Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan

2011

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan

CIS Coursework

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan

Under the Guidance of Prof. Rajeev Kumra

Prepared By: Arnab Bhattacharya

(WMP-5012)

Course of Independent Study

CIS Coursework

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan

Abstract
Marketing has been one of the major applications of data mining since the field emerged. Typically, the decision of whether or not to market to a particular person is based solely on their characteristics (direct marketing), or those of the population segment to which they belong (mass marketing). This often leads to suboptimal marketing decisions by not taking into account the effect that members of a market have on each other's purchasing decisions. In many markets, customers are strongly influenced by the opinions of their peers. Viral marketing takes advantage of this to inexpensively promote a product by marketing primarily to those with the strongest influence in the market. This is more applicable in the current age scenario, with Word-of-Mouth marketing taking prevalence with the tremendous growth for Social Networking. Richardson and Domingos proposed a viral marketing model in their paper [5], which modeled the viral interactions across the social network in a graph. Though seminal in nature, the model has some inherent deficiencies with respect to direct practical application. The model assumes the availability of the various psychological parameters required by the model. However, very limited work has been done to estimate these psychological parameters, empirically or statistically, to increase the practical relevance of the theoretical models. In this paper, we try to estimate the value of , which measures self-reliance of an individual, in model developed by Richardson and Domingos[5]. We claim that an individuals VALS classification could be used as a suitable proxy for the individuals value. A survey is conducted to prove this claim and also to estimate the value for the various VALS classification. We further propose a model based on collaborative filtering to typically predict an individuals VALS classification based on the online usage and behavior using web-analytics of social-bookmarking sites.

CIS Coursework

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan

Acknowledgement
This research work was carried out as a part of Course of Independent Study that provided an opportunity to explore the specialized fields related to Viral Marketing and Consumer behavior in the Marketing functions. I would like to express my gratitude to my supervisor Prof. Rajeev Kumra for his guidance and support during this research work. This has been a great learning experience, both from the studies that have been carried out in the marketing area in the past and the applications of the literature in enhancing a model that helps understand the consumer behavior with respect to adoption of a new technology.

CIS Coursework

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan

Introduction
Marketing has been one of the major applications of data mining since the field emerged. Typically, the decision of whether or not to market to a particular person is based solely on their characteristics (direct marketing), or those of the population segment to which they belong (mass marketing). This often leads to suboptimal marketing decisions by not taking into account the effect that members of a market have on each other's purchasing decisions. In many markets, customers are strongly influenced by the opinions of their peers. Viral marketing takes advantage of this to inexpensively promote a product by marketing primarily to those with the strongest influence in the market. The use of relation- between people makes viral marketing potentially more profitable than direct marketing. In this paper, we discuss the problem of viral marketing in context of a new product launch through internet marketing through web. Many studies have been done on finding out the optimal set of customers to market, so that the viral effect has the maximum effect involving minimum cost. The seminal papers by Richardson and Domingos , [5] and [9], are the most important of them. These papers theoretically model the viral effect and arrive at a mathematical formulation for achieving the optimum marketing plan. However, these and other studies, look at the problem from a theoretical background without a thorough analysis at the applicability of the approach. One such theoretical parameter is the value , which stands for the selfreliance of consumers while making purchase decision. In the current study, we try to come up with an empirical approach for estimating , by using the VALS psychographic classification scheme. We argue that there is a strong correlation between the VALS psychographic segments and their associated values. We claim that VALS segment of a consumer could be used as an accurate proxy for the value of a consumer while identifying his network worth in estimating the optimum marketing plan. We have done a quantitative study, within the limited scope of the study, to substantiate the claims made above. Further to this, we have proposed a mechanism to automatically classify consumers through the collaborative filtering mechanism using web-analytics of social-bookmarking sites. In Section-2, the relevant existing literature on the topics has been discussed. In Section3, we define following the various existing literatures available on this topic. In Section-4, we explain the importance of the value of in identifying the optimum marketing plan in the viral marketing context. In Section-5, we explain the VALS classification scheme in detail. Section-6 explains the relevance between the VALS classification and the associated value for the segments. Section-7 describes the dataanalysis plan and the results observed. Section-8 gives the proposal for automatic classification of consumers through the collaborative filtering mechanism using webanalytics of social-bookmarking sites. Sec-9 is the conclusion with the direction for future works.

