in this issue:

e-news l november 2006

Viral Marketing in a telco environment

Viral Marketing

‘Traditional’ mass marketing techniques are becoming less and less effective.
Modern consumers have become more immune to the majority of advertising fired at them via the various existing channels.
End result: traditional campaigns are becoming less effective, and the investment needed to make them have the desired impact is increasing. This means new channels must be found for getting the message across. One possibility is viral marketing.

What is Viral Marketing?

The formal definition is: “An effort by an organization to encourage, facilitate and amplify a marketing message to consumers”.
In practice, this results in two major approaches:

The first approach involves the network of consumers being used to spread a message.
This approach is very popular in online marketing:
  - Currently a popular application: the spreading of information about a new product using funny videos containing brand info.
  - The traditional example from Yahoo! and Hotmail, where each message contains an invitation to sign up for the free e-mail service.

The second approach targets key-influencers and encourages them to share good experiences and product/service endorsements. The idea is that consumers want to reduce the “search-cost” (mainly time) when looking for a new product. So instead of spending their own time searching, they look for a local expert (everybody knows someone in their vicinity that can fix a computer). It is clear that these experts’ opinions have a big influence on a lot of people. Knowing and targeting these “leaders” has a high leverage towards potential sales.

Why Viral Marketing works…

As the influence of traditional marketing channels decreases, research shows that the influence of friends/contacts has become a more dominant influence in the buying process for a wide variety of products and services.

An example to demonstrate the power of this influence:
Verbal complaints (such as bad experiences by customers, and also a poor image with non-customers) cause a reduction in market potential by an average of 14%. On the other hand (the good news), positive recommendation is 2-7 times more influential than direct sales.

These examples show that the sharing of positive experiences between consumers can be a powerful asset if managed well.

The Telecom Setting

We can see that telecom companies have a big advantage over FMCG (Fast Moving Consumer Goods) companies: in order to get to know their influencers, FMCG companies have to start from scratch and completely build up their social network from responses to specific campaigns and market surveys. Telecom companies, on the other hand, have a great approximation of this network: the network of people calling (texting, MMS…) each other. This means that with the correct techniques the “leaders” can be found and the viral effect can be tracked.

These techniques are based on a combination of social network analysis techniques and large graph (the technical name for a network) mathematics, and make it possible to analyze the network and discover various customer roles. The most important role in a telecom operator’s customer base is the opinion leader. Opinion leaders are ideal for spreading viral messages over the whole base.

Targeting Social Leaders with Viral Campaigns

What is the first approach?
Social leaders make up 1-2% of the customer base and can be seen as the interface that infects the rest of the base by spreading the ‘virus’ (e.g. a marketing message,..). The small size of the group means they can be targeted with high intensity.

What is the second approach?
  - Excel in customer service: Customer service is usually a trade-off between customer satisfaction and cost. A negative experience for a leader is very costly. It doesn’t only affect them; it also affects their contacts.
  - Give access to new technology: this doesn’t only mean just giving it for free, but also removing all usage barriers and motivating them to start the ‘virus’ (using the technology). …

If we look back at the first approach, where customers are used as the transport medium for campaigns, then the most logical argument is: If we use bulk SMS campaigns, there is no difference in cost if we send the promo to just a few customers or to the whole base.
Besides the cost argument (which is only valid for no cost bulk campaigns), we also see another advantage: Giving customers the opportunity to forward promotions generates positive word of mouth advertising. This positive effect will raise awareness and influence involvement, positively affecting bottom line churn and acquisition.

Difference between social network models and traditional prediction models

All telecom providers today have several prediction models such as: churn prediction, bad debt prediction, cross and up sell, etc. It is widely assumed that these techniques should be used with classic models like regression and prediction tree models, or more complex models such as neural or Bayesian networks, to score customers and flag those with the highest scores.

The problem with this reasoning is that people tend to have contacts with similar profiles. Pre-paid customers tend to call pre-paid customers, high value customers tend to call high value customers, customers from city X tend to call customers from city X, customers from segment A tend to call customers from segments A, and so on…

Traditional models look for customers with the maximum predicted score: customers with a specific profile. But customers with a similar profile have a high probability of calling each other. Using the high scoring customers as seeds for a viral marketing campaign would therefore result in local (relative to the network) results. This means this approach is not suitable for viral campaigns.

The social network models can take the whole network into account and so spread the seeding throughout the network, resulting in an optimized viral spread.

The social leader model

IKAN Consulting has been working on social network analysis for several years, ranging from detecting opinion leaders to assisting with campaign deployment. IKAN uses a model called the social leader model that was created together with academia, with a specific focus on call networks. This model is being implemented in several European operators and achieves excellent results.

Extra reading

LINKED by Albert-Làszló Barabàsi
This is standard literature for anyone interested in social networks. It is also very easy to read.

Social Network Analysis by John Scott
This book handles the social science aspects of social networks

Social Network Analysis – Methods and Applications by S. Wasserman and K. Faust
This can be considered as the bible for the mathematical aspects of social network analysis. (Reading material for die-hard fans)

The Secrets of Word-of-Mouth Marketing by G. Silverman
This book provides a thought and execution framework for viral marketing campaigns.

