The reaction of readers of my blog to the first post about NPS – a large number of views, live feedback, many private messages and phone calls from partners and colleagues – showed the importance of the topic under discussion. Everyone noted that the topic about NPS is interesting and relevant, many said that they were waiting for the continuation.
Only two questions
Feedback has shown that, when talking about NPS, it is necessary to clarify some of the nuances once again, to talk about errors in calculations, as well as how they can be avoided.
- Let’s agree right away and not overlook: the most important thing in working with NPS is your customers. Remember, NPS is not an index or a number, NPS is your customers. These are specific people with specific questions and problems who, for some reason, turned to you.
- And again about the methodology for calculating NPS. You only need to ask two questions to your clients. Once again: not five, not ten questions, and not even 33 questions, among which there are questions on NPS. Only two questions are needed. The first is formulated as “On a scale from 0 to 10, how likely are you to recommend our company to your friends, relatives and acquaintances? Where “0” is “I would never recommend”, and “10” is “definitely recommend.” The second question is: “Please explain your opinion.” And you need to ask it to each participant in the survey, regardless of what grade the client put on the first question.
Attention to individual groups
Let’s look at examples. So let’s say you had 1000 clients in a month. You need to call them and ask these two questions. In ideal conditions, you will make 1000 calls and get 1000 answers.
However, when there are a lot of clients, it will not be possible to call everyone. Then you need to form a random, representative and sufficient sample.
An important criterion here is the randomness of the sample . You can get a real NPS if you do not ignore certain categories of clients, for example, clients of a particular master or a certain department, or a client who purchases only one product / service. If we “throw” such clients out of the study, then we get the NPS of just a certain, separately taken group of people, and not our company. And as practice has shown, if you can easily separate one group of customers from another, then their level of satisfaction should be measured separately. You will never get a reliable result if you do not clearly track all categories of customers.
For example, such groups can be: individuals and corporate clients; car owners who bought a car at this dealership or drove along the road; new and repeat customers. It is best to measure the NPS separately for each category, and then calculate the average result taking into account the weight of each category.
Another nuance is a representative and sufficient sample . If you have ten clients, then interviewing five of them is not enough. But interrogating nine of these ten, according to mathematical laws, is also not enough, since the remaining one can ruin the whole picture. But if you have 1 million customers, then there is no need to interview 500 thousand of them either.
I will give you advice that is true from a practical point of view: if there are more than 1000 clients / visits / acts of purchase (and further increase in the number does not matter), then 200 successful interviews will be enough for a representative sample.
Soft or fair option
We have come to the most interesting point in our example. Ideally, you call 1000 customers and collect 1000 responses. Those who chose “9” and “10” in response to the first question become “supporters”, those who chose “7” and “8” become neutral clients, and those who answered “0-6” become “critics “. Let’s say you got 500 “9-10” responses, 400 “7-8” responses, and 100 “0-6” responses. In this case, the calculation will be as follows: 500/1000 = 50% “supporters”, 400/1000 = 40% neutral clients and 100/1000 = 10% “criticism”. And the NPS will be 50% -10% = 40%.
The above is the perfect picture. In reality, collecting 1000 responses from 1000 customers will not work. And what to do with those clients whom we could not get through, as well as those who did not give us answers? There are several options – soft and honest. In fact, there is also a third one, but it is very tough.
- The first option is soft: do not count such customers, but collect at least 200 customers who gave answers. This will help preserve the representativeness of the study, but with this approach we will get an overestimated NPS.
- The second option is honest: the category of “neutral customers” includes all those to whom we could not get through, as well as those from whom we could not collect answers.
And if we have 1000 clients, then, having gained 200 successful respondents, we get, for example, 100 “supporters”, 50 “neutral clients” and 50 “critics”. And in order to dial those 200 who answered, we had to make 300 calls. Then the calculation of NPS for the first, soft option will be as follows: (100-50) / 200 = 25% . And according to the second option: (100-50) / 300 = 16% . My choice is in favor of the second option, since it is more honest.
3. There is also a third, more rigid method of counting. In this case, all customers who could not be reached by phone are assigned a score of “0” and are considered in the category of “critics”. In my opinion, it is not always worth doing this. After all, it is clear that the inability to take part in the survey does not always mean that this respondent was dissatisfied. Of course, those who did not give answers should not be included in the category of “supporters”, but it is also wrong to class them as “critics”. My choice is in favor of an honest calculation option, according to which respondents who did not answer the question are counted in the “neutral customers” group.
And a few more important notes
The second question cannot be neglected. It must be asked to all survey participants. He will help you understand a lot in your work. And I will pay special attention to this issue, I will tell you how to process and analyze the received answers.
NPS for long term relationships. This is its main difference from other indices measuring satisfaction. There is no need to ask about the willingness to recommend the company after only a telephone conversation with the client or even immediately after the service has been provided. In these cases, the correct solution is to ask about satisfaction and difficulties encountered.
For example, if a client gives low NPS scores immediately after talking to a call center, this does not necessarily mean that there is a problem with that particular interaction. The call center may have done a great job and you got a 6 instead of a 2, but the client might still not like something else about the company. For example, he could contact a call center to solve a problem that arose on the site or in the course of personal communication.
Conversely, too fast or generally incorrect communication between the call center staff and the client can lead to lower NPS scores immediately after such communication, and will not reflect the client’s attitude towards the company in the long run. To accurately measure a long-term relationship, you need to give the client a few days for his emotions (both positive and negative) to subside. You will get smarter answers and therefore more reliable NPS results.
It seems to be all, but no. You must work out all the information received. Yes, you need to work with “critics”. Many people know about it and many do it (although not always). It is necessary to “extinguish” the negative, to transfer these clients at least to the category of neutral. And as practice shows, with good work, up to half of the “critics” can be transferred to the category of “supporters.” You also need to work out the “supporters”. If you give up on them, they say, “these are already happy, we did our job well,” then there is a risk of losing them the next time.
The most important thing in any research is to apply the findings, learn lessons and work with the results. By itself, NPS doesn’t mean anything, it’s just numbers on paper. Another thing is important, what this data motivates you to, as well as what you will do when you measure the NPS of your company.
What should be done, how to work out feedback? What is important to us: NPS itself or our clients? Do not be lazy, read point # 1 again. Right now. Yes, I know that you remember what is written there. But read it again to answer honestly to yourself why exactly you need to count NPS.