CIS Coursework

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan

Secondary Literature Analysis

A graph representation of a social network is often the basis for representing the spread of a new idea or product, with the nodes and edges in the graph representing the individuals and inter-individual interaction respectively. These interactions could include the recommendation of a particular product and such recommendation networks and their effects on consumer purchasing have been analyzed in [1] and [2]. Further, there has been recent statistical support that such network linkage can directly affect product adoption [3]. Based on these empirical studies, we can formulate assumptions on how people affect the people they interact with. We can then use these graphs to answer questions such as: If customers influence each other in their decisions to buy products, which customers should be targeted to maximize the expected profit of a new product? and How large of a consumer base needs to be targeted for a new technology, product, or idea to capture a significant share of the market? Motivated by the declining effectiveness of traditional mass marketing techniques [7], many recent papers have studied these and similar types of questions. The algorithmic problem of designing viral marketing strategies, marketing techniques, which exploit pre-existing social networks to reach consumers, was studied by Richardson and Domingos [5], and Kempe, Kleinberg and Tardos [4, 6]. Their research builds on a word-of-mouth approach examined in a marketing context by Goldenberg et al. in [8]. In the aforementioned works, the producer of a new product is assumed to have the ability to influence a particular set of consumers within the social network either through targeted advertising, providing free samples, or adding monetary incentive to adopt the new product. If these people influence some of their friends to also try the product, and these friends in turn recommend it to others, and so forth, the producer can create a cascade of recommendations. The question then becomes how to choose an initial subset of so-called early adopters to maximize the number of people that will eventually be reached, and hence be likely to purchase the product. The size of the subsets allowed is assumed to be limited due to marketing budget constraints. The model proposed by Richardson and Domingos [5], though seminal in nature, has some inherent deficiencies with respect to direct practical application. In one extent, this is because of the simplified scenarios in which the model works, and in the other extent, the various psychological parameters that the model assumes to be available. Various papers have tried to remove some of the constraints due to the first limitation. Notable among them is the study by Time Carnes et al.[13]. This paper tries to apply Richardson and Domingos[5] model in a competitive scenario where a rival product has also been launched. However, very limited work has been done to remove the second constraint, wherein the various psychological parameters need to be estimated, empirically or statistically, to increase the practical relevance of the theoretical models. In this paper, we try to estimate the value of , which measures self-reliance, in model developed by Richardson and Domingos[5]. In the next section, we try to introduce the concept of in the model through an abridged description of the model.

CIS Coursework

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan

Model: Definition of
Lets suppose there is a set of n potential customers. Xi 1 if customer i buys the product being marketed 0 otherwise Ni Neighbors of Xi directly influencing Xi ..(1) Mi 1 if customer i directly marketed ..(2) 0 otherwise M={Mi ..... Mn} be the marketing plan Then, for all Xi, the probability that Xi is 1 could be represented by P(Xi | X-{Xi},M) which could be described as the probability that Xi = 1 provided the state of the other customers represented by (X- Xi) and the marketing plan M. P(Xi | X-{Xi},M) = P(Xi | Ni, M) as Xi is only influenced by Ni = i* PO(Xi | Mi) + (1 i) * PN(Xi | Ni, M) (3) where, PO(Xi | Mi) Xis internal probability of purchasing the product i.e. the probability that Xi would be buying the product out of his own volition without the effect from his influencers PN(Xi | Ni, M) The effect that Xis neighbors have on Xi i scalar with 0 < i < 1 that measures self-reliance of Xi The importance of is that it amplifies or reduces the effect of the respective probabilities. The higher the value of more pronounced is the effect of PO. For many products, such as cellular telephones, multi-player computer games, and Internet chat programs, a customer's probability of purchasing depends strongly on whether his friends have also purchased the product. However, though PO and PN varies for different product categories, the value of would range across the individuals depending on his lifestyle and demographic based classification. This paper aims to analyze the effect of customer classification in determining the value of self-reliance i.e. for the customer. The value estimate of for neighbors is required to estimate the network worth of a customer. This statement is proved in the next section.

CIS Coursework

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan

Significance of
Continuing from equation 3, we define PN as follows PN(Xi | Ni, M) = wij * Xj .. (5) where wij represents how much customer i is influenced by customer j. when wij >= 0 and wij = 1. Combining 3 and 5, we get Pi(Xi | Ni, M) = i* PO(Xi | Mi) + (1 i) * wij * Xj For calculating the optimal marketing plan for a product that has not yet been introduced to the market, the state of the neighbors will not be known. To eliminate the effect of Ni in formula 5, a formula has been derived for computing PN(Xi | M) PN(Xi | M) = PN(Xi | N, M) * P(N | M) .. (7) Where the summation is done across all the configurations of Ni and N is one such configuration. Combining 6 and 7 we get P (Xi | M) =
i *

PO(Xi | M) * P(N | M) + (1 - i) * wij * P(N | M) .. (8)

Since the inner summation is over all possible values of N, 8 could be written as P (Xi | M) =
i *

