How should we take to know customer satisfaction as accurately as possible


Users satisfied with the product and service may bring a friend with "Why do not you try using it because it was good?" This can be expressed as being able to support the growth of the entire enterprise by increasing customer satisfaction. So how do you measure how satisfied users are with products and services? Services where experts search for the best flight from the planning of the tripFlightfoxTodd Sullivan, co-founder of the company, has published the method they have taken on a blog.

How We Measure Customer Happiness

◆ What is the net promoter score (NPS)?
Todd used to measure customer satisfaction "Net promoter score (NPS)". Asking "Do you have a possibility to recommend products / services to others?" Instead of asking straight as "Please evaluate the satisfaction of products / services with a maximum of 10 points" It is an indicator to be obtained.

In the case of a method of evaluating products / services with a maximum of 10 points, it is roughly classified as "I do not like" if it is 3 points or less, "Neither is 4" or "6 points", and "7 likes" can do. In the evaluation of a restaurant, if 10 out of 100 people got 9 points or more, 70 people got 7 to 8 points, 20 people got 0 to 6 points, "More than 80% of customers evaluate 7 points or more "It seems that it was popular.

However, in NPS, those who won 6 or fewer points are "Detractor", 7 to 8 points are "Passive", 9 points are "Promoter" "We calculate the score by subtracting the number of losers from the number of promoters. In the case of this restaurant, there are 10 promoters and 20 sluts, so NPS is "10/100" minus "20/100" and "-10" (score is not displayed because it is a score).

The calculation formula looks something like this. In order to make NPS "50", when there are 100 customers, promoters - slanders must exceed 50.

◆ What is important to increase NPS is "All"
When evaluating products and services, there are roughly three evaluation axes, Todd said.

The first one is "functional aspect". In the restaurant, the type of menu was enough, whether the price setting was appropriate, whether the waiting time until the dish came out was not too long.

The second one is "emotional aspect". The menu that was offered was delicious, or the attitude of the clerk was not bad.

The third is "Behavior bias (Behavior bias)". This was slightly different from the above two points "Were you happy or sad when you went to a shop?" "Because you are yourself an employer, there was a part that you sympathize with the shop's management policy." " It is biased to exist for each person to evaluate, such as "biased higher whenever you attach points / strictly lower".

There are also such things, some people criticize for NPS "it is a figure that is dependent on environmental factors and there are flaws," Todd said, "As customers are affected by their emotions We should accept the customer's satisfaction in the same way, "he said, accepting environmental factors as one factor.

◆ Eliminate sample error
In order to increase NPS, it is important to "increase the number of promoters" and "reduce the number of losers". Therefore, it is possible to deliberately raise NPS by conducting surveys on satisfied customers only. But what I should not forget here is not to "raise NPS" but to "figure out how customers actually feel".

Mr. Todd, Flightfox, I temporarily distorted NPS to the negative side unconsciously. In that method, we used a questionnaire only for people who used it without canceling it, but it was a simple thing to take a questionnaire also for those who canceled and withdrew. Of those who were satisfied with the service, 20% answered the questionnaire, while nearly 100% answered (inevitably the grading is biased to a lower degree) for cancelers, so NPS is low as a result That is why.

Currently, it seems that it is made possible to obtain figures with less error as much as possible by receiving a reply at the time of the conclusion of the talk, including those who eventually canceled the service.

◆ Fear of retaliation
There are various opportunities to "rate someone" such as Amazon's customer review, App Store · App review on Google Play, Evaluation of the highest bidder at the net auction, etc. However, the side who is grading is also done completely anonymously There are parts that are not. That way you come out is "afraid of retaliation".

"Retribution Evaluation for People with Low Scores" was often seen early in eBay, the leading online auction giant. In other words, if you lower your evaluation of someone, you will get low ratings from your partner, so you only have to score high points for you to be highly evaluated.

Flightfox also committed this mistake when ceasing crowdsourcing, and NPS jumped from 30 to 90 in the past. Todd, who felt that "90 is NPS at Apple but 90 is truly funny", investigated the cause and noticed that Flightfox's rating form displayed the expert's name and face picture. In other words, the customer did not evaluate Flightfox 's service, but evaluated the experts who did the other party.

Actually, as Todd and others checked with the customer, it turned out that they felt "concerned about retaliation" by evaluating someone in concrete terms, so from the evaluation form the expert's name And the photograph was to be removed. However, since NPS still maintained a high numerical value of 70, "Then, how about evaluating them both?", This time we evaluate expert individuals and Flightfox separately Change to a shape to receive.

As a result, the expert NPS was 75 and the Flightfox NPS was 26. It took a couple of months to settle in this form, it took more time to get enough samples, but after that, NPS seems to be stabilized with a slightly higher number.

◆ "Tip or Tweet (tip or tweet)"
Regardless of how much evaluation form is built, there is a person who gives priority to doing the next page without thinking deeply and can not do anything to put a rating point for the lie in the first place in the first place. So, in order to know that "I want to recommend to someone = customer satisfied? Does it lead to real numbers of referrals? Does it lead to growth?" Todd and others said more than 7 points in the evaluation form When it is turned on, share (SNS sharing) button is displayed immediately below.

It succeeded and about 10% of people clicked the share button. In addition, more people than that were sharing on Twitter and Facebook.

At the same time, we also added the ability to pay chips using PayPal. It seems that this function truly thought that even inside people "2 ~ 5 per month" Todd, who is unfamiliar with chip culture, thought "It is not a month" Lead developer Oscar is supposed to be "30 monthly". For that reason, we created a page called "Tip or Tweet (Chip or Tweet)" and, as with the Share button, we made it possible to choose whether to pay a tip or tweet when we add more than 7 points.

Chips were set at least 10 dollars (about 1000 yen), but 20% chose chips out of customers who gave 7 points or more. The average was $ 19 (about 1900 yen). In other words, "customers (promoters) who support high NPS were satisfied with the service to think that they could pay the chip."

◆ Conclusion obtained from the case of Flightfox
As a result of the above trial and error, Todd got the following three conclusions about NPS.

1: To be evaluated by all customers (including canceled persons)
2: Evaluation page should be universal / deceptive so as not to feel "retaliation"
3: To add "Tips or Tweet" function to confirm their NPS is accurate

Flightfox 's NPS transition released by Mr. Todd is Kore. Certainly, there are numbers above and below when changing functions, but the most recent figures are quite stable.

in Note, Posted by logc_nt