NPS Survey Best Practices | CustomerGauge NPS Survey Best Practices

Request Demo

NPS Survey Best Practices

Designing and conducting Net Promoter System (NPS) surveys isn’t complicated and it doesn’t require a team of engineers or analysts. However, with so many companies needing to put their own unique stamp on NPS, it’s not surprising that it has become confusing and taxing.

Picture a travel company called “Imaginary Travels” is in need of an NPS program. As they design and implement NPS, the company’s digital marketing manager says “we should ask how important are different features on our website.” The travel guides wish to know how satisfed customers are with their knowledge and service. And fnally, upper management wants surveys that cover each possible touchpoint. By the end of the planning phase, their NPS project becomes a bloated, multi-purpose yet inefcient undertaking.

At CustomerGauge, we love NPS. And we think it’s the number one way to measure and improve customer loyalty. But we also believe that things should be simple - meaning simple to do and simple to understand - because having a system that only a few understand and is slow to execute, will get little buy-in and fail to improve customer experience.

In this best practice guide, we teach you how to generate reliable data that you can make decisions on by asking the right questions, to the right people, at the right time. And with some help from our imaginary travel company, we show you how to do this in a way that is simple, fast and actionable.


How to design an NPS survey 

1. Where should I start?
2. Relationship and transactional surveys: What are they and which one is for me?
3. How many questions should I ask?
4. What questions should I ask?
4.1 How to formulate a transactional survey question.
4.2 You need to know the reasons for their score.
4.3 How to choose predefned reasons.
4.4 Add a comment box.
4.5 Give customers the option to be contacted 
5. Visuals

How to conduct an NPS survey

6. Customer data and automation
7. What customer data should I upload?
8. How many should I survey?
9. Relationship surveys - who, when and how often?
10. Transactional surveys - who, when and how often?
11. What’s the best day of the week or hour to send surveys?
12. Through what channels should I send surveys?
12.1 The many ways to send transactional surveys.
12.2 Are you mobile friendly? 
13. What do I say in a survey invite?
13.1 Email subject lines.
13.2 An effective invite.
13.3 Send reminders.
13.4 How to avoid email spam flter
14. Test your survey

How to design an NPS survey

The design component of an NPS survey is predominantly about what should you ask and how should you ask it. And while many make the process more difcult than it needs to be, follow the rules in the following chapter and you will have an NPS that is both effective for companies and easy and fast for customers to do.

1. Where should I start?

NPS is about being easy and fast to start. It’s about getting your frst survey out there as fast as possible and receiving your frst round of feedback.

Many choose to start a survey program by conducting a long planning period to decide the type of surveys, how many different surveys need to be created, and what goes into each one.

Implementing a survey program in this way, means investing a lot of time and resources before you have a single survey out there. And you won’t know if you made the right choices until you fnally get the program out there.

Roll out NPS in stages. Start with the core of NPS, the relationship survey. As you master your frst round of surveying, this will educate you about how to expand the program by adding new surveys, surveying larger numbers or different company segments.

2. Relationship and transactional surveys: What are they and which one is for me?

Relationship surveys

Relationship surveys seek to investigate a customer’s loyalty to a company/brand. These types of surveys ask customers to consider the overall experience and satisfaction they have with a company. They are typically carried out at regular intervals (e.g., quarterly, half-yearly); however, there are exceptions (see section 9).

Transactional surveys

Transactional surveys investigate the experience a customer has in a specifc transaction/interaction (touchpoint). This survey is designed, not to measure customer loyalty, but to measure satisfaction with a specifc company segment in order to improve it.

Whether you use only a relationship survey, or you add transactional surveys is dependent on your business and your needs. But it is important to be aware of what both mean and how and when to use them.

We recommend that you begin your NPS program with a relationship survey to investigate customer loyalty and brand perception, and uncover which touchpoints are affecting it - be it good or bad. Transactional surveys should then investigate those touchpoints that relationship surveys indicate as the most important to the customer experience.

One of Alaska’s biggest Internet providers and a CustomerGauge client, Alaska Communications, uses both. After starting out using only relationship surveys, they came to realize that they weren’t getting the whole picture. They needed to dig deeper into the touchpoints that customers indicated as strongly impacting the customer experience, such as their contact center, feld operations, and service delivery.