PO(Xi | M) * P(N | M) + (1 - i) * wij * P(Xj | M) (9)

Note, because 9 expresses the probabilities P(Xi | M) as a function of themselves, it can be applied iteratively to find them, starting from a suitable initial assignment. The marketer's goal is to find the marketing plan that maximizes profit. Let c be the cost of marketing to a customer (assumed constant), r0 be the revenue from selling the product to the customer if no marketing action is performed, and r1 be the revenue if marketing is performed, Let.fi0(M) be the result of setting Mi to 1 and leaving the rest of M unchanged, and similarly for. fi1(M). The expected lift in profit from marketing to customer i in isolation (i.e ignoring his effect on other customers) could be represented by ELPi1(M) = r1 * P(Xi | fi1(M)) r0 * P(Xi | fi0(M)) c .(10)

CIS Coursework

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan We also refer to this as the customer's intrinsic value. Let M0 be the null vector (all zeros). The global lift in profit that results from a particular marketing plan M is ELP (M) =
n i=1

[r1 * P(Xi | fi1(M)) r0 * P(Xi | fi0(M)) c ] . (11)

If we expand P terms, we would be getting their respective in the formula. This proves that estimating effectively for each customer is essential and paramount for choosing the most optimum marketing plan for introducing a product.

CIS Coursework

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan

VALS Framework

VALS assigns individuals a VALS type on the basis of their responses to questions in theVALS Survey. VALS-typing populations of interest, such as customers or constituents, is the first step in a VALS approach to achieving strategic marketing and communication goals.

Primary Motivation: Ideals, Achievement, and Self-Expression

The concept of primary motivation explains consumer attitudes and anticipates behavior. VALS includes three primary motivations that matter for understanding consumer behavior: ideals, achievement, and self-expression. Consumers who are primarily motivated by ideals are guided by knowledge and principles. Consumers who are primarily motivated by achievement look for products and services that demonstrate success to their peers. Consumers who are primarily motivated by self-expression desire

CIS Coursework

10

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan social or physical activity, variety, and risk. These motivations provide the necessary basis for communication with the VALS types and for a variety of strategic applications.

Resources
A person's tendency to consume goods and services extends beyond age, income, and education. Energy, self-confidence, intellectualism, novelty seeking, innovativeness, impulsiveness, leadership, and vanity play a critical role. These psychological traits in conjunction with key demographics determine an individual's resources. Various levels of resources enhance or constrain a person's expression of his or her primary motivation.

VALS classifications: Innovators:

Innovators are successful, sophisticated, take-charge people with high self-esteem. Because they have such abundant resources, they exhibit all three primary motivations in varying degrees. They are change leaders and are the most receptive to new ideas and technologies. Innovators are very active consumers, and their purchases reflect cultivated tastes for upscale, niche products and services. Image is important to Innovators, not as evidence of status or power but as an expression of their taste, independence, and personality. Innovators are among the established and emerging leaders in business and government, yet they continue to seek challenges. Their lives are characterized by variety. Their possessions and recreation reflect a cultivated taste for the finer things in life.

Thinkers:
Thinkers are motivated by ideals. They are mature, satisfied, comfortable, and reflective people who value order, knowledge, and responsibility. They tend to be well educated and actively seek out information in the decision-making process. They are wellinformed about world and national events and are alert to opportunities to broaden their knowledge. Thinkers have a moderate respect for institutions of authority and social decorum but are open to consider new ideas. Although their incomes allow them many choices, Thinkers are conservative, practical consumers; they look for durability, functionality, and value in the products that they buy.

Believers:
Like Thinkers, Believers are motivated by ideals. They are conservative, conventional people with concrete beliefs based on traditional, established codes: family, religion, community, and the nation. Many Believers express moral codes that have deep roots and literal interpretation. They follow established routines, organized in large part around home, family, community, and social or religious organizations to which they belong. CIS Coursework 11

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan As consumers, Believers are predictable; they choose familiar products and established brands. They favor U.S. products and are generally loyal customers.

Achievers:
Motivated by the desire for achievement, Achievers have goal-oriented lifestyles and a deep commitment to career and family. Their social lives reflect this focus and are structured around family, their place of worship, and work. Achievers live conventional lives, are politically conservative, and respect authority and the status quo. They value consensus, predictability, and stability over risk, intimacy, and self-discovery. With many wants and needs, Achievers are active in the consumer marketplace. Image is important to Achievers; they favor established, prestige products and services that demonstrate success to their peers. Because of their busy lives, they are often interested in a variety of time-saving devices.

