Viral Marketing :Decoded

viral marketing

Welcome to Viral Marketing: Decoding the Viral Loop

After the success of games like Candy Crush and FarmVille many firms have taken to word of mouth and viral marketing on Facebook and other social media websites. Companies have realized that it's word of mouth marketing that ultimately lends authenticity to a viral message.

Yet, will viral marketing approach work for other products?

When we look at research studying this phenomenon we find that they provide either descriptive accounts of particular initiatives, or, advice based on anecdotal evidence. What is missing is an analysis of viral marketing that highlights systematic processes that results in predictable desired outcomes.

In this article we take look at viral marketing from various different perspectives: the senderthe receiver and the message. We try to understand what motivates someone to pass on message and how the mechanism works.

Table of Contents

The first thing to know about Viral Marketing is that…

Viral spread happens reactively, not proactively

What this means is, viral spread is a reactive behaviour rather than a proactive one.

(Gil-Or, O. (2010) set out to test out this hypothesis.

An experiment was initiated to figure out the possibility of generating demand using viral marketing on Facebook, which included the promotion of a restaurant group page.  As a part of the experiment, the “Cramim restaurant” Facebook page was created using the pre-defined Facebook’s template for restaurant pages.

In order to start the distribution of the restaurant page virally, an invitation to become a fan of the “Cramim restaurant” page has been sent to 20 friends that were identified. The invitation was sent without any additional message attached, to avoid the influence of the audience by the recognition of the message source.

During the following few weeks two other actions were taken:

  • A message to the existing restaurant members was sent, in which they were told that the members of the group will get special offers from time to time and they were asked to invite their friend to join the group and enjoy these benefits.
  • A special coupon was offered exclusively to the group members where they got an offer to get a free glass of wine and an appetizer while dining in the restaurant. The coupon was used in order to attract more Facebook users and give them a reason to become a fan of the “Cramim restaurant” Facebook page.

The message path was tracked manually. This approach tracked which friends each group member got and in some cases, the researchers got in direct contact with the individuals for tracking cases where apparently different “routes” lead to the same person.

When the group members were asked about the way in which they first heard about the group. There were three ways that were mentioned:

  • A friend that became a group fan and sent an invitation to his friends
  • A message of the friend joining the group was published on his publicly-exposed wall page.
  • The logo of the “Cramim” group that appeared on a friend’s personal info page was seen by his friends that got curious about this group.

During this first month 80 new members (including the first 12 members) joined the “Cramim” group.

If one ignored the anomaly of the one long-branch that was developed, there were four branches with that were seven-members deep.

That meant that the viral message was transferred six times from the first sender of the message. It is important to reiterate that the “message” could be transferred reactively by someone within one’s network that gets exposed to the restaurant group information.

The first proactive approach was actually used very rarely. The majority of the members joined as a result of the reactive message transfer i.e. path no. 1. This is really surprising since it means that the community could grow without any special efforts from the marketer’s side.

To understand the spread mechanism better, researchers used the Facebook admin statistics tool. It was mainly used in order to track the fans’ activity and statistics. This graph shows the time-based progress of the total number of fans that joined the community during the first 5 weeks since the original message was distributed with an average of 2.3 members a day and with a convex graph that showed a reduced growth rate.

Having seen this, let’s proceed to see how the virus spreads…

Characteristics of viral spread

The stochastic process that shows influence is propagated from the seeding to their neighbours, and further.

However, all mentioned models and algorithms ignore one important aspect of influence propagation in the real world. That is influence propagation often happens only within a close proximity of the seeding.

For example, study of Chaet. al. in Flickr network reveals that the typical chain length is less than four; another study of Leskovec et. al. suggests that social influence happens on the level of direct friends. Moreover, shared information in social networks such as Facebook, in most cases, can be seen only by friends or friends of friends i.e. the propagation is basically limited within two hops from the source. (Dinh, T. N., Nguyen, D. T., & Thai, M. T. (2012, June), p 2)

This gives rise to two questions:

  • When the influence only propagates locally, is massively reaching customers via viral marketing still affordable?
  • In addition, can we speed up the spread of information for time-critical applications such as political campaigns or time-sensitive announcements?

To determine the best rate of message propagation we, then, need to first find out the optimum number of contacts to seed with our message.

