After you've collected Net Promoter surveys, it's time to learn from them. Analyzing your NPS results is a critical step towards taking action on your customer's feedback.
In this article, we break down your NPS results into three data points:
- The NPS score itself
- The NPS drivers
- The verbatim free-text comment box
Each one has several ways to analyze it.
From benchmarking to text analytics, we'll cover best practices for getting to the root cause of your score. We also deep dive NPS driver analytics, walking you step-by-step through an NPS driver analysis example.
Table of Contents:
Advice before starting:
How to analyze your NPS survey results:
How to get to the root cause of NPS:
NPS Analysis Best Practice—Our advice before you start
1. Is your net promoter representative?
To get results you can trust from an NPS results analysis, we recommend first ensuring that 40% or more of your customers respond to surveys.
Rob Markey of Bain & Co notes that this rate of response should provide a reliable, representative score. That being said, sample size and statistical significance is largely irrelevant with the Net Promoter Score.
Firstly, it’s not normally distributed. Customers respond to surveys in predictable ways based on how much they really like (or dislike) your product—thereby influencing the final results in a way that counteracts statistical techniques.
Secondly, we don’t recommend treating each customer equally. In the B2B customer experience space, where CustomerGauge specialises, we help our customers link their Net Promoter program to the revenue each of their accounts generate.
When looking at it through a revenue and growth lens, it becomes clear that your largest accounts (those that represent 20-40% of your company’s revenue) should be listened to more deeply than smaller ones. And, likely, their feedback should be integrated quickly to inspire loyalty and increase customer lifetime value.
When you look at feedback on an account-by-account basis, inferential statistics are irrelevant.
2. Which customers should you focus on first?
Not all customers have equal value to your business, and not all responses themselves should be treated equally either.
After selecting your most important and profitable customers, we recommend analyze and closing the loop with detractors, passives and promoters, in that order.
Detractors (leaving a score of 0 to 6) should be given the most attention. Our research suggests that companies should close the loop with NPS detractors within 48 hours to see the best results (those that do see a +6 NPS score and a higher retention rate). Being passionately negative is better than being totally disengaged, so by focusing on reaching out to detractors and showing you’ve listened to them can often lead to flipping detractors into promoters in your next round of surveying.
NPS passives (score of 7 or 8) also deserve attention. They’re in the messy middle and are likely to be tempted by competitor offers. By analyzing their results you’re likely to uncover some quick wins for improving your NPS score.
You would also be wise to close the loop with NPS promoters (score of 9 or 10). Our customer brought in $6m in referrals by identifying promoters and reaching out to them for ask for referrals—they’re one of your quickest routes to revenue growth from your NPS program.
3. What’s your response rate?
As one of our clients puts it so eloquently, “what is worse than apathy? Indifference.” Their research determined that those who did not respond to their surveys were actually much more likely to be detractors.
This is why response rate matters so much. If your customers are not giving you feedback, that is a red flag that must be addressed.
If your non-responders are not filling out email surveys, try text messages. If they visit your website often, hit them with a popup. Try and engage your customers on the channels they use the most.
Do not force them down what is easiest for you to manage. Nothing bad will ever come from getting a 100% response rate. Consider this a call to action for all B2B companies to get serious about maximizing response rates if they want to minimize churn.
Pre-NPS Analysis Setup: The Data You Should Have
Before we jump into NPS analysis techniques, we’ve made some assumptions about the data you’ve collected.
Data point #1: The Net Promoter Score
Every recipient should have answered the question “how likely are you to recommend [Brand name] to [Relevant people]?”
If you’re in the B2B space, we recommend collecting NPS scores across multiple level of your customer’s organisation. Front-line staff all the way up to the C-Suite have influence over buying decisions, so all their feedback is important to tackling customer churn. (More advice multi-stakeholder surveying on page 31).
As you’ll see later, it’s important that for each score you know which customer company they’re from and how much annual revenue they offer.
Data point #2: The NPS drivers
Every Net Promoter survey should have follow up questions to help you understand what’s driving the score.
Allowing customers to select drivers lets you analyze the overall impact of that driver on the NPS score, so you can quickly identify the impact of improving it.
Learn more about how and what to ask in your NPS drivers question in our award-winning course: Listening to feedback, Platinum (use the code NETPROMOTER25 to get 25% off today!)
Data point #3: Verbatim comments
The magic happens in the free-text field. Always leave room for a comment on your NPS survey—it may provide the extra information you didn’t know you needed.
Free-text comment fields make analysis hard, especially at-scale (i.e. in the B2C space). But, we’ll cover the techniques and methods below that help you make sense of them.
How to analyze your NPS survey results by data type (3 methods)
There are three datasets typically collected during the NPS survey process. Let's breakdown how to analyze each one.
