Market Research · Nielsen Sports
Read the case study →

The platform in detail

Go deeper

Common questions about NPS with DataLion

Exactly how does DataLion calculate the NPS?
Via a codebook formula: 100 × (share of promoters with 9–10 − share of detractors with 0–6) divided by valid answers. Passives (7–8) do not count toward the score. Missing values are excluded, and the range runs from −100 to +100.
Can I weight the NPS?
Yes. If a weight variable is present, DataLion computes the weighted NPS (each case multiplied by its weight). Weights come from the dataset, from a separate weights table (via a join), or are computed in DataLion to a target distribution.
Are segment and wave differences significant?
DataLion tests this right in the table: differences are tested at four confidence levels (80/90/95/99%) via z-test, chi² or t-test and shown as stars or letters, with an optional Yates correction for chi².
How do I understand what drives the NPS?
With relative-importance analysis — one of the predefined models on the R engine — you find which drivers explain the NPS most, complemented by regressions. You also analyze the open why-question in a structured way.
How do I track the NPS over time?
As a timeline with optional SMA/EMA smoothing and a freely chosen window. New waves import automatically, the dashboard refreshes itself, and wave-over-wave change can be significance-tested.

Ready to compute your NPS properly?

Try DataLion free with your own NPS survey — from the codebook formula to significance-checked tracking. Or book a personal demo.

Try it free Book a demo