Additional Resources:-

Viral Marketing : The Participants' View

Determining the optimum number of exposures

In epidemic models it is assumed that all individuals have an equal
probability of getting infected with every interaction.

Another discussed point is that traditional epidemic and innovation diffusion models
also often assume that individuals either have a constant probability of ‘converting’
every time they interact with an infected individual,
or,
that they convert once the fraction of their contacts that are infected exceeds a certain threshold.

In both cases, it is expected that an increasing number of infected
contacts results in an increased likelihood of infection.
But what has been observed in real life marketing scenarios is quite the
opposite. In real viral marketing campaigns researchers have observed that
the probability of infection decreases with repeated interaction. (Leskovec, J., Adamic, L. A., & Huberman, B. A. (2007), p 46).

This is an important finding for marketers who assume that more
the exposures (interactions) the better is the viral spread.
What Leskovic, Adamic and Huberman, 2007, found was that the probability of
purchasing a product increased with the number of recommendations received,
but quickly saturated to a constant and relatively low probability.

This means individuals are often impervious to the recommendations
of their friends, and resist buying items that they do not really want no
matter how many recommendations, or exposures they recieve.
(Leskovec, J., Adamic, L. A., & Huberman, B. A. (2007), p 46)
It was also observed that smaller and more tightly knit groups were
more conducive to viral marketing.  (Leskovec, J., Adamic, L. A., & Huberman, B. A. (2007), p 46).

We will go into details of groups and types of groups a little later in this article.
Let’s now understand the formula for viral propagation.
Therefore, it is important to find out how many seeds to influence to begin the viral marketing process.

The Viral Equation

It is common sense, that as long as every new seed is able to infect
at least one other new seed, the infected base continues to grow.

The number of new seeds that a current seed can bring to the infected base,
in turn, depends on the number of requests sent out by the user (N) and the
conversion rate (Cr) of those into the following equation:

Viral Index = =* Cr > 1

This simple equation proved to be a powerful diagnostic for Plaxo managers
when Plaxo ran a viral campaign to get new users to use their application.
(Kalyanam, K., McIntyre, S., & Masonis, J. T. (2007))

The Viral Index indicates the relative size of each new generation
of customers compared to the previous period.

For example, when = 1.1, each new generation of customers is 1.1 times
the size of the previous generation.

When = 1.0, each new generation is
smaller than the previous one and the number of new customers eventually goes to zero.

is indeed a magic number, and _1 a magic threshold.

For best results it is necessary to increase Cr as well as N.
But only increasing ‘N’ (the number of exposures) doesn’t increase
the viral index, as seen in the previous point.

Some Viral Marketing Models.

The first model explores the motivations of the sender and the receiver.
In this study Greg Nyilasy (2006), sets out to understand the phenomenon
of word-out-mouth propagation (viral marketing in its traditional form) in detail.

  • He studied things like how does it work?
  • How does it influence the consumers?
  • What are the factors that make it more effective?
  • And, what causes it and what effects does it have?

The results of the study are measure over four parameters the perspectives of the sender and the receiver and the triggers to word-of-mouth propagation and the effects. This is tabulated in a 4 X 4 matrix as shown below:
viral marketing quadrants

Quadrant 1. What initiates word of mouth

Word of mouth begins with either seeking of information or receiving unsolicited information.

What causes receivers to seek information or advice, etc.

 People usually seek information when searching for additional information, or searching for a product / service. They also come across new information when someone in their network sends across such information (the motivation for group members to send such information varies and is covered later in this article).

Why does such a need arise to look for information?

To reduce the perceived risk

viral marketing word of mouth

Early research proposed that one of the crucial predictors of what influenced people to seek more information; thereby generating input for word of mouth was perceived risk.

Perceived risk is the subjective assessment of potentially negative outcomes of acquiring or using a product.

Early studies and a more recent replication found empirical support for the hypothesis that the more risky the consumer perceives the purchase decision to be; the more likely he/she would be exposed to word of mouth.

A higher level of perceived risk is associated with services as opposed to products, according to ‘services marketing theory’. Risk is one of the reasons why consumers of services use word of mouth sources more often than purchasers of products, a proposition that has further empirical support.

The more difficult the decision, higher is the perceived risk by the individual.