Strivers:
Strivers are trendy and fun loving. Because they are motivated by achievement, Strivers are concerned about the opinions and approval of others. Money defines success for Strivers, who don't have enough of it to meet their desires. They favor stylish products that emulate the purchases of people with greater material wealth. Many Strivers see themselves as having a job rather than a career, and a lack of skills and focus often prevents them from moving ahead. Strivers are active consumers because shopping is both a social activity and an opportunity to demonstrate to peers their ability to buy. As consumers, they are as impulsive as their financial circumstance will allow.

Experiencers:
Experiencers are motivated by self-expression. Young, enthusiastic, and impulsive consumers, Experiencers quickly become enthusiastic about new possibilities but are equally quick to cool. They seek variety and excitement, savoring the new, the offbeat, and the risky. Their energy finds an outlet in exercise, sports, outdoor recreation, and social activities. Experiencers are avid consumers and spend a comparatively high proportion of their income on fashion, entertainment, and socializing. Their purchases reflect the emphasis that they place on looking good and having "cool" stuff.

Makers:

CIS Coursework

12

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan Like Experiencers, Makers are motivated by self-expression. They express themselves and experience the world by working on itbuilding a house, raising children, fixing a car, or canning vegetablesand have enough skill and energy to carry out their projects successfully. Makers are practical people who have constructive skills and value selfsufficiency. They live within a traditional context of family, practical work, and physical recreation and have little interest in what lies outside that context. Makers are suspicious of new ideas and large institutions such as big business. They are respectful of government authority and organized labor but resentful of government intrusion on individual rights. They are unimpressed by material possessions other than those with a practical or functional purpose. Because they prefer value to luxury, they buy basic products.

Survivors:
Survivors live narrowly focused lives. Because they have few resources with which to cope, they often believe that the world is changing too quickly. They are comfortable with the familiar and are primarily concerned with safety and security. Because they must focus on meeting needs rather than fulfilling desires, Survivors do not show a strong primary motivation. Survivors are cautious consumers. They represent a very modest market for most products and services. They are loyal to favorite brands, especially if they can purchase them at a discount.

CIS Coursework

13

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan

Application of VALS classification

VALS provides a systematic adults classification into eight distinct consumer segments. VALS is based on enduring psycho-logical characteristics that correlate with purchase patterns. Respondents are classified according to their primary motivation, which serves as one of VALs two dimensions. By the theory of consumer behavior, motives are critical determinants of behavior. Motives have strong linkages to personality and self-concept. A core premise behind VALS is that an individuals primary motivation determines what in particular about the self or the world is meaningful core that governs hir or her activities. The three primary motivations are: Ideals motivation: These consumers are guided in their choices by their beliefs and principles rather than by feeling or desire for social approval. They purchase functionality and reliability. Achievement motivation: These consumers strive for clear social position and are strongly influenced by the actions, approval and opinion of others. They purchase social symbols Self-expression motivation: These action-oriented consumers strive to express their individuality through their choices. They purchase experiences. These three orientations determine the types pf goals and behaviors that individuals will pursue. The second dimesion, termed resources, reflects the ability of individuals to pursue their dominant self-orientation. It refers to the full range of psychological, physical, demographic and material means on which consumers can draw. Resources generally increase from adolescence through middle age and they remain relatively stable until they begin to decline with their old age. Resources are an important pf VALS since they can aid or inhibit a consumers ability to act on his or her motivation.

Relevance of VALS to
Now, with the whole details of the VALS classification in light, we try to see how VALS classification is relevant to . The importance of is that it amplifies or reduces the effect of the respective probabilities as defined in Equation-8 in Section-1. The higher the value of more pronounced is the effect of PO. For many products, such as cellular telephones, multiplayer computer games, and Internet chat programs, a customer's probability of purchasing depends strongly on whether his friends have also purchased the product. However, though PO and PN varies for different product categories, the value of would

CIS Coursework

14

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan range across the individuals depending on his lifestyle and demographic based classification. This paper aims to analyze the effect of customer classification in determining the value of self-reliance i.e. for the customer. A study of secondary data and literature available on the VALS segmentation indicates that the consumers having primary motivation of Self-Expression would have the maximum self-reliance i.e a maximum value for . These consumers are spontaneous and their choices have significant emotional impact. They display authority in their decision-making, looks for experiences associated with adventure, excitement and novelty. They are driven by What do I feel like doing now? emotions. This clearly indicates a relatively higher value of . The next set of customers in a decreasing sequence of value would be those motivated by Ideals. They are basically information seeking and purchase functionality and reliability. Hence, they would be somewhat affected by the customers in their influence network. They also resist impulse and are driven by What should I do? emotions. However, they also make choices based on principles and pursue selfdevelopment. Hence, they would be having lesser value of than those driven by SelfExpression. The consumers having primary motivation of Achievement would be having the least value of . These people strive to reduce risk and look for social approval. They are goal oriented and make choices to enhance their positions. The basically purchase success symbols and are driven by What are others like me doing?. Hence, it is apparent that these consumers would be having the least self-reliance and are expected to have the minimum value of . With this background, we have gone for a quantitative research based on the ANOVA technique. The next section describes the data-analysis plan and the results obtained.