How to analyze your NPS score itself
You already know the basics (i.e. detractor = bad), so we’ll skip over them and take a look some novel ways of analyzing the NPS score itself.
1. Is your NPS score good?
The very first step of analyzing your NPS results is to calculate your NPS score. The calculation is simple (% promoters - % detractors).
But, answering the question ‘is this a good nps score?’ is a little harder.
Luckily we’ve answered this question plenty of times before and have 3 approaches you can use to help everyone at your company understand the score.
Absolute NPS benchmarking: This is a simplistic approach that says a score <0 is bad, a score of 0-30 is acceptable, a score 30-50 is pretty good, and a score of >50 is exceptional. NPS changes by industry so this method is quite inaccurate.
Relative NPS benchmarking: Relative NPS leverages competitor benchmarks so you're one step closer to comparing oranges to oranges.
Being your own benchmark: The most accurate benchmark of improvement is your score last year. The Net Promoter Score by itself is largely meaningless, unless you work on improving it. Assuming you survey customers every 6 months, then a good NPS is the one that is higher than the score you received during your last survey campaign.
We cover these (and more) in depth in our article, “what is a good NPS score?”.
2. Segment your survey responses by account, then by lifetime value
Assuming you’ve surveyed multiple stakeholders in each of your customer accounts, collect them together to provide one view of that account’s health.
As show in the CustomerGauge platform (image below), we integrate NPS feedback with revenue data so you can understand how well (or how poorly) your top customers are doing.
For example, MoonFaith Banks has a really bad NPS score of -50, only three contacts were surveyed and only two responded. This is a pretty good indicator this account is at-risk of churning—a BIG problem considering they’re generating the most revenue for the business.
In the B2C space, you can do this too by segmenting your customers by lifetime value. If you know that, in general, customers who live in London spend more and more frequently, then you can cluster them together and identify their group NPS score.
Another revenue-based NPS score analysis we offer our customers is a SWOT analysis. Based on both revenue and NPS score, the SWOT below plots which customers are a strength, weakness, opportunity or threat.
3. Plot a trend
Simple, but effective. Plot your overall or individual NPS scores on a graph to see whether they’re improving or declining. This should give you a pretty good understanding of your NPS program’s progress.
Related read: 13 ways to visualize NPS
How to do an NPS driver analysis (step-by-step with template)
So you’ve collected your NPS scores and each recipient told you which drivers contributed to the score they gave. Now, what?
It’s time for an NPS driver analysis to tell you the contribution of each driver to your overall score.
Take a quick watch of this video in our Analyzing Results Platinum course for an overview of what we’re about to guide you through.
Before we start, it’s important to know that NPS is calculated by using a simple average formula like in the below image.
In this example, we look at the customer experience of a holiday letting company. The average NPS score from all our recipients is 30 (% promoters - % detractors).
Each recipient left a score and they checked which drivers led them to give that score. We can use this information to calculate how much individual drivers contribute to the overall NPS.
Here’s where we’re going to enter step-by-step mode for an NPS driver analysis, so download the template here to follow along.
Step 1: Download the NPS driver analysis template
To get started with this step-by-step guide to calculating NPS, download the template here and follow along.👇
Step 2: Upload your data into the analysis template
As part of this step you should fill out your drivers and the corresponding collected data in the Excel spreadsheet. Check out Tab 1 for our example dataset, and I’ll walk you through it below.
In our holiday letting company example, we asked our 10 recipients to select which drivers contributed to their score. We gave them four options: holiday arrangements, the flight, the hotel, and the guided tours and asked them to tick the box if it contributed.
In our spreadsheet, we converted a tick to a ‘yes’ for ease of understanding.
Once your data is added in Tab 3 to the table on the left, our analysis template should automatically populate the table on the right. In doing so, it breaks down the contribution of each driver to the overall NPS score.
It’s important to know what’s going on here, so let's review.
Because NPS is a simple calculation of the % of promoters - % of detractors, and we ignore passives in the calculation, each respondent either counts as -100 or +100 to the average NPS score calculation.
A detractor contributes -100 (their specific score doesn’t matter, if they gave a score of less than 7 they count as a detractor and add equal weight to the question, ‘what percentage of our recipients are detractors?’). Whereas, a promoter contributes +100 for the same reasons.
Now, let’s zoom in on one specific recipient: Bob. Bob gave a score of 5, making him a detractor. He gave three reasons: flight, hotel, and guided tours.
In this NPS driver analysis, each driver has equal weighting and contributes a third of Bob’s overall score (we do this for simplicity, you could ask your customer to rank each but driver it often confuses them and extends the survey creating fatigue). Since Bob is a detractor, each driver contributes -100/3 = -33.3 towards his score.