The difficulty level of a decision is calculated on 2 dimensions:

1) Number of alternatives from which a choice has to be made, and

2) Number of attributes on which a choice has to be based.

Additional attribute is that of complexity of attributes to be considered.

What sources of information do the receivers seek?

The level of risk associated with decision tasks and the level of difficulty determine the type of influence sources that the customer seeks.

When difficulty levels of a decision task increases people feel less confident about making good judgements and they seek sources that are similar to themselves. (Duhan, D. F., Johnson, S. D., Wilcox, J. B., & Harrell, G. D. (1997), p 5)

E.g. High school seniors who are considering various colleges are often overwhelmed with the amount of information that is sent to them from various colleges. A common response is to ask the family members.

When perceived task difficulty increases, customers are more likely to seek information from strong-tie sources. (Duhan, D. F., Johnson, S. D., Wilcox, J. B., & Harrell, G. D. (1997), pp 10

Additional Resources:-

A different line of research, using ‘network analysis’ methods, suggests that the kind of social relationships the receiver has with potential communicators (preceding the communication event) is another important predictor of whether word of mouth will occur. (Nyilasy, G. (2006), p12)

If this relationship is a closely-knit and strong one, then the receiver will seek information from sender.

Relying on the ‘strength of ties’ theory, sociometric research in marketing found that the stronger the relationship between the members of a particular social network, the more likely that these strong-tie relationships will be used for word of mouth communication.

In other words, there is strong empirical evidence (also replicated recently) for the common sense notion that if word of mouth occurs, it tends to be between close friends and relatives rather than superficial acquaintances (although ‘weak ties’ are also important, especially for the spread of word of mouth between overlapping social networks, as we will see later). (Nyilasy, G. (2006), p12)

When it comes to topics related to technology, research, science, etc. the receivers seek out industry experts, experts in the community or influencers for the information. (Nyilasy, G. (2006), p 12)

Quadrant 2. Effects of word of mouth for the receiver

viral marketing WOM on receiver
The ‘two-step flow’ theory of media and advertising effects stated that instead of direct effects, commercial messages were channelled through ‘opinion leaders’ – individuals influential in social networks – and it is only through their output word of mouth that viral communication effects can occur. (Nyilasy, G. (2006), p12)
Similarly, individuals who are seen are ‘experts’ in a community are sought for their advice when it comes to reducing the perceived risk of purchase.

The groups’ predisposition towards your brand.

Favorable predisposition toward the brand moderates the effects of word of mouth on product evaluation, i.e. the new seeds, too, perceived the brand more favorably.
Similarly, highly diagnostic information such as prior impressions about the brand and the presence of extremely negative information also influence the extent to which word of mouth is related to communication effectiveness variables.
A study carried out in this direction reports that a favorable brand name reduces the persuasiveness of negative word-of-mouth (NWOM). This means when the brand is strongly perceived to be positive, negative word-of-mouth doesn’t have any effect on the new seeds in changing their already held perceptions.(Nyilasy, G. (2006), p13) .
Hence, a strongly favorable brand helps build immunity against negative word-of-mouth (NWOM).

Why you should seed the network first

Communicators that come from a primary group (family and friends) are more trustworthy and credible than impersonal or weaker personal ones. This is the case in terms of (Nyilasy, G. (2006), p 13)
“The greater the similarity (homophily) of the seeker and source is, the greater will be the influence of the source on the seeker’s decision.” (Gilly, M. C., Graham, J. L., Wolfinbarger, M. F., & Yale, L. J. (1998), p 9)

Seed only relevant groups

Another reason suggested by the two-step flow theorists is that normative influence (conformity to influencers and group norms) is also at play when word of mouth (informational influence) is passed along. (Nyilasy, G. (2006), p13)
What this means is messages which adhere to the norms of the group get passed along.
E.g. messages about new software in the market are more likely to get passed along in a software developers’ social group than a message about the latest fiction novel released by an author.