CIS Coursework

15

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan

Data Analysis and Results

To find the correlation between and the VALS classifier, the survey questionnaire consisted of two parts Questions to derive the VALS classification Question to estimate the for the respondent For finding the VALS classification, the Survey questions used in VALS survey by Strategic Business Insights group (http://www.strategicbusinessinsights.com/vals/surveynew.shtml) was used. The questions were used as it is in this survey. The respondent response was then fed back to the survey to determine its VALS classification. The Question number 1-34, and 39-41 in the survey in Appendix, is for indentifying the VALS classification of the respondent. For finding the value, three dimensions of external psychographic elements were used viz. family, friends and media. Question-35 is the screening question for Internet usage, while Question 36-38 are for the rating for family, friend and media influence in respondent decision making. We have given equal weights to the three elements. So for each respondent, the estimated value is the mean of the responses retrieved from the three dimensions. Sample Size and Description: The time constraint allowed me to consider the sample only within the fellow participants of my current coursework. Though the sample is not an accurate representation of the Internet population, the significant diversity in demographics and in certain extent psychographics has made the sample a effective one, if not an optimum one. The diversity present in the current sample includes: Age distribution: 23 years 45 years Annual Income Distribution: 6 lacs 40 lacs Representation from almost all possible religious and social sects. Initially, 58 people responded to the survey, out of which 6 had to be discarded due to the incompleteness of the responses. So, finally 52 responses were acquired for final data analysis. Though the sample size is not the optimum one, it is suitable enough to go ahead with the analysis. Hypothesis: The basic hypothesis for this study is that the estimates vary with the VALS classification. So, we can derive the NULL hypothesis as

CIS Coursework

16

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan HO: estimate does not vary with VALS classification We try to reject HO by the analysis and try to estimate the for each classifier. Data Analysis: The classification of the respondents arrived at by using the respondents data is as follows:
INNOVATORS THINKERS ACHIEVERS EXPERIENCERS BELIEVERS STRIVERS MAKERS TOTAL 9 7 9 6 7 9 5 52

Note, that there was no Survivors classification. This is expected because the target segment of the study would not typically be someone with very low resources. As noted previously, the estimate for each respondent is arrived at by separately considering influence from family, friends and media and then arriving at an aggregate value by assigning equal weights to all the 3 parameters. The values arrived at for each respondent along with their VALS classification for each respondent are shown below.

CIS Coursework

17

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan

INNOVATORS ACHIEVERS INNOVATORS STRIVERS ACHIEVERS EXPERIENCERS THINKERS THINKERS ACHIEVERS ACHIEVERS BELIEVERS INNOVATORS EXPERIENCERS EXPERIENCERS THINKERS MAKERS STRIVERS BELIEVERS BELIEVERS ACHIEVERS STRIVERS THINKERS THINKERS BELIEVERS INNOVATORS BELIEVERS

3.663 2.380 4.581 1.242 2.574 3.486 3.469 3.567 2.372 2.570 2.763 5.887 3.850 3.573 3.353 2.253 1.629 3.101 2.989 2.460 1.597 3.826 3.803 2.810 3.965 2.460

THINKERS ACHIEVERS ACHIEVERS INNOVATORS MAKERS INNOVATORS ACHIEVERS INNOVATORS STRIVERS STRIVERS ACHIEVERS STRIVERS STRIVERS MAKERS STRIVERS EXPERIENCERS MAKERS BELIEVERS BELIEVERS INNOVATORS STRIVERS INNOVATORS EXPERIENCERS THINKERS MAKERS EXPERIENCERS

3.315 2.546 2.312 4.646 2.385 4.206 2.493 4.663 1.423 1.213 2.174 1.537 1.528 2.184 1.532 3.469 1.962 2.845 2.726 4.261 1.629 4.150 3.393 3.710 1.883 3.510

Note, higher value corresponds to higher self-reliance. Each value is aggregate of 3 user ratings on a 5 point scale based on family, friend and media influence, where 1 stands for maximum influence 5 stands for minimum influence. The VALS classification was converted to a nominal scale 0-6 as follows.
INNOVATORS THINKERS ACHIEVERS EXPERIENCERS BELIEVERS STRIVERS MAKERS 0 1 2 3 4 5 6

ANOVA was run in SPSS, with VALS id as the independent factor, while the estimate values as the dependent metric variable. Following are the results obtained

CIS Coursework

18

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan The VALS classification was converted to a nominal scale 0-6 as follows.