Transactional surveys helped them understand the performance of singular company segments, but more than this both surveys complemented each other. Relationship surveys captured the voice of the low-contact customer and uncovered issues that transaction surveys were not capturing, such as internet speed issues. Transactional surveys helped them learn about those frequently occurring issues such as the quality of an installation or the level of knowledge of a technician.

3. How many questions should I ask? 

An NPS survey isn’t about investigating customer satisfaction across many company segments, which then requires analysis to fnd the key drivers of loyalty. It is about uncovering customer loyalty and what most affects it.

Customer feedback surveys identify what you need to focus on to improve the customer experience, so knowing how customers rate less important drivers has little if any value. Many people in your company might demand to know how customers rate certain touchpoints, but all they really need to know is whether it impacts customer loyalty.

Keep it short, a maximum of four or fve questions. Long surveys mean low response rates, survey fatigue resulting in inaccurate answers and data that isn’t immediately actionable.

Our best results have always occurred with just three questions. Shorter than this may produce higher response rates, but a survey limited to just the NPS question will produce data that has limited actionability. With just three questions, you can gather all the data you need to create actionable insights and still receive high response rates (40% - 60%). The following chapter will explain these three questions.

4. What questions should I ask? 

“On a scale of 0-10, how likely is it that you would recommend Imaginary Travels to your friends, family or business associates?”

We all know the question above as the Net Promoter question; developed in 2003 by Fred Reichheld. It is the basis for the relationship survey. His team tested a range of loyalty questions and found that the “would you recommend” question correlated the most with a customer’s loyalty: their likelihood of repurchasing or renewing with the same company.

There is a long debate about the best way to formulate this question to measure loyalty (some even promote asking directly “how loyal are you to this company”), but if you want to do it, save it for later when you are more experienced with NPS and can do reliable testing.

Important Note

4.1 How do I formulate a transactional survey question?

For NPS transactional surveys, the question changes. As asking someone’s likelihood to recommend a company is not related to a customer’s satisfaction with a transaction. To remedy this, many construct the NPS question as such:

“Based on your recent online experience with booking your holiday, how likely are you to recommend the experience to your friends, family or business associates?”

However, if you measure a high-volume transactional touchpoint, asking people about their likelihood to recommend may defate the value of the question. Consider alternative questions such as “how satisfed” or “please rate.” An even better option may be the “Customer Effort Score” (CES). For customer service issues CEB Towergroup’s research showed that it had a better correlation with loyalty than the NPS question.

The CES question is typically formulated as:

“To what extent do you agree or disagree with the following statement: The company made it easy for me to handle my issue with customer support.”

When to use which is dependent on the transaction. For example, our travel company sells large and expensive package holidays. The sales of these large holidays can be a long process with a booking agent. Asking customers how likely they are to recommend the sales service after a booking helps the company understand the performance of the agent.

However, for the customer support department of Imaginary Travels, it might be better to ask how much effort a customer’s transaction took. As customers can have multiple interactions with customer support, asking their likelihood to recommend after many interactions provides less and less benefts.

4.2 You need to know the reasons for their score 

Reichheld’s original “would you recommend” question is great, but you need more than a score. A score alone only tells you how you are performing not why. It is now generally agreed that asking customers what drove their score, is necessary.

The easiest way is to ask customers to identify only the most important reasons (drivers) for their score and present customers with a predefned selection of reasons, rather than have them submit their own reasons. 

Detractors and Passives

While, presenting predefned reasons provides you with a set of uniform answers, rather than having to continually group customer created responses under particular themes.

How you ask customers what affects their score should be different for promoters than detractors and passives. Returning to our travel company, take a look at the two relationship survey examples below to see how Imaginary Travels can discover with just a single question what is most impacting their customer experience.


In the examples, the customer is presented with a two-tiered set of reasons for their score. As a customer selects a “level 1” option (e.g. Our website) for their score, your survey should present a set of “level 2” options for selection (e.g. The overall appearance).

Constructing the survey this way keeps it short and simple for the respondent, as they only need to think about what is most important to them, which leads to higher response rates. 

