NPS

What is the NPS?

NPS (Net Promoter Score) is a metric of customers' willingness to recommend a company, obtained from a single question on a scale from 0 to 10. Respondents are asked how likely they are to recommend the company or product to a friend or colleague. The NPS is the share of promoters (rating 9 or 10) minus the share of detractors (rating 0 to 6) and lies between minus 100 and plus 100. The term “NPS score” is also common, although the S already stands for score.

Companies use the NPS as a management metric in product development, customer service and marketing, and show it in the pitch deck as an indication of their customers' satisfaction. Unlike revenue and retention metrics, it is based not on accounting or contract data but on a survey, for the conduct of which there are no binding requirements. In a financing round or a company sale it is therefore a supplementary figure whose significance depends on the methodology.

Calculating the NPS: scale, groups and formula

The responses on the scale from 0 to 10 are divided into three groups:

• Promoters: rating 9 or 10.
• Passives: rating 7 or 8.
• Detractors: rating 0 to 6.

Formula: NPS = share of promoters in percent minus share of detractors in percent

The result is stated as a number without a percentage sign. The passives do not enter the numerator but count towards the total number of responses and thus dilute both shares. A value above 0 means there are more promoters than detractors.

An open follow-up question about the reason for the rating is customary. The relationship survey covers the customer base at regular intervals, while the event-based survey follows a specific contact, such as onboarding or an enquiry to customer service. In business with corporate customers, the company must also decide who responds: the daily user, the person responsible for the subject matter or the person who decides on the contract renewal.

Limits of the NPS

• Loss of information: The three-way split treats a 6 like a 0. A shift from 7 to 8 remains invisible, whereas a shift from 8 to 9 changes the value markedly. In addition, very different customer bases can have the same value: 50% promoters and 30% detractors give 20, as do 20% promoters without any detractors.
• Statistical uncertainty: With few responses, the value fluctuates considerably. A change of a few points between two surveys can be pure chance.
• Selection of respondents: Response rate and selection co-determine the result. Responses often come from particularly satisfied or particularly dissatisfied customers, and customers who have already left are generally no longer surveyed.
• Different response behaviour: Rating scales are used differently depending on country, industry and customer group. Comparisons across these boundaries are of limited significance.
• Attitude instead of behaviour: The NPS measures a stated intention. High switching costs (lock-in effect) can retain dissatisfied customers, and budget cuts can lead satisfied customers to cancel.

The NPS in a financing round and due diligence

Investors treat the NPS as a complement to the metrics that reflect actual customer behaviour: churn rate, retention rate and cohort data. If the figures fit together, the NPS supports the statement of strong customer loyalty. If they contradict each other, for example with a high NPS and high churn, it loses weight and raises questions about the methodology.

In due diligence, therefore, it is less the value itself that is examined than how it came about. The survey period, population, selection of respondents, response rate, wording of the questions and the raw data including the free-text responses are requested. In the market-related review before a company acquisition, advisers often conduct their own customer conversations or surveys. If their result deviates markedly from the reported NPS, the credibility of the other statements suffers.

What can be shaped above all is the selection. A survey of active users only, a mailing immediately after positive contacts or the presentation of the most favourable survey wave raise the value without anything changing in customer loyalty. If remuneration or target agreements are tied to the NPS, employees have an incentive to ask for good ratings.

Worked example: calculating the NPS of practice software

A manufacturer of practice software for tax advisers surveys 1,000 customers, and 250 respond. This corresponds to a response rate of 25%. 100 respondents give 9 or 10 points, 100 give 7 or 8, and 50 give 0 to 6.