Descriptives

Estimate 95% Confidence Interval for Mean N Mean 9 7 9 6 7 9 5 52 4.32657 3.75765 2.44238 3.40835 2.75829 1.41339 2.15727 2.89402 Std. Deviation .602089 .213420 .132708 .170747 .228881 .175269 .184527 1.027497 Std. Error .200696 .080665 .044236 .069707 .086509 .058423 .082523 .142488 Lower Bound 3.86376 3.56027 2.34037 3.22916 2.54661 1.27867 1.92815 2.60796 Upper Bound 4.78937 3.95503 2.54439 3.58754 2.96997 1.54811 2.38639 3.18008 Minimum 3.680 3.524 2.278 3.229 2.545 1.178 1.947 1.178 Maximum 5.335 4.031 2.646 3.643 3.178 1.757 2.444 5.335

0 1 2 3 4 5 6 Total

Note, that the means are different across the classifications. Moreover, it could be noted that INNOVATOR (Id 0) has the highest mean as expected, because they have the highest resources and more driven by self-beliefs. Similarly, Maker (Id-6) has the lowest mean. Following is the significance report for the ANOVA.

ANOVA

Estimate Sum of Squares df 6 45 51 Mean Square 8.281 .092 F 89.659 Sig. .000

Between Groups Within Groups Total

49.687 4.156 53.843

The results show that the mean value difference across the groups is significant at 5% level of significance. This allows us to reject HO, and allow us to infer that the VALS classification is indeed a good proxy for estimating values.

CIS Coursework

19

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan Further we normalize the mean values calculated to arrive at a normalized value to be used to find network worth of users to find the most optimum marketing plan. Since, only the relative strength of matters across the classifications, we divide all the means by 5 (which is the least integer greater than all the means), to arrive at a normalized value from all classifications.
INNOVATORS THINKERS ACHIEVERS EXPERIENCERS BELIEVERS STRIVERS MAKERS Counts SUM 9 7 9 6 7 9 5 Mean 38.94 26.30 21.98 20.45 19.31 12.72 10.79 4.33 3.76 2.44 3.41 2.76 1.41 2.16 Std. Mean 0.87 0.75 0.49 0.68 0.55 0.28 0.43

So, we suggest that the std. Mean values could be taken as suitable proxies for values for the optimum marketing plan derivation.

CIS Coursework

20

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan

Application of Collaborative Filtering for automatic classification of users

In traditional lifestyle segmentation methods, the VALS classification task can be performed upon the data collected by means of the psychographic questionnaire. However, the need of using questionnaires presents a significant obstacle: although it can initially assess the consumer's need and desire for products, the evaluation of the accuracy of this assessment and its consequent adjustments cannot be performed unless the user is engaged in (difficult to perform in practice and annoying for the users) recompletions of the questionnaire at regular time intervals. Instead, to identify lifestyle neighbors, we propose a segmentation-based approach where only a portion of the population provides the lifestyle data through questionnaires as given in Appendix-A which are utilized to infer the cluster membership. Having acquired the ground truth data concerning membership of the sample users, the idea is to exploit their behavioral data, namely their interactions with the web, and some basic demographic data (which are collected easily and are used as lifestyle segment descriptors) to produce classification rules that are able to classify the rest of the users without requiring the completion of the lifestyle questionnaire. As more and more data are collected from the sample, along with adjustment of the training set size, rules are being refined up to the point where the noise is minimized. The benefit of this approach is two-fold: the disengagement of the process from the need for completing timeconsuming questionnaires, and the continuous evaluation and update of the segmentation process using dynamic behavioral data. Despite the benefits of the approach described above, there are also important limitations: The sparsity problem affects the performance of the classification process both when developing the classification rules and when applying the rules on sparse user data. This problem also applies to first-time users (known as the cold start problem). VALS instruments are of a proprietary nature, making reliability and validity difficult to assess. This can seriously limit the generalizability of the approach by anchoring the whole process in proprietary data of doubtful quality. Recommender System In this section, we propose an automated recommender system to recommend the VALS classification of users through web-analytics. Recommender systems methods and techniques have been characterized as "the lightweight model for heavyweight applications", since even one-click interactions can be utilized in the user model and processed in order to produce the recommendations. This makes recommender systems appropriate for the web domain, since they allow for preserving the requirement for short interactive sessions. These systems have been successfully employed for various domains, one being personalized advertisement selection in interactive-television