When constructing the question in this way, we have found that 82% of respondents answer this question. Finally, it is also important not to restrict the customer to selecting only one reason; customers should be able to select as many reasons as they deem important. 

4.3 How to choose predefined reasons. 

For the relationship survey, start by mapping out the customer journey to fnd customer touchpoints, and select those touchpoints that receive the highest customer trafc to be used as your level 1 options. Level 2 options then are the strengths and weaknesses of every level 1 option.

Pay attention, the level 1 options of your transactional survey will cover a much narrower group of touchpoints than your relationship survey. For Imaginary Travels, a transactional survey after a trip would expand the relationship touchpoint “your holiday experience,” we see in the example above. Level 1 options then become:

The level 2 options for the example above, would then once again be the strengths and weaknesses of each level 1 option.

4.4 Add a comment box.

“Level 2” options aren’t a complete list of strengths and weaknesses. As you can see in the example in 4.2, you always add the option: “Other (please elaborate in the comment box below).” For this reason, always remember to add a comment box.

In our experience, 44% of survey respondents leave a comment. Frame the question as voluntary, to ensure the survey appears as short as possible and convey that you do care about their feedback, like so:

Comments are also an invaluable source of direct information from your customers, which is hard to argue against. If a customer scores a company a fve its accuracy can be debated. However, when a customer says, “your product is great, but each time I need to ring customer support it takes forever,” it is clear and inarguable how the customer perceives the company. 

Of course, analyzing free text is always a challenge, even with advanced text analytics. Your NPS survey tool should, in real-time, identify keywords to be submitted as they are or changed by the customer. Customer comments can then act as an additional means to identify key drivers behind customer scores. 

4.5 Give customers the option to be contacted 

Add a simple tick-box asking customers “may we contact you to discuss your experience in more detail?” Doing this conveys to customers that you really are listening and willing to act on their feedback.

5. Visuals 

You need to think about the appearance of your survey. Customers can just as easily be turned off by how it looks, as they can be by a poorly worded question.

Pay attention to colors, font, font size, images and the layout of your survey. The type of market you serve will help guide this process and the type of atmosphere you want to create.

Industries like banking and fnance might choose to create a sense of seriousness by keeping it minimal with survey colors aligned with the company colors, a neat and well-placed company logo, and formal text.

While for our travel company, it would be more effective to aim for a more relaxed feel to create familiarity, by incorporating holiday related images, multiple color schemes and text that is more familiar and casual.

Later, when you are established with NPS, you can test different surveys to see what works best. 

How to conduct an NPS survey

The following chapters explain how best to conduct an NPS survey to ensure that, (1) you send the right survey to the right people at the right time, (2) you create a survey process that is easy, and (3) you receive high response rates.

6. Customer data and automation

Whether you use a Customer Relationship Management (CRM) tool or simply a spreadsheet, customer data needs to be uploaded into your NPS survey tool to avoid manually entering contact details when the time comes to send your surveys.

For those using a CRM, make sure you use survey technology that can not only integrate with your CRM but also automate as much of the conducting and collecting process as possible.

Read the following to learn more about what makes a good automated system. 

7. What customer data should I upload?

To get started with your NPS survey, upload customer contact details and basic demographic information (e.g. age, gender, etc.) to create a profle for each piece of feedback.

Later, if you are processing and analyzing customer feedback in your NPS tool, you will need to upload more extensive customer data such as purchase history, support history, etc., to create a better understanding of each customer’s response.

8. How many should I survey?

What is a good sample size, what error of margin or confdence level should I aim for with NPS? These questions unnecessarily plague many who frst start out with NPS, and who can blame them, as these concerns have been drilled into everyone who does surveys.

Sample size and statistical signifcance are important in NPS, but not for your frst surveys. While one part of NPS is about aggregating feedback to drive organizational change (meeting statistical standards), its biggest strength is as a voice of the customer program. It is a program designed to raise loyalty by resolving individual customer issues.

Release your survey to your customers over weeks and months (depending on the size of your customer base) and if you are worried about sample size, you will over time create the sample size you need, as NPS surveying is a continual process.

