NPS Best Practices: The most effective way to send a Net Promoter Score℠ survey

In Zendesk Support, you can send Net Promoter Score℠ (NPS®) surveys to your customers. Although the sending of the actual NPS survey is very simple, you might want some best practices to make sure you go about sending the survey and interpreting the results in the best possible way. This will make NPS a valuable asset to your support team and your company as a whole.

The power of NPS is not in the feature itself but in how you use it. We’ve created a 3-part series to walk you through best practices on how to best utilize NPS:

• What is NPS and how does it help me?
• The most effective way to send an NPS survey (this article)
• Analyze your NPS results and take action

This second article in the series walks you through the best ways to send out your NPS survey campaigns to gather the most valuable feedback without bias.

Customizing your survey

Sending your first NPS survey takes just a few steps.

One of those steps is customizing the NPS survey to fit your company’s brand and use case. When you create your first NPS campaign, you’ll notice that you are limited in the customization of the NPS question, “How likely are you to recommend company x to a friend or colleague?”

The reason Zendesk does not give you free reign to edit this question is so that we can eventually create an NPS benchmark that you can use to compare your NPS against your peers in your industry. Keeping the NPS question consistent in our product ensures the integrity of the survey and enables us to create a Zendesk benchmark for our customers.

However, you can customize the NPS question to fit your company’s brand as well as the type of person that your company would be referred to. There are four options:

• Friend or colleague
• Friend
• Friend or family member
• Colleague

Additionally, on the confirmation page, you can fully customize the follow-up question. The default follow-up question is “Will you share why?” but you can use this as an opportunity to gather additional feedback or ask any question you want.

Your customers are...	NPS question: Referred customer dropdown	Ideas for follow-up question: Ask anything you want
Businesses	Friend or colleague Colleague	What could we do to improve our products and services? How satisfied or dissatisfied are you with the customer care we provide?
Consumers	Friend Friend or family member	What could we do to make you more likely to recommend us? What could we do to make your life as our customer even better?
Employees/coworkers	Colleague	What information or tools would help you do your job better? What could the information technology support team do to serve you better?

Sending out the survey

Ready to ask the NPS question? Be sure to survey a group of customers that is both large enough to allow for statistically meaningful conclusions and varied enough to be representative of your overall customer base.

Survey the right number of customers

What is the “right” number of customers for your organization? That depends on how concerned you are that your sample data are accurate. You’d probably like your survey results to be error free, but, unless you’re willing to survey every customer, that preference is unrealistic. Instead, you can save time and money by surveying only a subset—or sample—of your customers.

Let’s walk through an example for your company. Suppose you are interested in surveying a sample of your customer base and you’re willing to accept a moderate amount of error—about 10%—in your data.

Assuming a 10% acceptable error rate and using the sample size equation in the “Numbers, numbers, numbers” section, you’ll need about 250 customers to respond to your NPS survey.

In order to obtain this recommended sample size, you’ll need to invite a larger number of customers to take the survey than will actually respond to it. We’ll assume a response rate of 15%, which is the typical response rate that other Zendesk Support customers have experienced with their NPS surveys. It’s usually better to invite too few people than too many at first. You can always invite more customers in subsequent surveys.

So, assuming a 15% survey response rate and using the invitation equation in the “Numbers, numbers, numbers” section, we see that you should send your NPS survey to  1,700 customers. What if you’re a smaller company and don’t have enough customers to send the recommended number of invitations? In that case, we recommend you invite all of your customers with the aim of collecting responses from as many of them as possible.

So, if you are not a statistical wizard, your key takeaways are:

• Send the NPS survey to a minimum of 1,700 customers.
• Given a 15% response rate, you can expect to receive approximately 250 responses.

Survey a representative sample of customers

Now you know how many customers you should survey, but surveying just the right number of customers is not enough. You must also survey customers who are representative of your entire population of customers.

Surveys are a way of inferring the characteristics of a large group—for example, all of your customers—from a small sample of it. In order for this inference to be valid, your sample of customers must accurately represent the total population of your customers. But how do you create such a sample?

We recommend using one of the various randomizer tools available on the Internet.

Adopt a regular survey schedule

After you’ve sent your first NPS survey, chances are you’ll want to send follow-up surveys in order to track changes in your NPS over time. Plan to send follow-up surveys on a regular schedule, controlling for events like product launches or website downtime, which can bias your NPS score.

NPS is a long-term measure of customer loyalty, so consider picking one or two dates each year to collect NPS scores and stick with that schedule. If you have a very large customer base, you can run NPS surveys monthly if you’d like. Just be sure to only ask the same customer for their rating no more frequently than every six months.

To help facilitate a regular survey schedule, Zendesk’s NPS feature comes pre-built with a best-practice survey setup. You’ll be able to send an NPS survey once a day. Additionally, a customer will receive your survey only once every 90 days. This way you’re not accidentally spamming your customers.

Don’t survey more frequently than you can respond

If you’re looking for guidance on how frequently you should survey, a great rule of thumb is to only survey as frequently as you can digest the customer feedback, respond to your customers, and create an action plan. It does your team and company no good if you do not have enough time to analyze the customer feedback and responses, understand what’s going on with your business and customer experience, and set forth next steps for improvement.

When you apply these best practices to your NPS survey campaign, you’ll be ready to start asking customers for their NPS ratings. As you gather responses and feedback, check out the final article in our best practices series, Analyzing your NPS results and taking action.

Numbers, numbers, numbers

So where did all these numbers come from? Recommended sample size was calculated using the equation:

where:

• n represents the recommended sample size.
• z represents the standard deviation associated with the chosen confidence level. We recommend using a 95% confidence level, which essentially means you’d expect to get the same NPS score 19 times if you surveyed your customers 20 times within the same time period.
• e represents the margin of error you’re willing to accept. We recommend a margin of error of +/-3% if you’re willing to accept only a low degree of data inaccuracy, +/-5% if you’re willing to accept a moderate degree, and +/-10% if you’re willing to accept a high degree of data inaccuracy. A margin of error of +/-5% is most common.
• represents the variance in NPS scores and was calculated using the equation

  

  where:

• represent the population proportions for Promoters, Passives, and Detractors,

  with = and NPS = .

  For your first NPS survey, we recommend assuming that your customers are equally distributed across the three NPS groups, so that ⅓ are Promoters, ⅓ are Passives, and ⅓ are Detractors. You can adjust these proportions as you get a better idea of how your own customers are distributed across the three groups.

Using the formula:

and applying the guidelines noted above, we calculated recommended sample size using a high level of tolerance:  +/-10%. If you’re willing to accept only a moderate level of error, you’d replace “0.10” in the denominator below with “0.05.” If you need especially accurate data, replace “0.10” with “0.03.””

n =

Recommended number of invitations was calculated using the equation I = n/r where: n represents the recommended sample size and r represents the expected response rate.

I = 256/.15 =1706

Again, if you are not a statistical wizard, your key takeaways are:

• Send the NPS survey to a minimum of 1,700 customers.
• Given a 15% response rate, you can expect to receive approximately 250 responses.

**Net Promoter, NPS, and the NPS-related emoticons are registered trademarks, and Net Promoter Score and Net Promoter System are service marks, of Bain & Company, Inc., Satmetrix Systems, Inc. and Fred Reichheld.