Positioning of the sender

On the other hand, the perception that the communicator is an expert source enhances word of mouth effectiveness, even if the message is initiated by the sender. (Nyilasy, G. (2006), p13)

“The greater the expertise of the source, the more likely he or she will be perceived to be an opinion leader. The greater the expertise of the source, the greater will be the influence on the seeker’s decision.” (Gilly, M. C., Graham, J. L., Wolfinbarger, M. F., & Yale, L. J. (1998), p 9)

 ‘Attribution theory’- The motives

This theory proposes is that if the receiver thinks that the communicator is unbiased, i.e. un-incentivized, or isn’t likely to benefit personally by propagating the message, the causal attribution implies the communicator really believes in the message; which makes the receiver accept the message.
On the other hand, the attribution for paid-for messages is that the motivation of the source is commercial in nature and, not genuine evaluation. Consequently, the message is received sceptically.
“The findings of the study suggest that the receiver’s causal attribution about the motives behind the communicator’s behaviour is a key mediator in the Negative Word-Of-Mouth –brand evaluation relationship.” (Nyilasy, G. (2006), p14)

Quadrant 3. What initiates word of mouth output

viral marketing what starts word of mouth

The literature reviewed in this area investigates the factors influencing the likelihood and extent to which communicators engage in positive or negative word of mouth. The two-step flow theory proposes that mass media messages are not directly influential; rather they should be filtered through opinion leaders.

Are some motivations better for certain product categories?

To understand this we study Taylor’s six segment strategy wheel

Taylor’s six segment strategy wheel (1999)

In 1999 Golan and Zaidner (Golan, G. J., & Zaidner, L. (2008)) tried and applied the Taylor’s six segment strategy wheel to viral advertisements.

viral marketing taylor's six segments

The research question in the study asked if different product categories use different messages strategies in viral advertisements. The results in Table 5 indicate that such indeed was the case.

Successful viral advertisements for some of the product categories based their creative strategies on the ritual view more predominantly than on the transmission view. The ritual view was more widely used in the fashion (84.6%), alcohol and tobacco (82%), food and beverage (66.6%), entertainment and media (65.3%), and automotive (64.1%) product categories.

A combination of the ritual and transmission views was common in the travel (50%), banking (50%) and non for profit (41.6%) product categories. The only product categories that were relatively evenly balanced between the ritual and transmission views were the electronic and communications (46.7%, 36.3%), other (44.7%, 39.5%) and household products (50%, 35.7%) product categories.

viral marketing product category wise ritual vs transmission

The results of the study indicate that viral ads were primarily designed around the ritual view as a whole and across product categories. The results also indicated that most viral ads employed the ego strategy accounting for more than 50% of total ads.

Viral ads were not absent of the transmission view approach as nearly a quarter of the viral advertisements employed the ration strategy.

When synthesized with the results of the advertising appeal results, one could argue that viral ads were often based on an individual appeal (ego rather than social) that was based largely on humor while attempting to provide some information to the user.

In layman terms, it could be argued that viral advertising strategies target users through the gut rather than the brain. The prominence of this approach will likely advance rather than wane. Viral advertising competes daily with thousands of video clips that are user generated rather than professionally produced.

Target opinion leaders first

Opinion leaders process the information first, since they tend to be more frequently exposed to mass media. So the opinion leadership trait can be understood as a knowledge-based antecedent to output word of mouth.

The rationale for this type of investigation has been that the identification of opinion leaders is the key for marketers to be able to manage word of mouth. In the framework of the two-step flow theory this has meant that advertising campaign should target opinion leaders first.

Quadrant 4. Effects of word of mouth for the communicator

word of mouth after effects

While it is conceivable that word of mouth has an effect on the communicator – not only on the receiver of the communication – not much research has investigated this possibility.

The most important effects are ego-enhancement and the reduction of cognitive dissonance.

A word of mouth episode on the one hand might reassure the communicator that (s)he has made the right purchase decision. Just by talking about it, the communicator can get rid of negative feelings associated with cognitive dissonance.

On the other hand, the communicator might also feel good about him/herself as (s)he helps out a fellow human being. (S)he might also entertain the thought that (s)he was knowledgeable and competent in something. These forms of ego enhancements can be powerful emotional effect on the communicator.

 

Viral Marketing framework based on the participation levels.

viral marketing framework

3 comments

  1. Excellent pieces. Keep writing such kind of info on your page.

    Im really impressed by your site.
    Hi there, You have performed an excellent job. I will definitely digg it and for my part recommend to my friends.
    I am sure they will be benefited from this website.

Leave a comment

Your email address will not be published. Required fields are marked *