9. Relationship surveys

9.1 Who should I survey? 

In B2C, all customers are eligible for an NPS relationship survey. However, what you use NPS for will determine which customers should be surveyed. For example, an airline might choose only to survey its economy class customers because it knows that’s where improvement is needed. Generally though, in B2C, customer bases are pretty uniform and companies will survey all their customers.

In B2B, sending a relationship survey means sending surveys to many contacts within an account. All customers in B2B should be surveyed, but it is important to determine which contacts in each account need to be surveyed. To learn how to do this, read:

“Why “Voice of the User” is not “Voice of the Customer” in B2B.”

9.2 When should I send?

When to send a relationship survey during the year is almost arbitrary; however, there are some things to be aware of.

Don’t send out a relationship survey to a customer, if they have only been a customer for 24 hours. Wait until there is a relationship to talk about, before you send a relationship survey. How long this takes, depends on how frequently customers use your products or services and interact with your company. 

You need to create rule-based sending. If a transaction survey is due just days before a relationship survey, consider the effect of sending two surveys close in time.

Relationship surveys need to take priority over transactional surveys. They are more valuable to your NPS program (particularly for B2B), and if they follow shortly after a transactional survey, the response rate and even the answers themselves may be affected.

In some instances, you will know when a transaction is going to happen, e.g. a sales call, and you can plan accordingly. However, there are many other transactions that you won’t see coming, such as a call to customer support. Set up rules then that stop transactional surveys from being sent just before scheduled relationship surveys or have relationship surveys automatically rescheduled to a later date.

Lastly, do not survey your customer base all at once. It will cause a bottleneck of data that you will not be able to process and respond to quickly. Either survey each customer at a set time from their starting date or survey all your customers in one block, but spread over weeks or months. 

9.3 How often? 

Relationship surveys should be sent at least once a year, preferably twice. Depending on your industry, you may be able to get away with surveying every 3 or 4 months. Be careful, though, survey too often and you will annoy customers, which will decrease response rates.

Relationship surveys are generally sent periodically, but they can be sent after an interaction if needed. For companies in single purchase models such as in-store retail, there is little opportunity to conduct a relationship survey outside of the interaction itself.

10. Transactional surveys

10.1 Who should I survey?

What transactions you should survey, really depends on your industry context and the aim of your NPS program.

Transactional surveys are no more suited to B2C or B2B, as is sometimes claimed. Transactional surveys allow companies to further investigate how particular touchpoints affect a customer’s relationship with a company and make service improvements. As such, it is something that benefts all industries.

As we advise NPS to be rolled out incrementally, start by doing transactional surveying for just one or two of your most important touchpoints. Important touchpoints are those that customers continually select in your relationship survey.

Later, when you have mastered a small number of transactional surveys, expand your transactional NPS to cover other touchpoints that your relationship survey uncovers.

10.2 When should I send? 

You should send a transactional survey immediately after a touchpoint. Waiting a day or two before sending a survey will mean the customer’s memory of the transaction will have changed, and the customer’s incentive to do the survey will have diminished.

Be mindful though of how you defne the boundaries of the touchpoint. If you want to measure the whole experience from browsing/ordering to receiving the item, you should time your survey for when the item should arrive, not after the purchase.

10.3 How often? 

Sending a transactional survey too frequently can be damaging to your data and response rate.

For example, a customer calls a support center to resolve an issue, but to resolve the issue requires 3 or 4 interactions in the space of a day. If you survey after each interaction, this isn’t going to raise new insights, and it will produce a low response rate.

Leave sufcient time between transactional surveys, so when completing a survey, the customer is not thinking about a previous transaction but about the one they just experienced. This is particularly important for customers that experience two different touchpoints in a short period. As the experience with one company segment could affect the feedback about another company segment. 

11. What’s the best day of the week or hour to send surveys?

We felt compelled to answer this question because it is commonly asked about improving response rates. However, for those just starting out, our answer is: It doesn’t matter!

Experimenting with timing is only something you should attempt when you have an operationalized NPS, where you have the time and skills to test which days or hours work better. However, even this is largely pointless, as we have found.