As an Ikan Consulting contact, we want to keep you up to date on evolutions in our company and product range.
If you would prefer to discontinue your subscription, unsubscribe here.

in this issue: Viral Marketing Our organisation Customer Events
	e-news l november 2006
	Viral Marketing in a telco environment Viral Marketing ‘Traditional’ mass marketing techniques are becoming less and less effective. Modern consumers have become more immune to the majority of advertising fired at them via the various existing channels. End result: traditional campaigns are becoming less effective, and the investment needed to make them have the desired impact is increasing. This means new channels must be found for getting the message across. One possibility is viral marketing. What is Viral Marketing? The formal definition is: “An effort by an organization to encourage, facilitate and amplify a marketing message to consumers”. In practice, this results in two major approaches: The first approach involves the network of consumers being used to spread a message. This approach is very popular in online marketing: - Currently a popular application: the spreading of information about a new product using funny videos containing brand info. - The traditional example from Yahoo! and Hotmail, where each message contains an invitation to sign up for the free e-mail service. The second approach targets key-influencers and encourages them to share good experiences and product/service endorsements. The idea is that consumers want to reduce the “search-cost” (mainly time) when looking for a new product. So instead of spending their own time searching, they look for a local expert (everybody knows someone in their vicinity that can fix a computer). It is clear that these experts’ opinions have a big influence on a lot of people. Knowing and targeting these “leaders” has a high leverage towards potential sales. Why Viral Marketing works… As the influence of traditional marketing channels decreases, research shows that the influence of friends/contacts has become a more dominant influence in the buying process for a wide variety of products and services. An example to demonstrate the power of this influence: Verbal complaints (such as bad experiences by customers, and also a poor image with non-customers) cause a reduction in market potential by an average of 14%. On the other hand (the good news), positive recommendation is 2-7 times more influential than direct sales. These examples show that the sharing of positive experiences between consumers can be a powerful asset if managed well. The Telecom Setting We can see that telecom companies have a big advantage over FMCG (Fast Moving Consumer Goods) companies: in order to get to know their influencers, FMCG companies have to start from scratch and completely build up their social network from responses to specific campaigns and market surveys. Telecom companies, on the other hand, have a great approximation of this network: the network of people calling (texting, MMS…) each other. This means that with the correct techniques the “leaders” can be found and the viral effect can be tracked. These techniques are based on a combination of social network analysis techniques and large graph (the technical name for a network) mathematics, and make it possible to analyze the network and discover various customer roles. The most important role in a telecom operator’s customer base is the opinion leader. Opinion leaders are ideal for spreading viral messages over the whole base. Targeting Social Leaders with Viral Campaigns What is the first approach? Social leaders make up 1-2% of the customer base and can be seen as the interface that infects the rest of the base by spreading the ‘virus’ (e.g. a marketing message,..). The small size of the group means they can be targeted with high intensity. What is the second approach? - Excel in customer service: Customer service is usually a trade-off between customer satisfaction and cost. A negative experience for a leader is very costly. It doesn’t only affect them; it also affects their contacts. - Give access to new technology: this doesn’t only mean just giving it for free, but also removing all usage barriers and motivating them to start the ‘virus’ (using the technology). … If we look back at the first approach, where customers are used as the transport medium for campaigns, then the most logical argument is: If we use bulk SMS campaigns, there is no difference in cost if we send the promo to just a few customers or to the whole base. Besides the cost argument (which is only valid for no cost bulk campaigns), we also see another advantage: Giving customers the opportunity to forward promotions generates positive word of mouth advertising. This positive effect will raise awareness and influence involvement, positively affecting bottom line churn and acquisition. Difference between social network models and traditional prediction models All telecom providers today have several prediction models such as: churn prediction, bad debt prediction, cross and up sell, etc. It is widely assumed that these techniques should be used with classic models like regression and prediction tree models, or more complex models such as neural or Bayesian networks, to score customers and flag those with the highest scores. The problem with this reasoning is that people tend to have contacts with similar profiles. Pre-paid customers tend to call pre-paid customers, high value customers tend to call high value customers, customers from city X tend to call customers from city X, customers from segment A tend to call customers from segments A, and so on… Traditional models look for customers with the maximum predicted score: customers with a specific profile. But customers with a similar profile have a high probability of calling each other. Using the high scoring customers as seeds for a viral marketing campaign would therefore result in local (relative to the network) results. This means this approach is not suitable for viral campaigns. The social network models can take the whole network into account and so spread the seeding throughout the network, resulting in an optimized viral spread. The social leader model IKAN Consulting has been working on social network analysis for several years, ranging from detecting opinion leaders to assisting with campaign deployment. IKAN uses a model called the social leader model that was created together with academia, with a specific focus on call networks. This model is being implemented in several European operators and achieves excellent results. Extra reading LINKED by Albert-Làszló Barabàsi This is standard literature for anyone interested in social networks. It is also very easy to read. Social Network Analysis by John Scott This book handles the social science aspects of social networks Social Network Analysis – Methods and Applications by S. Wasserman and K. Faust This can be considered as the bible for the mathematical aspects of social network analysis. (Reading material for die-hard fans) The Secrets of Word-of-Mouth Marketing by G. Silverman This book provides a thought and execution framework for viral marketing campaigns.
		Schaliënhoevedreef 20A, 2800 Mechelen l +32 (0)15 44 80 20 l ikanconsulting.com