CIS Coursework

21

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan domain, however no previous reference of applying recommender system for webdomain based customer classification has been observed. Recommender systems rely on data such as explicit or implicit expression of users' interest in observed items. These expressions are measured on binary nominal (e.g. interesting/not-interesting) or numerical ordinal rating scales (e.g. 1 to 5) The major approach utilized for the prediction task is collaborative filtering. Collaborative Filtering Collaborative filtering (CF) is the process of filtering for information or patterns using techniques involving collaboration among multiple agents, viewpoints, data sources, etc. In collaborative filtering, entities are recommended to a new user based on the stated preferences of other, similar users. (For example, a CF system might suggest the classification Experiencers for the user Deepak after noticing that Ankesh's tastes are similar to Deepak's tastes, and that Ankesh has already been classified as and Experiencer) We propose a mechanism to implement Collaborative Filtering in the context of social networking by using Social Bookmarking Mechanism on the web. Social bookmarking is a method for Internet users to organize, store, manage and search for bookmarks of resources online. Unlike file sharing, the resources themselves aren't shared, merely bookmarks that reference them. Case Study : Delicious We explain the mechanism by using a case-study of how the process would be done for a popular website www.delicious.com. Recommendations generated by Delicious are less personal than those of most other CF sites. But the database maintained by Delicious is more transparent, and that database is not just on the Web, it is about the Web (not about books, songs, or movies). So Delicious makes a good case study. In the following case, the tags done for a particular site by users are studied. The neighbors for a user is decided who have made similar tags for the same website.

CIS Coursework

22

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan

Delicious allows many users to bookmark the same page, and each user can specify her own set of tags for that page. Each bookmark includes the following elements: User: the ID of the person who created the bookmark (e.g., MardiW) Page: the URL of the Web page being bookmarked (e.g., http://wordpress.org) Tags: the set of tags associated by that user with that page (e.g., {blog, blogging, wordpress}) We represent the user, page, and tags of a Delicious bookmark as a 3-tuple: (u,p,T) u = the user who created the bookmark p = the page being bookmarked T = the set of tags associated by that user with that page For example, the bookmarks illustrated above include these 3-tuples: (valentinote, http://wordpress.org, {webdesign, blog, programming}) (Pisces6, http://wordpress.org, {blog, software}) The association between an user, a webpage and tags could be represented by a colored graph as represented by the following

CIS Coursework

23

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan

WebSit e

Users Users Users Users Users Users

WebSit e WebSit e WebSit e WebSit e WebSit e WebSit e

Tags Tags Tags Tags Tags Tags

The collaborative filtering is applied on the idea that two users have similar neighborhood w.r.t the tags applied to the various websites. These could be also termed as the structural equivalence. When the classification for an user is required, the nearest neighbor for the user is determined. The nearest neighbor is the one with the most similar website-tag mapping to that of the given user.

CIS Coursework

24

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan

Conclusion and Future Work

In the current study, we have tried to come up with an empirical approach for estimating , by using the VALS psychographic classification scheme. We claim that VALS segment of a consumer could be used as an accurate proxy for the value of a consumer while identifying his network worth in estimating the optimum marketing plan. We have done a quantitative study, within the limited scope of the study, to substantiate the claims made above. Further to this, we have proposed a mechanism to automatically classify consumers through the collaborative filtering mechanism using web-analytics of social-bookmarking sites. However, due to the limited scope of our study, we could not do an empirical study on the proposed collaborative filtering mechanism. A suitable future work on this current study would be to perform a study on the collaborative filtering approach on the database of a bookmarking website and substantiate or even take forward the conjecture as proposed by this paper.

CIS Coursework

25

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan

References
[1]. H.-K. Ahn, S.-W. Cheng, O. Cheong, M. Golin, and R. van Oostrum. Competitive facility location: the Voronoi game. Theoret. Comput. Sci., 310(1-3):457467, 2004. [2] W. B. Arthur. Competing technologies, increasing returns, and lock-in by historical events. Economic Journal, 99(394):11631, March 1989. [3] J. Goldenberg, B. Libai, and E. Muller. Talk of the network: A complex systems look at the underlying process of word-of-mouth. Marketing Letters, 12(3):211223, 2001. [4] D. Kempe, J. Kleinberg, and E. Tardos. Maximizing the spread of influence through a social network. In 9th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 246257, 2003. [5] M. Richardson and P. Domingos. Mining knowledge-sharing sites for viral marketing. In Eighth Intl. Conf. on Knowledge Discovery and Data Mining, pages 6170, 2002. [6] M. Tomochi, H. Murata, and M. Kono. A consumer-based model of competitive diffusion: the multiplicative effects of global and local network externalities. Journal of Evolutionary Economics, 15:273295, 2005. [7] T. W. Valente. Network Models of the Diffusion of Innovations. Quantitative Methods in Communication Subseries. Hampton Press, New York, NY, 1995. [8] D. Kempe, J. Kleinberg, and E. Tardos. Influential nodes in a diffusion model for social networks. In 32nd International Colloquium on Automata, Languages and Programming (ICALP), pages 11271138, 2005. [9] P. Domingos and M. Richardson. Mining the network value of customers. In Proceedings of the 7th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 5766, 2001. [10] D. Gruhl, R. V. Guha, D. Liben-Nowell, and A. Tomkins. Information diffusion through blogspace. In Proceedings of the 13th international conference on World Wide Web, pages 491501, 2004. [11] W. Chen, Y. Wang, and S. Yang. Efficient influence maximization in social networks. In Proceedings of the 15th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 2009. [12] M. Kimura and K. Saito. Tractable models for information diffusion in social networks. In Proceedings of the 10th European Conference on Principles and Practice of Knowledge Discovery in Databases, pages 259271, 2006.

