Net Promoter Score Calculator
Enter counts for promoters, passives and detractors to calculate Net Promoter Score, category shares and response total. The page keeps the −100 to +100 scale distinct from an ordinary percentage.
Classify the survey responses
Report the category shares, valid response count, survey period and collection method with the headline score.
How response scores become NPS groups
Net Promoter Score begins with a specific recommendation question answered on a scale from 0 to 10. Responses of 9 or 10 are promoters, 7 or 8 are passives, and 0 through 6 are detractors. The grouping is not based on local school percentages or an ordinary satisfaction grade. Apply the published cut-offs consistently before entering the category counts.
The calculation is percentage promoters minus percentage detractors. Passives are included in the total response count, which lowers the shares of both active groups, but they are not separately added or subtracted. If 72 of 100 valid respondents are promoters and 10 are detractors, the score is 72 − 10 = +62. The result is written as a number from −100 to +100, not “62%”.
Why NPS is not a standard percentage
A percentage normally describes a part of one whole and stays between 0% and 100%. NPS is the difference between two percentages and can therefore be negative. A score of zero might mean equal promoter and detractor shares, but it does not say how many passives there were. Several very different response distributions can produce the same NPS.
Always retain the promoter, passive and detractor shares beside the headline score. Also report the valid response count. A +40 based on 20 responses does not have the same precision or coverage as +40 based on 20 000 properly sampled responses. The calculator performs the arithmetic but cannot determine whether the sample reflects the customer population.
Preparing survey data
Use one response per eligible respondent for the stated period and touchpoint. Remove test records and investigate duplicate submissions using a pre-agreed rule. Do not discard low scores merely because the comment is blank. Missing answers to the recommendation question should be counted separately as non-response, not placed into the detractor category or silently treated as zero.
Keep the original 0–10 response data where governance permits. Category totals alone cannot show movement within a group, such as a change from 0 to 6. Store the question wording, language, survey channel, dates and eligibility definition with the extract so a later period can be compared on the same basis.
Comparing South African customer segments
A national result can hide differences by province, language, product, branch, digital access or customer tenure. Segment comparisons are useful only when the survey process is comparable and each group has enough responses. A small branch can move sharply because of a handful of people. Publishing a ranked league table without uncertainty and context may encourage staff to chase the metric rather than address customer problems.
Where survey invitations use email, app notifications or SMS, coverage can differ across customer groups. Response rates and channel availability should be examined before treating the respondents as representative. Weighting may be appropriate in a formal research design, but a weighted NPS requires respondent-level weights and cannot be reproduced by entering unweighted category counts into this simple calculator.
Interpreting change over time
A higher score is not automatically evidence that an intervention caused improvement. Seasonality, changes in customer mix, survey timing, response rate and random sampling variation can move the result. Compare like periods and preserve the same question and classification. A rolling average can reduce noise but also delays recognition of a sudden problem.
Show the underlying counts and shares for every period. If detractors fall while promoters remain steady, the operational response may differ from a situation where promoters rise but detractors are unchanged. Link scores to coded reasons, service data and qualitative comments while protecting personal information. NPS is a signal for investigation, not the diagnosis itself.
Rounding and score presentation
The page calculates category shares from unrounded counts, subtracts them, and then rounds the displayed result. Subtracting already-rounded percentages can differ by a tenth or a whole point. Establish a reporting convention, such as one decimal for analysis and a whole number for an executive chart, and apply it consistently.
Include a plus sign for a positive score and retain the minus sign for a negative score. Avoid adding a percent symbol to the headline. If a dashboard uses colour bands, document who set the thresholds and why; there is no universal colour boundary that makes one score “good” across every industry, relationship type and survey method.
From score to action
Close-the-loop programmes should prioritise customer consent, safe handling of personal details and realistic service recovery. Contacting a detractor may reveal a solvable issue, but it can also feel intrusive when the survey did not explain follow-up. Aggregate themes can guide process changes without exposing an individual response more widely than necessary.
Track operational measures that relate to the identified issue, such as first-contact resolution, delivery reliability or waiting time. If those improve and the survey remains comparable, NPS may follow. Reward systems based solely on the score can create pressure to coach respondents or suppress invitations, damaging the measurement. Use a balanced view and audit the invitation process.
Relationship and transactional surveys
A relationship survey asks about the customer’s broader view of an organisation, while a transactional survey follows a particular interaction such as a delivery or support call. Their scores answer different questions and should not be combined without a designed weighting method. The invitation timing also matters: asking immediately after a resolved complaint can produce a different perspective from a periodic survey sent to all active customers.
Name the survey type in reports and avoid comparing a branch’s post-transaction score with a national relationship score as though the populations were identical. If the programme changes its trigger, keep a break in the time series or analyse both methods in parallel.
Question wording and translation
Small wording changes can affect how people interpret the 0–10 scale. Use a stable, tested translation and preserve the reference brand, product or service named in the question. Where respondents choose among South African languages, monitor whether response patterns differ because of customer mix or wording. Translation should aim for equivalent meaning, not a literal word-for-word construction that changes tone.
Follow-up questions should invite the reason for the score without pressuring the respondent to change it. Store the language and version with the response so revisions remain auditable.
Questions that affect this result
Are passives excluded from the total?
No. Passives are included in the denominator used for promoter and detractor percentages. They simply do not contribute directly as a positive or negative group.
Should NPS have a percent symbol?
No. The score is a difference in percentage points on a scale from −100 to +100. Report category shares separately with percent symbols.
Can two surveys with the same NPS be different?
Yes. They can have different passive shares, response counts and customer mixes. Keep all category shares and sample context beside the score.
Does a higher score prove service improved?
No. Survey mix, timing, response rate and sampling variation can change the score. Combine the trend with operational and qualitative evidence.
How should missing recommendation answers be treated?
Exclude them from valid NPS category counts and report non-response separately. Do not convert a missing answer into a score of zero.