THE INFLUENCE OF THE NET PROMOTER SCORE (NPS) ON ONLINE WORD-OFMOUTH BEHAVIOR Hans Haans

NPS
Net Promoter Score introduced in 2003 by Reichheld
One question on a scale from 0 – 10 to measure loyalty and the number one question you need to grow

“How likely is it that you would recommend our company to a friend or colleague?”
Score 0 – 6 Detractors
Score 7 – 8 Passives
Score 9 – 10 Promoters

2

Practice vs. Academics
Embraced by many companies such as Philips, Google and Apple as the corporate metric due to its simplicity (Forbes 2011; Gupta and Zeithaml 2006; Reichheld and Markey 2011)


Academics question the quality of the NPS (e.g., Keiningham et al. 2008; Morgan and Rego 2006; Van Doorn, Leeflang, and Tijs 2013)

 Is NPS accurately measuring growth predictions  NPS is an attitudinal measure of intention to recommend rather than actual behavior (Sheeran 2002 – 28%)  Data loss through use of three segments

3

Electronic Word of Mouth (eWOM)
NPS measures intention to spread word of mouth
WOM is important in consumer purchase decisions and as a result many companies have deflected traditional marketing approaches in favor of WOM (East, Hammond, and Lomax 2008; Keiningham et al. 2008; Trusov, Bucklin, and Pauwels 2009)

    

low costs interactivity speed lack of commercial bias higher sense of credibility

4

Electronic Word of Mouth (eWOM)
“Word-of-mouth marketing is a particularly prominent feature on the Internet” (Trusov, Bucklin, and Pauwels 2009)
Online word of mouth (eWOM) has some advantages over traditional WOM (Chen and Xie 2008; Dellarocas 2003)  fast and convenient  available for an indefinite period of time  can reach far beyond the local community


This makes companies more dependent than ever on cultivating positive WOM and getting rid of negative WOM (thesis proposals) 5

Research Objectives
The question arises if the NPS is the right tool to be used for this purpose.  Measurement Issue 1: Test whether the NPS (i.e. intention to recommend) actually represents true promotion behavior (i.e. online word of mouth) of customers.  Measurement Issue 2: Test whether the segmentation into promoters, passively satisfied customers, and detractors is justified.

6

Sentiment Analysis
The main goal of a sentiment analysis is to determine the polarity of a written text (Jansen et al 2009; Liu 2010)
Is there a relationship between NPS scores and online sentiment

7

Sentiment Analysis
Specifically, is there a relation between customers’ individual NPS scores and their online WOM.
Manually determined the sentiment of the customers’ social media messages. In particular, two coders content-analyzed the messages to identify whether the eWOM was positive, negative, or neutral  Cohen’s Kappa to judge inter-judge reliability, was sufficiently large (Neuendorf 2002)  94% for company 1  91% for company 2 8

Descriptives
Average NPS for customers with a positive, neutral, and negative sentiment. All differences are significant at p < .01

Sentiment Positive Neutral Negative

Total Sample 8.43 6.73 4.64

Company 1 8.44 6.98 4.54

Company 2 8.40 6.47 4.72

9

Model Estimation
Ordered logit model is used because eWOM is measured with an ordinal scale (positive, neutral, negative)
eWOM is modeled as a function of the individual customers’ NPS (one month)     

The main model is significant (χ2(1) = 53.77, p < .01) Pseudo R2 value of .23 NPS has a significant positive effect on eWOM (β = .41, p < .01) The odds ratio is 1.50 Predictability of our model => 60% of the cases are correctly predicted 10

Robustness Checks
Company-Specific Model Estimation
Three-Month, Six-Month, and One-Year Time Frame


Automated Sentiment Analysis

Reichheld’s Classification
Detractors (0-6) do not differ from each other (p’s > .10), but differ from customers that give a higher NPS
Promoters (9-10) differ from the other groups of customers (p < .01) but are similar to each other (p = .29)
Passives do not appear to be homogenous group 100% 90% 80%

Percentage

70% 60% Positive

50%

Neutral

40%

Negative 30% 20% 10% 0% 0

1

2

3

4

5 NPS

6

7

8

9

10

Reichheld’s Classification
55 eWOM messages are negative (27% of all messages) and are provided for 70 percent by the detractors (NPS score of 0-6). Still 16% of the negative eWOM message come from customers with an NPS score of 7.


42 messages contain positive eWOM (20% of all messages. Especially the promoters (NPS 9-10), with 50 percent, are responsible for the positive eWOM. 41 percent of positive eWOM comes from customers with a NPS of 8.
With 60% most neutral eWOM messages are spread by the passives (NPS score 7-8).

Conclusion
NPS is a good proxy for customers’ actual eWOM behavior  What service components determine the NPS  firms should invest more in better service to increase their NPS and as a result eWOM.


Categorization Reichheld holds for the detractors and promoters, but the group of passives is scattered and ignored by Reichheld  especially the passives should get more attention.

14