CIS Coursework

26

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan [13] Time Carnes, Chadrasekhar Nagarajan, Stefan M. Wild. Maximizing Influence in a Competitive Social Network: A Followers Perspective

CIS Coursework

27

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan

Appendix
Survey Questionnaire Q1. I am interested in theories Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q2. I like outrageous people and things. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q3. I like a lot of variety in my life. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q4. I love to make things I can use every day. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q5. I follow the latest trends and fashions. Mostly disagree Somewhat disagree Somewhat agree Mostly agree CIS Coursework 28

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan Q6. Just as in Hindu mythology, I believe Lord Visnu is the creator the world. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q7. I like being in charge of a group. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q8. I like to learn about art, culture, and history. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q9. I often crave excitement. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q9. I am really interested in only a few things. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q10. I would rather make something than buy it. Mostly disagree

CIS Coursework

29

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan Somewhat disagree Somewhat agree Mostly agree Q11. I dress more fashionably than most people. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q12. The government should encourage prayers in schools. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q13. I have more ability than most people. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q14. I consider myself an intellectual. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q15. I must admit that I like to show off. Mostly disagree Somewhat disagree Somewhat agree Mostly agree CIS Coursework 30

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan Q16. I like trying new things. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q17. I am very interested in how mechanical things, such as engines, work. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q18. I like to dress in the latest fashions. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q19. There is too much sex on television today. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q20. I like to lead others. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q21. I would like to spend a year or more in a foreign country.

Mostly disagree

CIS Coursework

31

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan Somewhat disagree Somewhat agree Mostly agree Q22. I like a lot of excitement in my life. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q23. I must admit that my interests are somewhat narrow and limited. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q24. I like making things of wood, metal, or other such material. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q25. I want to be considered fashionable. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q26. A woman's life is fulfilled only if she can provide a happy home for her family. Mostly disagree Somewhat disagree Somewhat agree Mostly agree CIS Coursework 32

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan Q27. I like the challenge of doing something I have never done before. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q28. I like to learn about things even if they may never be of any use to me. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q29. I like to make things with my hands. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q30. I am always looking for a thrill. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q31. I like doing things that are new and different. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q32. I like to look through hardware or automotive stores. Mostly disagree

CIS Coursework

33

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan Somewhat disagree Somewhat agree Mostly agree Q33. I would like to understand more about how the universe works. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q34. I like my life to be pretty much the same from week to week. Mostly disagree Somewhat disagree Somewhat agree Mostly agree Q35. I make purchases on the Internet (If answer is Never, please ignore questions 37, 38 and 39) Heavily Frequently Occassionally Once in a blue moon Never Q36. I consider the opinion of my friends when I make decisions about making purchases on the internet Always true Mostly true Not consistent Mostly false Always false

CIS Coursework

34

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan Q37. I consider the opinions of my family (eg. parents, spouses, children) when I make decisions about making purchases on the Internet. Always true Mostly true Not consistent Mostly false Always false Q38. I consider information I see on the media when I make decisions about making purchases on the internet. Always true Mostly true Not consistent Mostly false Always false Q39. Sex: Male Female Q40. Age 1824 2529 3034 3539 4044 4549 5054 5559 6064 6569 7074

CIS Coursework

35

Application of VALS classification for estimating customer self-reliance in defining optimum viral marketing plan 75 or older Q41. Education Below 12th standard Passed 12th standard Graduate Post-Graduate Doctorate and above Q42. Annual Income: Below 5 lac 5 - 10 lac 10 - 15 lac 15 - 20 lac 20 - 25 lac 25 - 30 lac Above 30 lac

CIS Coursework

36