12. Through what channels should I send surveys?

There are many ways to send NPS surveys to customers, but deciding which channel to use is
about convenience for your customers and what will receive the best response rate.

Two of the most common methods are to send survey links by email or SMS. This is particularly useful for relationship surveys, where the customer has not had any recent contact with a company. 

12.1 The many ways to send transactional surveys  

Transactional surveys offer a lot of options for sending, as surveys can be presented to customers immediately after a transaction. Think about the industry you are in, and how you can “grab” a customer straight after a transaction.

Retail stores often put a survey link at the bottom of receipts while others ask customers if they would like to complete the survey via dedicated in-store tablets. Online, many companies present an NPS survey link on the fnal screen of a transaction, while in the software industry some present it during the use of an app or software tool.

What opportunities exist immediately after your touchpoints? Where could you invite the customer to do a survey?

We strongly recommend using email over all other channels, as it is not a dead channel of communication (as some claim). Email gives you the space to construct a proper invitation and it is the best way to conduct relationship surveys.

12.2 Are you mobile friendly? 

No matter what channel you use to send your surveys, make sure your survey can open on mobile devices. Increasingly people are living by their phones, and although computers still dominate, a survey invite is often something a customer might leave for the bus ride home. 

13. What do I say in a survey invite? 

How do you interest customers to click the link to your survey? What do you say to them in an email, SMS or post transaction screen?

In your invite, pay careful attention to each word you write and eliminate everything that isn’t necessary. Are you repeating yourself or communicating things that aren’t relevant or interesting to the customer?

13.1 Email subject lines 

There are psychological triggers to get the attention of customers, such as using emotion or asking respondents a question. Your subject line should be brief, straightforward and friendly. Examples of good subject lines: 

Email Subject Lines

To learn more about this read our article: “6 email subject lines strategies to increase your survey feedback.”

13.2 An effective invite

Whatever channel you use to invite customers, there will always be a body of text. To create an invite that gets the highest possible click-rates, follow these rules:

An effective invite
An effective invite

13.3 Send reminders 

Sending survey invite reminders is a good way to improve your response rate. As many customers, particularly in email, may not see or be able to respond to your survey the frst time around.

A standard waiting time for CustomerGauge clients is 3 - 7 days. You don’t want to remind the customer too soon after the frst request, but if it is a transactional survey you don’t want to wait too long either.

13.4 How to avoid email spam filters 

Various things may trigger spam flters. Avoid these common red fags in your invitation email:

How to avoid email spam filters

14. Test your survey 

Before you send anything out to customers, test your survey internally. Send it to other employees within the company, to test technical problems.

Make sure your email engine is working, the emails open in the format and layout intended, links to the survey work and the survey itself runs smoothly for other employees.

Test your survey across different browsers. And although we think everyone uses Gmail or Outlook, make sure your email invitation works in all major email services.

Ask employees what they think of the survey and the process around it. Do they have any issues or ideas about such things as the invitation or the survey itself?

Testing NPS surveys isn’t difcult and it will help ensure you create the most reliable data possible.

NPS needs to be simple and fast.

The aim of this guide was to demonstrate how simple and quick it is to get started with NPS. Unlike other survey types, designing and conducting an NPS survey does not require a long planning process. At just three questions long, an NPS survey can be designed in a day and sent out the next. For the longer you wait, the more insights you miss and the more customers you lose.

Don’t worry about surveying all your customers or designing a survey for each division of your company. Start small, so that you get started fast.

Use a simple relationship survey on a select group of customers, to show upper management how fast it is to identify customer experience weaknesses and turn data into action.

And as you master your frst round of relationship surveys you will learn while benefting from it, how to increase the scope and the scale of your NPS. 

To fnd out more about To fnd out just how easy and fast NPS can be, try CustomerGauge’s for free or visit us at to learn more about our mission to help companies create happier customers.

Go Beyond Surveying.
Take Action.

You’ve measured your customer experience, now act on the insights. Demo CustomerGauge’s robust reporting platform and see how to turn survey data into streamlined, intelligent workflows.

Next Up: NPS® Guide for the Retail Industry

Want a PDF copy of "The Fine Art of Surveying" whitepaper?

Get the eBook

Might we also interest you in...