By doing this exercise for each driver and customer (we automated that for you in the spreadsheet), we can then take the average of each driver to understand how each one contributes to the overall NPS score. You'll find that summarized in the final row.
Step 3: Interpret Your Score
The final row of our driver analysis template shows the contribution of each driver to NPS. We can now clearly see that column one (holiday arrangement) has the largest positive influence, whereas column three (their hotel) has the most negative influence on NPS.
Final note: It’s likely you have more than 10 NPS responses. You can extend your table by selecting the bottom row, right-clicking and choosing ‘insert one above’. You can do this many times to adjust the table to your needs.
To ensure that the table on the right continues to calculate NPS for you, simply select the bottom row and click and drag the small blue square down over the new rows that you inserted.
How To Do An NPS Verbatim Analysis (at-scale)
As we recommend to all our customers, every NPS survey should contain a free-text field that allows your customers to elaborate should they wish to.
This does complicate things. If you have high volumes of surveys you’re likely to have far too many responses to go through by hand, and even if you could it’s hard to be objective in your approach to extracting common themes. This is where text analytics comes in.
The term text analytics describes a set of linguistic, statistical and machine learning techniques that model and structure the comments from surveys. The term is also called text mining referring to the idea of exploration of business information within unstructured text.
A common method in NPS text analytics is tagging—or categorization—of comments, which assigns keywords or drivers to comments. Tagging comments allows companies to report on driver frequencies and split frequencies on promoters, passives and detractors.
In an ideal world, you’d want to run a topic and sentiment analysis (two parts of text analytics) on all your NPS verbatim text responses. Utilising these techniques, you’re able to extract the topic the customer mentions at-scale and then understand their feeling towards that topic at-scale.
Here’s an example of a topic and sentiment analysis done by CustomerGauge’s text analytics application. (Read more about top NPS softwares like ours here)
For all our customer’s NPS responses, this widget automatically reads the comment left by the customer and notes what comments they discussed and the sentiment in which they discussed it.
Watch our video module on voice of customer text analytics to get a strong grounding in this NPS analysis technique.
While we don’t recommend that you do this manually, you can certainly test the waters on a handful of NPS responses. Practice creating a tagging taxonomy and analyzing some of verbatim responses to see what common themes you find.
How To Do A Net Promoter Score Root Cause Analysis—Dig Deeper
Step 1, watch our introduction to root cause analysis (this video one clip from the full course: Analyzing Results, if you'd like to do the full course, use NETPROMOTER25 for a 25% discount).
Even with good driver analysis, you may still need to do root cause analysis. You typically do this by interviewing a group of customers about why they marked a certain driver.
One of our customers, an IT company, found that the quality of a new feature created detraction among clients. At first, the company was surprised, as they thought they had set expectations by requesting clients to test it. Later, they were even more surprised upon learning that some customers hardly found any bugs during their test.
Performing root cause analysis, they found that the real issues were that clients hadn’t assigned sufficient time to testing and following up on bugs. Consequently, they changed their delivery model by offering onsite test and incident management services
NPS Root Cause Analysis Technique: The 5 Whys
The 5 Whys is a deduction technique developed by Toyota and used to explore the underlying root cause of a problem.
When you need to understand the root cause(s) of an NPS driver, select a subset of respondents and interview them using this technique.
The technique repeats the question “Why?” a number of times with each answer forming the basis of the next question. The reason why it’s called the 5 Whys is that it often takes 5 questions to get to the root cause.
In reality, you can ask other questions than “Why?” as long as you use open-ended questions.
5 Whys Example:
Problem: The Jefferson Memorial monument in Washington D.C. is deteriorating.
1st Why: Why is the monument deteriorating?
Answer: Because water and chemicals are used to clean it.
2nd Why: Why are water and chemicals being used to clean the monument?
Answer: There is a lot of bird droppings on it.
3rd Why: Why are there bird droppings on the monument?
Answer: Because birds feed on the many spiders on the monument.
4th Why: Why are there many spiders?
Answer: Because vast swarms of insects are drawn to the monument at dusk.
5th Why: Why are insects drawn to the monument at dusk?
Answer: Because the lighting on the monument attracts insects at dusk.
Solution: Delay the lighting by an hour to prevent attracting insects.
While talking about spiders and bird poop is not the most attractive example, it provides a clear scenario of how the ‘5 Whys’ can be used to identify the root cause of a problem.
So there you have it. In this article, we covered several NPS analysis techniques, including:
How to analyze your NPS score itself
How to conduct an NPS driver analysis
Why text analytics is so helpful for breaking down free-text
How to do a root cause analysis with your customers