Messages les plus consultés

mardi 7 juin 2011

Social networks and Opinion Leaders : two keys for adoption of new products

A recent article published in Marketing Science (2011 March-april), by Iyengar, Van den Bulte et Valente gives us some interesting highlights about opinion leadership and network.

Main research questions

1- Does word-of-mouth in social networks really affect how quickly members (hereafter physicians) adopt a new product (ie. a new drug against viral infection) ?
 2- Do actual opinion leaders exist and can one identify them ?
3- Where are good seedings points for a viral marketing campaign ?

Two measures of opinion leadership

1- Self-assessment opinion leadership scale (adapted from Childers, 1986, Journal of Marketing Research) : six questions for self-reported opinion (on a semantic differential scale)

Q1: In general, do you talk to other doctors about (viral infection) : 
Never — — — — — —— Very Often
Q2 : When you talk to your colleagues about (viral infection), do you : 
Offer very little information— — — — — ——Offer a great deal of information
Q3: During the past 6 months, how many physicians have you instructed about ways to treat (viral infection) : Instructed no one — — — — — ——Instructed multiple physicians
Q4 : Compared to your circle of colleagues, how likely are you to be asked about ways to treat (viral infection) ? Not at all likely to be asked— — — — — ——Very likely to be asked
Q5 : In discussions of (viral infection), which of the following happens more often?
Your colleagues tell you about treatments— — — — — ——You tell your colleagues about treatments);
Q6 In general, when you think about your professional interactions with colleagues, are you
Not used as a source of advice— — — — — ——Often used as asource of advice

2- Sociometric leadership (centrality network = indegree)
How many times a physician have been named by his or her colleagues as "someone with whom they feel confortable discussing" about treatment, or "someone to whom they typically refer patients"


Physicians from three large US cities: San Francisco (SF), Los Angeles (LA) and Los Angeles  (NYC). Response rate varie from  San Francisco (SF) (44.5 % = 150 respondents) Los Angeles (28.9 % : 197 respondents) Los Angeles (24.3 % : 284 respondents).

Within sociometric condition, the 67 respondents in SF generated 37 unique nominees for discussion and 24 unique nominees for referral. In LA, the 57 respondents generated 38 and 24 unique nominees, and in NYC the 69 respondents generated 43 and 22 unique nominees. 

For self assessment measures, the average score was closed to 26 points (mean = 4,5)

Main Results

- Correlation between the two leadership measures are significant (p < .001), but rather average : .45 for SF sample, .33 and .41 for or LA et NY samples.

- Correlation between early adoption and each measure of opinion leadership were stronger for sociometric measure (25%) than for self-evaluation leadership measure (11%).

- Evidence of contagion was found over network ties, even after controlling for marketing effort and arbitrary system-wide changes. Moreover, adoption is affected by peers’ usage volume, rather than by whether peers have adopted or are prescribing. 

 - Concerning the opinion leadership measures, authors find that self-reported leadership and sociometric leadership are distinct characteristics. More precisely "(i) they are weakly correlated, (ii) the tendency to adopt early is more pronounced for sociometric than for self-reported leaders, and (iii) self-reported opinion leaders are less responsive than others to their contacts’ behavior, whereas sociometric opinion leaders are not differentially responsive. "

-  Self-assessment measure of leaderhip does not equals sociometric leadership. Sociometric leadership was associated with early adoption even after controlling for contagion, and sociometric leaders were equally sensitive to contagion as non-leaders. Sociometric opinion leaders are also opinion followers, but self-assessed opinions leaders are not. Why ? These two measures probably tap some different underlying concepts : sociometric would be closed to the original leadership essences, self-assessment would rather measure self-confidence than true influence.

Two main marketing advantages for targetting opinion leadership:

1 : the “stand-alone” customer lifetime value (CLV) of opinion leaders will be higher than of other people because they tend to be early adopters and heavy users. 
 2 : their “network” value may be higher : they reach more people and also, because there are early adopters and heavy users, they influence others sooner and more effectively than less connected people.


1- Network ties affect the adoption of a new product
2- Focusing on opinion leader is worth of value
3- Sociometric would be a more valid measure of leadership

Read more

Iyengar R, Van del Bulte C and Valente W, How social networks and opinion leaders affect the adoption of new products, Marketing Science, 30, March, April, 195-212

Aucun commentaire:

Enregistrer un commentaire