• Share of promoters: 100 / 250 = 40%.
• Share of detractors: 50 / 250 = 20%.
• NPS: 40 minus 20 = 20.
• Random fluctuation: The standard error measures how much the value fluctuates purely because of which customers happen to respond. Formula: square root of ((promoter share + detractor share minus NPS to the power of 2) / number of responses), with the NPS as a decimal. Here: square root of ((0.40 + 0.20 minus 0.20 to the power of 2) / 250) = square root of (0.56 / 250) = around 0.047, that is, 4.7 points.
• Margin of fluctuation (95% confidence interval): 20 plus or minus 1.96 × 4.7 = 20 plus or minus around 9.2, that is, approximately 11 to 29. The value for the entire customer base lies in this range with high probability, provided that the respondents are a random selection.
• If 25 passives move from 7 to 8 points, the NPS stays at 20. If 25 passives move from 8 to 9 points, the share of promoters rises to 125 / 250 = 50% and the NPS to 30.
• Revenue-weighted: If the promoters account for 30% and the detractors for 35% of the respondents' recurring revenue, the result is 30 minus 35 = minus 5.

A rise from 20 to 25 in the next survey would, with this sample, lie within the uncertainty and would not be reliable progress. In business with corporate customers, weighting by revenue can give a different picture: if the detractors are predominantly large customers, the revenue risk is higher than the unweighted value suggests. The value says nothing about the 750 customers who did not respond. If mainly particularly satisfied customers respond, it is biased, and the margin of fluctuation does not capture this bias.

Distinctions: NPS, CSAT, churn rate and referral rate

• CSAT: measures satisfaction with a specific service or contact, usually on a short scale. The NPS targets the overall relationship and the willingness to recommend.
• Churn rate: measures actual customer loss. The NPS captures an attitude that may, but need not, precede churn.
• Share of new customers from referrals: shows how many customers were actually won through a recommendation. This figure is based on recorded transactions rather than on a survey and at the same time lowers the cost of customer acquisition.

Common mistakes

• Selective survey. If only active or particularly satisfied customers are contacted, the result does not represent the customer base.
• Sample and response rate withheld. Without the number of responses and the response rate, the reliability cannot be assessed.
• Small changes over-interpreted. Fluctuations within the statistical uncertainty are presented as a trend.
• Linkage to remuneration. Targets set on the NPS change survey behaviour and devalue the metric.
• Unsuitable comparisons. Values from other countries, industries or survey formats serve as a yardstick in benchmarking although the methodology differs.
• Reasons not analysed. The free-text responses contain the causes of the rating. A report that states only the number leaves the actual insight unused.

Frequently asked questions

What NPS counts as good?

There is no universally valid benchmark. A positive value means more promoters than detractors. More informative than a comparison with other companies' values are the company's own development using the same methodology and the distribution of the responses.

Does the NPS measure customer satisfaction?

Only indirectly. The NPS captures the willingness to recommend, not satisfaction with a particular service. The two are related, but a satisfied customer does not necessarily recommend, for example where recommendations are unusual in that customer's industry. For satisfaction with individual contacts, the CSAT is usually collected.

How often is the NPS collected?

As a rule of thumb, relationship surveys take place once or twice a year, while event-based surveys run continuously after specific contacts. Surveys that are too frequent can reduce the willingness to respond.

How many responses does a reliable NPS need?

That depends on the desired accuracy. With a distribution like that in the worked example, the uncertainty is around plus or minus 9 points with 250 responses and around plus or minus 15 points with 100 responses. For comparisons between segments, each segment needs a sufficient number of responses of its own.

Does the NPS play a role in business valuation?

Not directly. Valuations are based on revenue, earnings, growth and customer loyalty. The NPS can support the assumptions about customer loyalty if it fits the observed churn data, but it does not enter the valuation as a figure in its own right.

The Roemer Capital view

Roemer Capital reconciles a reported NPS with the company's churn and retention figures before a financing round or a sale. If the two do not fit together, it establishes the cause before the value appears in the pitch deck or data room. Alongside the value, it documents the methodology: period, population, response rate and wording of the questions.

Roemer Capital sees the greatest benefit of the survey in the free-text responses. The reasons given by promoters feed into the equity story, and the reasons given by detractors into preparation for the customer conversations that investors may hold in due diligence. That way, management goes into these conversations aware of its known weak points.