What data does it need?

Two fields: Score (0–10) with the raw likelihood-to-recommend answer as a number, and Value, a respondent count or a Sum of Weight measure. Wave / Period, Segment and Sum of Weight² (significance) are optional. Non-numeric answers in Score are ignored.

Why does it need the raw 0 to 10 answers instead of a computed score?

Because the margin of error and the significance tests come from the distribution of answers, not from the score. A pre-computed NPS measure cannot tell whether a change is real; the raw answers can.

How is a wave-over-wave change tested?

With a z-test on the NPS variance of the two waves, on effective bases when weighted. The headline says ✓ when the change is significant and (n.s.) when it is not, and the caption says how big a wave a non-significant change would have needed. The formulas are on the methodology page.

Can I use eNPS or a different cut-off?

Yes. Detractors Up To and Promoters From in Format → Score Mapping set the cut-offs, so a 1 to 10 scale or another convention works. Score Label is a text field: type eNPS or tNPS and the headline, caption and tooltips use it.

Which wave is the headline?

The last wave in the Wave / Period field's sort order. Change vs in the Display card compares it with the previous wave or the first wave; comparing with the target needs the premium Target card. Sort the wave field chronologically, for example by a date or a wave number.

Can I use weighted survey data?

Yes. Put a Sum of Weight measure in Value and the score and split are weighted, in the free tier too. Map Sum of Weight² (significance), SUM of the squared weights, and the margin of error and every test use Kish effective bases; that part is premium.

Do viewers without a licence see a broken visual?

No. The score, its margin of error, the split, the histogram, the trend with its confidence band and rolling average, the smart caption, the footnote and every cosmetic option are free for every viewer. The wave-over-wave tests, segments with letters and movement, effective bases from Sum of Weight², and the target (the line on the trend and the gap to target) need a per-user licence in reading view; in edit mode everything previews free. Where Power BI reports that licences cannot be checked, for example Report Server, the premium features stay on.

Why is the 0 to 10 histogram not shown?

It is off by default. Switch on Show 0–10 Histogram in the Display card, or pick the Executive summary Quick Style. It needs a tile at least about 260 pixels tall for the plot; in a smaller tile it stays hidden.

How much data can it take?

Power BI hands the visual up to 2,000 rows, one per combination of Score, Wave and Segment, which covers 11 scores across many waves and segments. If your data has more combinations, the visual shows Power BI's warning icon instead of silently leaving rows out.

Which languages does it support?

English and Czech for the format pane, the smart caption, the footnote and the empty-state help. Other Power BI locales show the English strings.

Does it send my data anywhere?

No. The visual requests no network access, makes no external calls and keeps no data in the browser. Every statistic is computed inside the visual, and the licence check is done by Power BI itself.

How do I report a bug?

Use the Report a bug block on the support page, or write to hello@smartvisuals.io. A short description, your Power BI version and, if you can, a screenshot or a sample file help us reproduce it.

NPS Visual for Power BI | Net Promoter Score with Margin of Error, Trend & Tested Changes

NPS in Power BI: the score is easy, the statistics are not

Computing a Net Promoter Score in DAX takes three measures: the share of 9s and 10s, the share of 0 to 6, and the difference. What DAX does not give you is the margin of error on that score, or an answer to the question every quarterly review asks: the score moved from 41 to 44, is that real? NPS has an unusual variance, because it is the difference of two proportions from the same sample, and a card visual cannot know it.

SmartVisuals NPS carries the statistics. Map the 0 to 10 score and a value and the headline appears with its margin of error and the promoter / passive / detractor split; the 0 to 10 histogram is one switch away. Add a Wave field and every wave-over-wave change is tested, so a +3 is reported as noise when it is noise, and the caption says how big a wave a non-significant change would have needed.

The headline, the split and the trend

The score renders as a number, a badge or a face, coloured by band, placed left, centre or right, counting up when it changes. Under it: the margin of error, the change against the previous wave, the first wave or your target, the 0 to 10 histogram, the split bar with shares and bases. The trend over waves comes with a confidence band and a three-wave rolling average.

Cut-offs are remappable for eNPS or tNPS conventions (Detractors Up To, Promoters From), and the Score Label field takes NPS, eNPS or tNPS. Four themes, fonts, bar shapes and three fills of your own cover the cosmetics, all free.

Gap to target: the score as an action

A target of +45 and a score of +41 is a four-point gap, which tells nobody what to do. With a target set (premium), the visual turns the gap into respondents at the current base: "To reach +45: 60 more promoters or 60 fewer detractors". The smart caption also names the best and worst segment and whether the last change was significant, offline, in your language, and read out by screen readers.

Segments with letters and movement

Add a Segment field in the premium tier and each segment gets its score, its split bar and its tested change since the previous wave, lettered against the other segments and sortable by movement. Low bases are hatched and flagged, Bonferroni correction is optional, and a target line sits on the trend.

Map Sum of Weight and Sum of Weight squared and the score, its standard error and every test use Kish effective bases, so a weighted CX study reports the same numbers in Power BI as in the agency's tables.

What the visual expects from your data

One column with the 0 to 10 likelihood-to-recommend answer and one value, a respondent count or a weight. Wave and Segment are optional grouping fields. The visual works from the raw answers, not from a pre-computed score, which is what lets it compute the variance and the tests. The same engine powers the Likert and Crosstab visuals and the all-in-one SmartVisuals Research visual; the formulas are on the methodology page.

Role	Display name	Kind
score	Score (0–10)	Grouping
wave	Wave / Period	Grouping
segment	Segment	Grouping
value	Value	Measure
weightsq	Sum of Weight² (significance)	Measure

Role

Display name

Kind

score

Score (0–10)

Grouping

wave

Wave / Period

Grouping

segment

Segment

Grouping

value

Value

Measure

weightsq

Sum of Weight² (significance)

Measure

NPS

Find the answer in the NPS docs.

Where the line is, exactly.

NPS in Power BI, explained.

NPS in Power BI: the score is easy, the statistics are not

The headline, the split and the trend

Gap to target: the score as an action

Segments with letters and movement

What the visual expects from your data

Where NPS earns its place.

Customer NPS headline with the 0 to 10 distribution

Monthly eNPS pulse for HR

Regions ranked by movement (premium)

Leadership target with the gap in respondents (premium)

From empty canvas to first chart.

Two ways in.

Readable and usable for everyone.

What changed, and when.

Asked before you did.

When it does not look right.

What goes in, and every setting.

Tell us what is missing.